Bench on the same disk Start benchmark with little-text Native Copy results : 2.212 ms action statistics: first = 2.492 ms, mean = 2.212 ms (CI deltas: -15.039 us, +28.537 us), sd = 1.791 ms (CI deltas: -690.216 us, +1.630 ms) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 512 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 512 --block statistics: mean = 1.133 s (CI deltas: -7.700 ms, +14.611 ms), sd = 40.536 ms (CI deltas: -15.618 ms, +36.873 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.130 s, interquantileRange = 46.619 ms --1 are EXTREME (on the high side): #59 = 1.373 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 4: r[1] is 3.7274133468601915 sigma above its mean, r[2] is 2.898209287152889 sigma above its mean, r[3] is 2.5185438319610163 sigma above its mean, r[4] is 2.2591939368453864 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 10.669321155463322% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 70.83302818347738% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 512.0, muB = 1.1325084997500001, sigmaB = 0.04053574450049242, muA = 0.002211930663574219, sigmaA = 0.0017914437385464679, tMin = 0.0, muGMin = 0.0011059653317871095, sigmaG = 2.7649133294677737E-4, cMax1 = 310, cMax2 = 143, cMax = 143, cOutMin = 143, varOutMin = 0.001163890481672089, muG(cOutMin) = 0.001106331476150905, U(cOutMin) = 0.005064840454897316 Naive Streams results : 10.826 ms action statistics: first = 11.869 ms, mean = 10.826 ms (CI deltas: -15.482 us, +19.685 us), sd = 777.493 us (CI deltas: -161.398 us, +226.312 us) WARNING: execution times have mild outliers, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 128 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 128 --block statistics: mean = 1.386 s (CI deltas: -1.982 ms, +2.520 ms), sd = 8.796 ms (CI deltas: -1.826 ms, +2.560 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.385 s, interquantileRange = 9.570 ms --4 are mild (on the high side): #56 = 1.407 s, #57 = 1.407 s, #58 = 1.410 s, #59 = 1.415 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 54.38561472711181% of the measured sd Naive Readers results : 491.949 us action statistics: first = 12.519 ms, mean = 491.949 us (CI deltas: -676.975 ns, +784.847 ns), sd = 130.474 us (CI deltas: -21.437 us, +40.065 us) WARNING: execution times have mild outliers, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 2048 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 2048 --block statistics: mean = 1.008 s (CI deltas: -1.386 ms, +1.607 ms), sd = 5.905 ms (CI deltas: -970.137 us, +1.813 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.007 s, interquantileRange = 7.862 ms --1 are mild (on the high side): #59 = 1.027 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is 2.345345825579906 sigma above its mean, r[4] is 2.0388953366443454 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 69.08591949165664% of the measured sd ---------- --action sd values likely INFLATED by outliers --they cause at least 20.193867067079037% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 2048.0, muB = 1.0075106942666667, sigmaB = 0.0059045866047145665, muA = 4.919485811848959E-4, sigmaA = 1.3047416338427881E-4, tMin = 0.0, muGMin = 2.4597429059244793E-4, sigmaG = 6.149357264811198E-5, cMax1 = 1915, cMax2 = 1614, cMax = 1614, cOutMin = 1614, varOutMin = 7.040418685958112E-6, muG(cOutMin) = 2.463299656098161E-4, U(cOutMin) = 5.579947268847624E-4 Buffered Streams results : 107.030 us action statistics: first = 1.130 ms, mean = 107.030 us (CI deltas: -741.002 ns, +1.393 us), sd = 500.415 us (CI deltas: -198.741 us, +265.971 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16384 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16384 --block statistics: mean = 1.754 s (CI deltas: -12.141 ms, +22.826 ms), sd = 64.053 ms (CI deltas: -25.439 ms, +34.044 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.744 s, interquantileRange = 33.946 ms --3 are EXTREME (on the high side): #57 = 1.980 s, #58 = 2.008 s, #59 = 2.047 s --1 are mild (on the high side): #56 = 1.837 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is 3.0064336741291067 sigma above its mean, r[2] is 2.0978754703280544 sigma above its mean, r[3] is 2.5430199733031413 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 8.401875299043374% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 98.80097720186836% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16384.0, muB = 1.7535778016833332, sigmaB = 0.06405315038496943, muA = 1.0702989512227376E-4, sigmaA = 5.004152373825737E-4, tMin = 0.0, muGMin = 5.351494756113688E-5, sigmaG = 1.337873689028422E-5, cMax1 = 717, cMax2 = 185, cMax = 185, cOutMin = 185, varOutMin = 0.004053612494046248, muG(cOutMin) = 5.357114831295772E-5, U(cOutMin) = 0.004787993352225573 Buffered Readers results : 142.572 us action statistics: first = 1.660 ms, mean = 142.572 us (CI deltas: -179.906 ns, +191.540 ns), sd = 66.247 us (CI deltas: -9.141 us, +18.111 us) WARNING: execution times have mild outliers, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 8192 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 8192 --block statistics: mean = 1.168 s (CI deltas: -1.474 ms, +1.569 ms), sd = 5.996 ms (CI deltas: -827.358 us, +1.639 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.168 s, interquantileRange = 9.454 ms --1 are mild (on the high side): #59 = 1.187 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 73.25005161959915% of the measured sd ---------- --action sd values likely INFLATED by outliers --they cause at least 44.02044924515337% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 8192.0, muB = 1.1679504218000003, sigmaB = 0.005995956530157201, muA = 1.4257207297363285E-4, sigmaA = 6.624658628396748E-5, tMin = 0.0, muGMin = 7.128603648681642E-5, sigmaG = 1.7821509121704106E-5, cMax1 = 6752, cMax2 = 4463, cMax = 4463, cOutMin = 4463, varOutMin = 1.582600948236517E-5, muG(cOutMin) = 7.130207099379162E-5, U(cOutMin) = 2.0212077057229474E-4 Custom Buffer Streams results : 47.027 us action statistics: first = 177.56 us, mean = 47.027 us (CI deltas: -400.467 ns, +432.331 ns), sd = 295.884 us (CI deltas: -22.741 us, +17.935 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32768 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32768 --block statistics: mean = 1.541 s (CI deltas: -13.123 ms, +14.167 ms), sd = 53.561 ms (CI deltas: -4.117 ms, +3.247 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 4: r[1] is -5.039043878277568 sigma below its mean, r[4] is -2.82704342565004 sigma below its mean, r[5] is 3.608063062617954 sigma above its mean, r[10] is 2.00047746359176 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 9.419054163151728% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.33541116689631% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32768.0, muB = 1.5409895797666666, sigmaB = 0.05356076403346057, muA = 4.702726989034017E-5, sigmaA = 2.9588421448119144E-4, tMin = 0.0, muGMin = 2.3513634945170084E-5, sigmaG = 5.878408736292521E-6, cMax1 = 807, cMax2 = 205, cMax = 205, cOutMin = 205, varOutMin = 0.002849690015519175, muG(cOutMin) = 2.3555203133110946E-5, U(cOutMin) = 0.0037754219519179266 Custom Buffer Readers results : 63.356 us action statistics: first = 229.921 us, mean = 63.356 us (CI deltas: -248.123 ns, +227.607 ns), sd = 120.479 us (CI deltas: -15.649 us, +17.353 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16384 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16384 --block statistics: mean = 1.038 s (CI deltas: -4.065 ms, +3.729 ms), sd = 15.421 ms (CI deltas: -2.003 ms, +2.221 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 4: r[2] is -3.4160503480309394 sigma below its mean, r[4] is 4.191038978228379 sigma above its mean, r[6] is -1.9687768797876608 sigma below its mean, r[8] is 2.287935600658531 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 26.849523155305345% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 93.13412364188973% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16384.0, muB = 1.0380251956833335, sigmaB = 0.01542132100040295, muA = 6.335603000996909E-5, sigmaA = 1.2047907031564805E-4, tMin = 0.0, muGMin = 3.167801500498454E-5, sigmaG = 7.919503751246136E-6, cMax1 = 3558, cMax2 = 1063, cMax = 1063, cOutMin = 1063, varOutMin = 2.2148891051072654E-4, muG(cOutMin) = 3.16854867477705E-5, U(cOutMin) = 5.198230039705942E-4 Custom Buffer Buffered Streams results : 53.750 us action statistics: first = 211.418 us, mean = 53.750 us (CI deltas: -250.299 ns, +430.546 ns), sd = 228.852 us (CI deltas: -77.282 us, +152.954 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32768 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32768 --block statistics: mean = 1.761 s (CI deltas: -8.202 ms, +14.108 ms), sd = 41.427 ms (CI deltas: -13.989 ms, +27.688 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.755 s, interquantileRange = 37.656 ms --1 are EXTREME (on the high side): #59 = 1.975 s --2 are mild (on the high side): #57 = 1.861 s, #58 = 1.880 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 13.019392931933163% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 98.55505303912932% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32768.0, muB = 1.761288681266667, sigmaB = 0.04142654786566592, muA = 5.3750264931233735E-5, sigmaA = 2.2885150716376148E-4, tMin = 0.0, muGMin = 2.6875132465616868E-5, sigmaG = 6.718783116404217E-6, cMax1 = 1714, cMax2 = 446, cMax = 446, cOutMin = 446, varOutMin = 0.001691361282658473, muG(cOutMin) = 2.6879044235793225E-5, U(cOutMin) = 0.002001130523491835 Custom Buffer Buffered Readers results : 81.306 us action statistics: first = 287.224 us, mean = 81.306 us (CI deltas: -620.144 ns, +1.445 us), sd = 448.822 us (CI deltas: -226.699 us, +375.216 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16384 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16384 --block statistics: mean = 1.332 s (CI deltas: -10.160 ms, +23.677 ms), sd = 57.449 ms (CI deltas: -29.017 ms, +48.028 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.317 s, interquantileRange = 16.516 ms --5 are EXTREME (on the high side): #55 = 1.381 s, #56 = 1.382 s, #57 = 1.501 s, #58 = 1.514 s, #59 = 1.671 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is 4.322762842137024 sigma above its mean, r[2] is 2.3699815912737763 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 8.164736512767146% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.13778493522551% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16384.0, muB = 1.3321195770666667, sigmaB = 0.057449154070098145, muA = 8.130612652994792E-5, sigmaA = 4.4882151617264176E-4, tMin = 0.0, muGMin = 4.065306326497396E-5, sigmaG = 1.016326581624349E-5, cMax1 = 520, cMax2 = 133, cMax = 133, cOutMin = 133, varOutMin = 0.003271948711645603, muG(cOutMin) = 4.071332981289864E-5, U(cOutMin) = 0.0050412575509567745 NIO Buffer results : 53.441 us action statistics: first = 4.324 ms, mean = 53.441 us (CI deltas: -703.149 ns, +700.145 ns), sd = 502.277 us (CI deltas: -14.784 us, +23.072 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32768 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32768 --block statistics: mean = 1.751 s (CI deltas: -23.041 ms, +22.942 ms), sd = 90.922 ms (CI deltas: -2.676 ms, +4.177 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is -6.195554996857976 sigma below its mean, r[2] is 3.452685223971835 sigma above its mean, r[3] is -2.492303134317408 sigma below its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 5.914896031675407% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.70164947770736% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32768.0, muB = 1.7511465190000004, sigmaB = 0.09092186471265334, muA = 5.344075070190431E-5, sigmaA = 5.022770866909612E-4, tMin = 0.0, muGMin = 2.6720375350952155E-5, sigmaG = 6.680093837738039E-6, cMax1 = 366, cMax2 = 92, cMax = 92, cOutMin = 92, varOutMin = 0.008242121485161213, muG(cOutMin) = 2.6791528864901154E-5, U(cOutMin) = 0.009518549150114025 NIO Transfer results : 52.106 us action statistics: first = 1.323 ms, mean = 52.106 us (CI deltas: -1.029 us, +1.286 us), sd = 821.491 us (CI deltas: -131.027 us, +136.420 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32768 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32768 --block statistics: mean = 1.707 s (CI deltas: -33.703 ms, +42.154 ms), sd = 148.706 ms (CI deltas: -23.718 ms, +24.695 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is -2.5460470307830665 sigma below its mean, r[2] is -2.2475883314191485 sigma below its mean, r[7] is 2.6293911099277807 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 3.571039220848684% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.89606983201922% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32768.0, muB = 1.7074050633333335, sigmaB = 0.14870578680028, muA = 5.210586741129558E-5, sigmaA = 8.214911738137419E-4, tMin = 0.0, muGMin = 2.605293370564779E-5, sigmaG = 6.513233426411947E-6, cMax1 = 131, cMax2 = 32, cMax = 32, cOutMin = 32, varOutMin = 0.02209042852266276, muG(cOutMin) = 2.642253058201334E-5, U(cOutMin) = 0.026326159443767025 Start benchmark with medium-text Native Copy results : 26.407 ms action statistics: first = 33.746 ms, mean = 26.407 ms (CI deltas: -364.619 us, +935.720 us), sd = 17.324 ms (CI deltas: -9.215 ms, +17.984 ms) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 64 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 64 --block statistics: mean = 1.690 s (CI deltas: -23.336 ms, +59.886 ms), sd = 138.596 ms (CI deltas: -73.724 ms, +143.870 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.646 s, interquantileRange = 20.894 ms --10 are EXTREME (on the high side): #50 = 1.728 s, #51 = 1.739 s, #52 = 1.752 s, #53 = 1.762 s, #54 = 1.787 s, #55 = 1.798 s, #56 = 1.933 s, #57 = 1.941 s, #58 = 1.990 s, #59 = 2.589 s --1 are mild (on the high side): #49 = 1.715 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is 3.897452990938686 sigma above its mean, r[2] is 2.4811850768329275 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 3.8120111923051057% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 62.57263124627522% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 64.0, muB = 1.6900528037833338, sigmaB = 0.1385958341388209, muA = 0.02640707505911459, sigmaA = 0.017324479267352613, tMin = 0.0, muGMin = 0.013203537529557295, sigmaG = 0.003300884382389324, cMax1 = 44, cMax2 = 23, cMax = 23, cOutMin = 23, varOutMin = 0.012019454870038073, muG(cOutMin) = 0.013583106432629999, U(cOutMin) = 0.04926719304545669 Naive Streams results : 107.995 ms action statistics: first = 110.854 ms, mean = 107.995 ms (CI deltas: -186.627 us, +235.982 us), sd = 3.291 ms (CI deltas: -789.497 us, +1.498 ms) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.728 s (CI deltas: -2.986 ms, +3.776 ms), sd = 13.166 ms (CI deltas: -3.158 ms, +5.990 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.729 s, interquantileRange = 12.537 ms --1 are EXTREME (on the high side): #59 = 1.783 s --2 are mild (on the high side): #57 = 1.754 s, #58 = 1.764 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is 6.328239469222116 sigma above its mean, r[2] is 2.7571717755721616 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 40.576705567098294% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859374999999994% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.7279177484333335, sigmaB = 0.013165548457304575, muA = 0.10799485927708334, sigmaA = 0.003291387114326144, tMin = 0.0, muGMin = 0.05399742963854167, sigmaG = 0.003291387114326144, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 1, varOutMin = 1.0156152315330158E-5, muG(cOutMin) = 0.10778240115242156, U(cOutMin) = 0.11118173114701005 Naive Readers results : 4.423 ms action statistics: first = 78.201 ms, mean = 4.423 ms (CI deltas: -5.955 us, +7.008 us), sd = 405.416 us (CI deltas: -59.766 us, +75.354 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 256 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 256 --block statistics: mean = 1.132 s (CI deltas: -1.524 ms, +1.794 ms), sd = 6.487 ms (CI deltas: -956.261 us, +1.206 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 5: r[1] is 4.894987054938894 sigma above its mean, r[2] is 2.628868190104773 sigma above its mean, r[9] is -2.0727679424671663 sigma below its mean, r[10] is -2.0566514187309837 sigma below its mean, r[11] is -2.0134232384598176 sigma below its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 66.66875082409828% of the measured sd Buffered Streams results : 668.932 us action statistics: first = 793.493 us, mean = 668.932 us (CI deltas: -1.289 us, +1.830 us), sd = 271.327 us (CI deltas: -62.738 us, +108.569 us) WARNING: execution times have mild outliers, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 2048 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 2048 --block statistics: mean = 1.370 s (CI deltas: -2.640 ms, +3.749 ms), sd = 12.279 ms (CI deltas: -2.839 ms, +4.913 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.364 s, interquantileRange = 14.792 ms --3 are mild (on the high side): #57 = 1.399 s, #58 = 1.401 s, #59 = 1.419 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is 2.7082138750402116 sigma above its mean, r[14] is 2.836954502034727 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 38.739400601190646% of the measured sd ---------- --action sd values likely INFLATED by outliers --they cause at least 37.3872471844534% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 2048.0, muB = 1.3699725796166673, sigmaB = 0.012278837371772768, muA = 6.689319236409508E-4, sigmaA = 2.71326536583354E-4, tMin = 0.0, muGMin = 3.344659618204754E-4, sigmaG = 8.361649045511885E-5, cMax1 = 1761, cMax2 = 1253, cMax = 1253, cOutMin = 1253, varOutMin = 5.6368695453200277E-5, muG(cOutMin) = 3.346388342269825E-4, U(cOutMin) = 8.81033285240396E-4 Buffered Readers results : 859.084 us action statistics: first = 1.049 ms, mean = 859.084 us (CI deltas: -3.403 us, +6.759 us), sd = 826.590 us (CI deltas: -392.790 us, +509.973 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 2048 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 2048 --block statistics: mean = 1.759 s (CI deltas: -6.969 ms, +13.841 ms), sd = 37.407 ms (CI deltas: -17.776 ms, +23.079 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.751 s, interquantileRange = 21.067 ms --3 are EXTREME (on the high side): #57 = 1.867 s, #58 = 1.919 s, #59 = 1.945 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is 3.8467317902796747 sigma above its mean, r[2] is 2.264707140573679 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 14.410596453843663% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 77.47516945404914% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 2048.0, muB = 1.7594047974333336, sigmaB = 0.03740719435193024, muA = 8.590843737467449E-4, sigmaA = 8.265900247316559E-4, tMin = 0.0, muGMin = 4.2954218687337245E-4, sigmaG = 1.0738554671834311E-4, cMax1 = 1067, cMax2 = 440, cMax = 440, cOutMin = 440, varOutMin = 0.0010841086433145087, muG(cOutMin) = 4.295709937570296E-4, U(cOutMin) = 0.0024287605442546135 Custom Buffer Streams results : 114.767 us action statistics: first = 223.254 us, mean = 114.767 us (CI deltas: -589.177 ns, +755.523 ns), sd = 336.587 us (CI deltas: -64.833 us, +54.619 us) WARNING: execution times have mild outliers, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16384 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16384 --block statistics: mean = 1.880 s (CI deltas: -9.653 ms, +12.378 ms), sd = 43.083 ms (CI deltas: -8.299 ms, +6.991 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.859 s, interquantileRange = 32.112 ms --11 are mild (on the high side): #49 = 1.957 s, #50 = 1.959 s, #51 = 1.961 s, #52 = 1.962 s, #53 = 1.964 s, #54 = 1.966 s, #55 = 1.969 s, #56 = 1.969 s, #57 = 1.970 s, #58 = 1.972 s, #59 = 1.973 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[2] is -1.9735794470674584 sigma below its mean, r[5] is 4.731892436298876 sigma above its mean, r[10] is 2.523964107488952 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 12.934983994575203% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 97.00253374117135% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16384.0, muB = 1.8803443929166663, sigmaB = 0.043083106530745464, muA = 1.1476711382548012E-4, sigmaA = 3.3658676977144894E-4, tMin = 0.0, muGMin = 5.738355691274006E-5, sigmaG = 1.4345889228185015E-5, cMax1 = 1708, cMax2 = 463, cMax = 463, cOutMin = 463, varOutMin = 0.0018005164764292089, muG(cOutMin) = 5.7419057499183204E-5, U(cOutMin) = 0.0020867723077152714 Custom Buffer Readers results : 202.990 us action statistics: first = 379.746 us, mean = 202.990 us (CI deltas: -562.804 ns, +624.330 ns), sd = 211.280 us (CI deltas: -26.491 us, +31.518 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 8192 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 8192 --block statistics: mean = 1.663 s (CI deltas: -4.610 ms, +5.115 ms), sd = 19.123 ms (CI deltas: -2.398 ms, +2.853 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 5: r[1] is -3.018830181000338 sigma below its mean, r[2] is 2.159970823765408 sigma above its mean, r[5] is 2.824693930163439 sigma above its mean, r[6] is -2.781442086189558 sigma below its mean, r[7] is 2.130524860481189 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 27.405278737757715% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 80.13096048038621% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 8192.0, muB = 1.66289563465, sigmaB = 0.01912284389387194, muA = 2.0299018977661132E-4, sigmaA = 2.1127957176450955E-4, tMin = 0.0, muGMin = 1.0149509488830566E-4, sigmaG = 2.5373773722076414E-5, cMax1 = 3947, cMax2 = 1550, cMax = 1550, cOutMin = 1550, varOutMin = 2.930254272926964E-4, muG(cOutMin) = 1.0152440195023822E-4, U(cOutMin) = 6.377874560622694E-4 Custom Buffer Buffered Streams results : 123.188 us action statistics: first = 260.659 us, mean = 123.188 us (CI deltas: -457.657 ns, +1.394 us), sd = 368.914 us (CI deltas: -218.744 us, +425.364 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16384 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16384 --block statistics: mean = 2.018 s (CI deltas: -7.498 ms, +22.838 ms), sd = 47.221 ms (CI deltas: -27.999 ms, +54.447 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 2.017 s, interquantileRange = 19.762 ms --1 are mild (on the low side): #0 = 1.974 s --1 are EXTREME (on the high side): #59 = 2.347 s --1 are mild (on the high side): #58 = 2.058 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 12.226836372796276% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 97.11904650251036% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16384.0, muB = 2.018318331166667, sigmaB = 0.047220973790332216, muA = 1.2318837470499675E-4, sigmaA = 3.6891385773697044E-4, tMin = 0.0, muGMin = 6.159418735249838E-5, sigmaG = 1.5398546838124594E-5, cMax1 = 1645, cMax2 = 445, cMax = 445, cOutMin = 445, varOutMin = 0.0021655802778936632, muG(cOutMin) = 6.159895975304756E-5, U(cOutMin) = 0.0023291966554221163 Custom Buffer Buffered Readers results : 233.627 us action statistics: first = 407.169 us, mean = 233.627 us (CI deltas: -336.540 ns, +335.748 ns), sd = 121.055 us (CI deltas: -16.822 us, +24.154 us) WARNING: SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 8192 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 8192 --block statistics: mean = 1.914 s (CI deltas: -2.757 ms, +2.750 ms), sd = 10.957 ms (CI deltas: -1.523 ms, +2.186 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 51.313758855408174% of the measured sd ---------- --action sd values likely INFLATED by outliers --they cause at least 49.521656833562204% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 8192.0, muB = 1.9138701707833337, sigmaB = 0.010956620379815955, muA = 2.336267298319499E-4, sigmaA = 1.2105469639772792E-4, tMin = 0.0, muGMin = 1.1681336491597495E-4, sigmaG = 2.920334122899374E-5, cMax1 = 6474, cMax2 = 4011, cMax = 4011, cOutMin = 4011, varOutMin = 5.944952591676173E-5, muG(cOutMin) = 1.1683277589455119E-4, U(cOutMin) = 3.553708139536812E-4 NIO Buffer results : 134.238 us action statistics: first = 318.667 us, mean = 134.238 us (CI deltas: -2.144 us, +3.078 us), sd = 910.359 us (CI deltas: -213.053 us, +264.538 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 8192 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 8192 --block statistics: mean = 1.100 s (CI deltas: -17.567 ms, +25.216 ms), sd = 82.396 ms (CI deltas: -19.283 ms, +23.943 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.062 s, interquantileRange = 14.816 ms --11 are EXTREME (on the high side): #49 = 1.217 s, #50 = 1.219 s, #51 = 1.234 s, #52 = 1.234 s, #53 = 1.241 s, #54 = 1.254 s, #55 = 1.254 s, #56 = 1.285 s, #57 = 1.300 s, #58 = 1.311 s, #59 = 1.387 s --1 are mild (on the high side): #48 = 1.115 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 5.172248082514038% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.42928068806526% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 8192.0, muB = 1.0996770987166669, sigmaB = 0.08239625061453311, muA = 1.3423792708943687E-4, sigmaA = 9.103585555294154E-4, tMin = 0.0, muGMin = 6.711896354471844E-5, sigmaG = 1.677974088617961E-5, cMax1 = 174, cMax2 = 44, cMax = 44, cOutMin = 44, varOutMin = 0.006750395170166048, muG(cOutMin) = 6.735120445828183E-5, U(cOutMin) = 0.012520442836149693 NIO Transfer results : 93.562 us action statistics: first = 256.579 us, mean = 93.562 us (CI deltas: -1.761 us, +3.570 us), sd = 1.206 ms (CI deltas: -489.804 us, +928.415 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16384 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16384 --block statistics: mean = 1.533 s (CI deltas: -28.859 ms, +58.485 ms), sd = 154.425 ms (CI deltas: -62.695 ms, +118.837 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.484 s, interquantileRange = 7.493 ms --1 are mild (on the low side): #0 = 1.468 s --7 are EXTREME (on the high side): #53 = 1.766 s, #54 = 1.777 s, #55 = 1.791 s, #56 = 1.812 s, #57 = 1.812 s, #58 = 1.999 s, #59 = 2.396 s --1 are mild (on the high side): #52 = 1.501 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 3.2583230831184418% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.8441459571853% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16384.0, muB = 1.5329130112333327, sigmaB = 0.15442542913776666, muA = 9.356158515828447E-5, sigmaA = 0.001206448665138802, tMin = 0.0, muGMin = 4.678079257914223E-5, sigmaG = 1.1695198144785558E-5, cMax1 = 97, cMax2 = 24, cMax = 24, cOutMin = 24, varOutMin = 0.023810046318568062, muG(cOutMin) = 4.735515236202033E-5, U(cOutMin) = 0.031590946607945 Start benchmark with big-text Native Copy results : 33.544 ms action statistics: first = 44.503 ms, mean = 33.544 ms (CI deltas: -424.718 us, +487.539 us), sd = 10.123 ms (CI deltas: -1.362 ms, +1.526 ms) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32 --block statistics: mean = 1.073 s (CI deltas: -13.591 ms, +15.601 ms), sd = 57.265 ms (CI deltas: -7.704 ms, +8.630 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 5: r[1] is 7.473924853779416 sigma above its mean, r[2] is 4.219478330498874 sigma above its mean, r[3] is 3.14714501145979 sigma above its mean, r[4] is 2.5731801987885503 sigma above its mean, r[5] is 2.1798073921163206 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 7.352725253031327% of the measured sd ---------- --action sd values likely INFLATED by outliers --they cause at least 26.76791399433096% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32.0, muB = 1.0734059527, sigmaB = 0.05726481583152096, muA = 0.033543936021875, sigmaA = 0.010123084899466808, tMin = 0.0, muGMin = 0.0167719680109375, sigmaG = 0.004192992002734375, cMax1 = 29, cMax2 = 23, cMax = 23, cOutMin = 23, varOutMin = 8.777892641633617E-4, muG(cOutMin) = 0.017756320514636062, U(cOutMin) = 0.03972169861166415 Naive Readers results : 441.997 ms action statistics: first = 541.966 ms, mean = 441.997 ms (CI deltas: -2.237 ms, +5.957 ms), sd = 26.748 ms (CI deltas: -15.911 ms, +28.301 ms) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 4 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 4 --block statistics: mean = 1.768 s (CI deltas: -8.947 ms, +23.826 ms), sd = 53.495 ms (CI deltas: -31.823 ms, +56.602 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.754 s, interquantileRange = 21.292 ms --2 are EXTREME (on the high side): #58 = 1.925 s, #59 = 2.119 s --2 are mild (on the high side): #56 = 1.825 s, #57 = 1.829 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 10.101305511996198% of the measured sd Buffered Streams results : 65.394 ms action statistics: first = 66.573 ms, mean = 65.394 ms (CI deltas: -98.792 us, +113.844 us), sd = 1.667 ms (CI deltas: -229.882 us, +431.250 us) WARNING: execution times have mild outliers, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.046 s (CI deltas: -1.581 ms, +1.822 ms), sd = 6.667 ms (CI deltas: -919.529 us, +1.725 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.046 s, interquantileRange = 9.637 ms --1 are mild (on the high side): #59 = 1.067 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is 3.4715790808459555 sigma above its mean, r[3] is -2.1557845422536506 sigma below its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 62.34838027832078% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859375% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.0463098273166664, sigmaB = 0.006667441031124669, muA = 0.06539436420729165, sigmaA = 0.0016668602577811673, tMin = 0.0, muGMin = 0.032697182103645826, sigmaG = 0.0016668602577811673, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 15, varOutMin = 2.604771674034656E-6, muG(cOutMin) = 0.05914363824061228, U(cOutMin) = 0.06581107927173695 Buffered Readers results : 82.482 ms action statistics: first = 82.166 ms, mean = 82.482 ms (CI deltas: -231.104 us, +280.565 us), sd = 4.008 ms (CI deltas: -666.517 us, +961.992 us) WARNING: execution times have mild outliers, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.320 s (CI deltas: -3.698 ms, +4.489 ms), sd = 16.032 ms (CI deltas: -2.666 ms, +3.848 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.316 s, interquantileRange = 22.237 ms --1 are mild (on the high side): #59 = 1.368 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 29.121214050105333% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.8593749999999964% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.3197127498333334, sigmaB = 0.01603187937143052, muA = 0.08248204686458334, sigmaA = 0.00400796984285763, tMin = 0.0, muGMin = 0.04124102343229167, sigmaG = 0.00400796984285763, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 1, varOutMin = 1.5059833369927694E-5, muG(cOutMin) = 0.08222333352369185, U(cOutMin) = 0.08636274697795562 Custom Buffer Streams results : 9.495 ms action statistics: first = 9.847 ms, mean = 9.495 ms (CI deltas: -17.026 us, +24.913 us), sd = 903.922 us (CI deltas: -230.425 us, +543.396 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 128 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 128 --block statistics: mean = 1.215 s (CI deltas: -2.179 ms, +3.189 ms), sd = 10.227 ms (CI deltas: -2.607 ms, +6.148 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.212 s, interquantileRange = 11.490 ms --1 are EXTREME (on the high side): #59 = 1.264 s --1 are mild (on the high side): #58 = 1.238 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is 2.7650702199719146 sigma above its mean, r[2] is 2.257622733907912 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 43.810568794051385% of the measured sd Custom Buffer Readers results : 17.406 ms action statistics: first = 26.373 ms, mean = 17.406 ms (CI deltas: -24.516 us, +29.449 us), sd = 847.674 us (CI deltas: -141.540 us, +202.821 us) WARNING: execution times have mild outliers, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 64 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 64 --block statistics: mean = 1.114 s (CI deltas: -1.569 ms, +1.885 ms), sd = 6.781 ms (CI deltas: -1.132 ms, +1.623 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.113 s, interquantileRange = 9.289 ms --2 are mild (on the high side): #58 = 1.133 s, #59 = 1.133 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is 5.878284099588966 sigma above its mean, r[2] is 3.6050066376780863 sigma above its mean, r[3] is 2.220866473039748 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 63.25172195692221% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 1.5380859374999998% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 64.0, muB = 1.11397022985, sigmaB = 0.006781388820080942, muA = 0.01740578484140625, sigmaA = 8.476736025101177E-4, tMin = 0.0, muGMin = 0.008702892420703125, sigmaG = 8.476736025101177E-4, cMax1 = 63, cMax2 = 63, cMax = 63, cOutMin = 63, varOutMin = 7.073231842613484E-7, muG(cOutMin) = 0.010730355221639074, U(cOutMin) = 0.017511744041720016 Custom Buffer Buffered Streams results : 9.236 ms action statistics: first = 9.851 ms, mean = 9.236 ms (CI deltas: -8.585 us, +11.261 us), sd = 435.777 us (CI deltas: -98.575 us, +121.710 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 128 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 128 --block statistics: mean = 1.182 s (CI deltas: -1.099 ms, +1.441 ms), sd = 4.930 ms (CI deltas: -1.115 ms, +1.377 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.181 s, interquantileRange = 4.300 ms --1 are EXTREME (on the high side): #59 = 1.197 s --4 are mild (on the high side): #55 = 1.193 s, #56 = 1.195 s, #57 = 1.195 s, #58 = 1.196 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 89.62412642662085% of the measured sd Custom Buffer Buffered Readers results : 18.754 ms action statistics: first = 19.207 ms, mean = 18.754 ms (CI deltas: -17.688 us, +23.173 us), sd = 633.156 us (CI deltas: -137.034 us, +156.097 us) WARNING: execution times have mild outliers, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 64 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 64 --block statistics: mean = 1.200 s (CI deltas: -1.132 ms, +1.483 ms), sd = 5.065 ms (CI deltas: -1.096 ms, +1.249 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.199 s, interquantileRange = 4.978 ms --6 are mild (on the high side): #54 = 1.210 s, #55 = 1.210 s, #56 = 1.213 s, #57 = 1.214 s, #58 = 1.214 s, #59 = 1.215 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 87.90133958309386% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 1.5380859375% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 64.0, muB = 1.2002854477333333, sigmaB = 0.005065251146554643, muA = 0.018754460120833333, sigmaA = 6.331563933193304E-4, tMin = 0.0, muGMin = 0.009377230060416667, sigmaG = 6.331563933193304E-4, cMax1 = 63, cMax2 = 63, cMax = 63, cOutMin = 63, varOutMin = 3.9462315873862474E-7, muG(cOutMin) = 0.013768353523443607, U(cOutMin) = 0.018833604669998248 NIO Buffer results : 10.896 ms action statistics: first = 11.520 ms, mean = 10.896 ms (CI deltas: -77.171 us, +173.334 us), sd = 4.882 ms (CI deltas: -2.377 ms, +4.010 ms) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 128 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 128 --block statistics: mean = 1.395 s (CI deltas: -9.878 ms, +22.187 ms), sd = 55.231 ms (CI deltas: -26.888 ms, +45.364 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.380 s, interquantileRange = 15.497 ms --4 are EXTREME (on the high side): #56 = 1.466 s, #57 = 1.541 s, #58 = 1.575 s, #59 = 1.718 s --3 are mild (on the high side): #53 = 1.422 s, #54 = 1.423 s, #55 = 1.427 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is 3.9406203894036653 sigma above its mean, r[2] is 3.064807836678888 sigma above its mean, r[3] is 2.47122685084124 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 8.68979959360594% of the measured sd ---------- --action sd values likely INFLATED by outliers --they cause at least 42.26013555079401% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 128.0, muB = 1.394663638516667, sigmaB = 0.05523052673265598, muA = 0.010895809675911462, sigmaA = 0.004881734997645742, tMin = 0.0, muGMin = 0.005447904837955731, sigmaG = 0.0013619762094889327, cMax1 = 106, cMax2 = 72, cMax = 72, cOutMin = 72, varOutMin = 0.0012891078586026602, muG(cOutMin) = 0.00545550967493989, U(cOutMin) = 0.015127154121111572 NIO Transfer results : 6.484 ms action statistics: first = 7.455 ms, mean = 6.484 ms (CI deltas: -8.031 us, +10.375 us), sd = 573.847 us (CI deltas: -117.465 us, +262.749 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 256 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 256 --block statistics: mean = 1.660 s (CI deltas: -2.056 ms, +2.656 ms), sd = 9.182 ms (CI deltas: -1.879 ms, +4.204 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.659 s, interquantileRange = 10.202 ms --1 are EXTREME (on the high side): #59 = 1.698 s --1 are mild (on the high side): #58 = 1.680 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 57.026077200800074% of the measured sd Start benchmark with fat-text Native Copy results : 103.466 ms action statistics: first = 104.012 ms, mean = 103.466 ms (CI deltas: -879.676 us, +2.489 ms), sd = 20.800 ms (CI deltas: -18.145 ms, +23.007 ms) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.655 s (CI deltas: -14.075 ms, +39.828 ms), sd = 83.199 ms (CI deltas: -72.582 ms, +92.030 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.641 s, interquantileRange = 7.938 ms --2 are mild (on the low side): #0 = 1.623 s, #1 = 1.625 s --3 are EXTREME (on the high side): #57 = 1.686 s, #58 = 1.929 s, #59 = 2.224 s --1 are mild (on the high side): #56 = 1.660 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 6.284881842407584% of the measured sd ---------- --action sd values likely INFLATED by outliers --they cause at least 17.39072654131697% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.6554589970333329, sigmaB = 0.08319865466939104, muA = 0.1034661873145833, sigmaA = 0.02079966366734776, tMin = 0.0, muGMin = 0.05173309365729165, sigmaG = 0.012933273414322913, cMax1 = 15, cMax2 = 13, cMax = 13, cOutMin = 13, varOutMin = 0.0012037888978439415, muG(cOutMin) = 0.06176719229342504, U(cOutMin) = 0.11308903231946599 Buffered Streams results : 663.213 ms action statistics: first = 666.091 ms, mean = 663.213 ms (CI deltas: -2.447 ms, +5.738 ms), sd = 19.477 ms (CI deltas: +-14.414 ms) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 2 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 2 --block statistics: mean = 1.326 s (CI deltas: -4.893 ms, +11.477 ms), sd = 27.545 ms (CI deltas: +-20.385 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.321 s, interquantileRange = 9.980 ms --2 are EXTREME (on the high side): #58 = 1.466 s, #59 = 1.473 s --1 are mild (on the high side): #57 = 1.345 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 16.9922212471041% of the measured sd Buffered Readers results : 830.938 ms action statistics: first = 835.504 ms, mean = 830.938 ms (CI deltas: -2.509 ms, +5.382 ms), sd = 19.673 ms (CI deltas: -11.354 ms, +13.221 ms) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 2 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 2 --block statistics: mean = 1.662 s (CI deltas: -5.019 ms, +10.765 ms), sd = 27.821 ms (CI deltas: -16.057 ms, +18.697 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.657 s, interquantileRange = 16.334 ms --2 are EXTREME (on the high side): #58 = 1.789 s, #59 = 1.807 s --1 are mild (on the high side): #57 = 1.712 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 18.83099419910041% of the measured sd Custom Buffer Streams results : 102.260 ms action statistics: first = 103.284 ms, mean = 102.260 ms (CI deltas: -967.359 us, +2.437 ms), sd = 21.623 ms (CI deltas: -18.522 ms, +17.829 ms) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.636 s (CI deltas: -15.478 ms, +38.986 ms), sd = 86.493 ms (CI deltas: -74.086 ms, +71.315 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.615 s, interquantileRange = 12.984 ms --2 are EXTREME (on the high side): #58 = 2.019 s, #59 = 2.165 s --3 are mild (on the high side): #55 = 1.649 s, #56 = 1.654 s, #57 = 1.656 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 6.0101582549322705% of the measured sd ---------- --action sd values likely INFLATED by outliers --they cause at least 17.52145719830841% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.6361544668333332, sigmaB = 0.08649289831961154, muA = 0.10225965417708333, sigmaA = 0.021623224579902886, tMin = 0.0, muGMin = 0.05112982708854166, sigmaG = 0.012782456772135416, cMax1 = 15, cMax2 = 13, cMax = 13, cOutMin = 13, varOutMin = 0.0013107839730622742, muG(cOutMin) = 0.058746959158771504, U(cOutMin) = 0.11230104533515528 Custom Buffer Readers results : 180.946 ms action statistics: first = 181.822 ms, mean = 180.946 ms (CI deltas: -260.463 us, +287.580 us), sd = 3.059 ms (CI deltas: -321.334 us, +395.816 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 8 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 8 --block statistics: mean = 1.448 s (CI deltas: -2.084 ms, +2.301 ms), sd = 8.651 ms (CI deltas: -908.870 us, +1.120 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 5: r[1] is 5.806804064286864 sigma above its mean, r[2] is 3.242755705347858 sigma above its mean, r[3] is 2.6228507624366424 sigma above its mean, r[4] is 2.578754552002478 sigma above its mean, r[5] is 2.1597729950083098 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 56.519507932865906% of the measured sd Custom Buffer Buffered Streams results : 99.988 ms action statistics: first = 100.980 ms, mean = 99.988 ms (CI deltas: -111.470 us, +155.805 us), sd = 2.068 ms (CI deltas: -516.936 us, +664.541 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.600 s (CI deltas: -1.784 ms, +2.493 ms), sd = 8.271 ms (CI deltas: -2.068 ms, +2.658 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.597 s, interquantileRange = 8.135 ms --1 are EXTREME (on the high side): #59 = 1.628 s --4 are mild (on the high side): #55 = 1.616 s, #56 = 1.617 s, #57 = 1.623 s, #58 = 1.626 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 62.14548771917787% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859375% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.59981351165, sigmaB = 0.008271405003209631, muA = 0.099988344478125, sigmaA = 0.002067851250802408, tMin = 0.0, muGMin = 0.0499941722390625, sigmaG = 0.002067851250802408, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 15, varOutMin = 4.008758245729765E-6, muG(cOutMin) = 0.09223390228761598, U(cOutMin) = 0.1005053072908256 Custom Buffer Buffered Readers results : 195.537 ms action statistics: first = 198.422 ms, mean = 195.537 ms (CI deltas: -260.545 us, +335.561 us), sd = 3.292 ms (CI deltas: -634.055 us, +1.049 ms) WARNING: execution times have mild outliers, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 8 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 8 --block statistics: mean = 1.564 s (CI deltas: -2.084 ms, +2.684 ms), sd = 9.311 ms (CI deltas: -1.793 ms, +2.967 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.562 s, interquantileRange = 12.684 ms --2 are mild (on the high side): #58 = 1.591 s, #59 = 1.597 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 54.58938951275181% of the measured sd NIO Buffer results : 114.963 ms action statistics: first = 120.179 ms, mean = 114.963 ms (CI deltas: -151.679 us, +191.621 us), sd = 2.690 ms (CI deltas: -513.501 us, +700.170 us) WARNING: execution times have mild outliers, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.839 s (CI deltas: -2.427 ms, +3.066 ms), sd = 10.758 ms (CI deltas: -2.054 ms, +2.801 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.837 s, interquantileRange = 12.192 ms --3 are mild (on the high side): #57 = 1.868 s, #58 = 1.868 s, #59 = 1.872 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 5: r[1] is 2.770299724933579 sigma above its mean, r[2] is 2.6567520585953166 sigma above its mean, r[3] is 2.1860209386782845 sigma above its mean, r[4] is 2.1560411569013134 sigma above its mean, r[6] is 3.296168629247441 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 51.23388266935939% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859374999999999% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.8394079879666672, sigmaB = 0.010758109774992105, muA = 0.1149629992479167, sigmaA = 0.0026895274437480263, tMin = 0.0, muGMin = 0.05748149962395835, sigmaG = 0.0026895274437480263, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 15, varOutMin = 6.7814605037566806E-6, muG(cOutMin) = 0.1048772713338616, U(cOutMin) = 0.1156353811088537 NIO Transfer results : 74.167 ms action statistics: first = 74.539 ms, mean = 74.167 ms (CI deltas: -52.410 us, +74.100 us), sd = 974.984 us (CI deltas: -261.479 us, +373.775 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.187 s (CI deltas: -838.559 us, +1.186 ms), sd = 3.900 ms (CI deltas: -1.046 ms, +1.495 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.186 s, interquantileRange = 4.213 ms --1 are EXTREME (on the high side): #59 = 1.201 s --2 are mild (on the high side): #57 = 1.198 s, #58 = 1.200 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 100.0% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859375% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.186677364533333, sigmaB = 0.003899936358277061, muA = 0.07416733528333332, sigmaA = 9.749840895692652E-4, tMin = 0.0, muGMin = 0.03708366764166666, sigmaG = 9.749840895692652E-4, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 15, varOutMin = 8.911818514811334E-7, muG(cOutMin) = 0.07051114494744858, U(cOutMin) = 0.07441108130572563 Start benchmark with little-binary Native Copy results : 34.911 ms action statistics: first = 38.222 ms, mean = 34.911 ms (CI deltas: -17.164 us, +17.484 us), sd = 389.365 us (CI deltas: -52.331 us, +70.602 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32 --block statistics: mean = 1.117 s (CI deltas: -549.244 us, +559.500 us), sd = 2.203 ms (CI deltas: -296.028 us, +399.385 us) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is -3.526678167664108 sigma below its mean, r[2] is 2.8180019661449993 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 100.0% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 3.02734375% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32.0, muB = 1.1171602127000002, sigmaB = 0.0022025820517288554, muA = 0.03491125664687501, sigmaA = 3.8936517622431316E-4, tMin = 0.0, muGMin = 0.017455628323437503, sigmaG = 3.8936517622431316E-4, cMax1 = 31, cMax2 = 31, cMax = 31, cOutMin = 31, varOutMin = 1.4686757669193448E-7, muG(cOutMin) = 0.03277750528426268, U(cOutMin) = 0.03498008733599153 Naive Streams results : 10.823 ms action statistics: first = 10.984 ms, mean = 10.823 ms (CI deltas: -11.063 us, +14.746 us), sd = 566.368 us (CI deltas: -120.216 us, +222.760 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 128 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 128 --block statistics: mean = 1.385 s (CI deltas: -1.416 ms, +1.887 ms), sd = 6.408 ms (CI deltas: -1.360 ms, +2.520 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.385 s, interquantileRange = 6.477 ms --1 are EXTREME (on the high side): #59 = 1.411 s --2 are mild (on the high side): #57 = 1.400 s, #58 = 1.400 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 74.6514329424937% of the measured sd Buffered Streams results : 107.561 us action statistics: first = 206.172 us, mean = 107.561 us (CI deltas: -799.698 ns, +1.498 us), sd = 535.666 us (CI deltas: -215.128 us, +303.087 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16384 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16384 --block statistics: mean = 1.762 s (CI deltas: -13.102 ms, +24.550 ms), sd = 68.565 ms (CI deltas: -27.536 ms, +38.795 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.754 s, interquantileRange = 46.835 ms --3 are EXTREME (on the high side): #57 = 1.998 s, #58 = 2.014 s, #59 = 2.093 s --1 are mild (on the high side): #56 = 1.863 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is 4.616956786906825 sigma above its mean, r[2] is 2.499066142593592 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 7.8684385311349425% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 98.94337356383195% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16384.0, muB = 1.762287020133333, sigmaB = 0.0685652354333772, muA = 1.0756146363118487E-4, sigmaA = 5.356659018232594E-4, tMin = 0.0, muGMin = 5.3780731815592436E-5, sigmaG = 1.3445182953898109E-5, cMax1 = 635, cMax2 = 163, cMax = 163, cOutMin = 163, varOutMin = 0.004651517477724532, muG(cOutMin) = 5.3881340113795724E-5, U(cOutMin) = 0.0054495570683892785 Custom Buffer Streams results : 47.547 us action statistics: first = 144.53 us, mean = 47.547 us (CI deltas: -407.945 ns, +442.922 ns), sd = 303.654 us (CI deltas: -25.797 us, +23.252 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32768 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32768 --block statistics: mean = 1.558 s (CI deltas: -13.368 ms, +14.514 ms), sd = 54.967 ms (CI deltas: -4.670 ms, +4.209 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 5: r[1] is -4.7390933003472195 sigma below its mean, r[4] is -3.3637079188905923 sigma below its mean, r[5] is 4.2218619456979845 sigma above its mean, r[6] is -2.1560301186634745 sigma below its mean, r[10] is 2.6695560899465813 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 9.228608996575215% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.35485467146766% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32768.0, muB = 1.5580140622166665, sigmaB = 0.05496720413072625, muA = 4.754681586354573E-5, sigmaA = 3.036537717476698E-4, tMin = 0.0, muGMin = 2.3773407931772865E-5, sigmaG = 5.943351982943216E-6, cMax1 = 784, cMax2 = 199, cMax = 199, cOutMin = 199, varOutMin = 0.003001901150733885, muG(cOutMin) = 2.3815582627950312E-5, U(cOutMin) = 0.003931479153803783 Custom Buffer Buffered Streams results : 54.296 us action statistics: first = 153.249 us, mean = 54.296 us (CI deltas: -319.453 ns, +700.536 ns), sd = 321.839 us (CI deltas: -201.627 us, +225.878 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32768 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32768 --block statistics: mean = 1.779 s (CI deltas: -10.468 ms, +22.955 ms), sd = 58.259 ms (CI deltas: -36.498 ms, +40.888 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.769 s, interquantileRange = 36.477 ms --2 are EXTREME (on the high side): #58 = 2.068 s, #59 = 2.075 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 9.304681714833778% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.25119688149341% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32768.0, muB = 1.7791831245333336, sigmaB = 0.05825899791707871, muA = 5.429636000162761E-5, sigmaA = 3.218385350960883E-4, tMin = 0.0, muGMin = 2.7148180000813806E-5, sigmaG = 6.7870450002034516E-6, cMax1 = 907, cMax2 = 231, cMax = 231, cOutMin = 231, varOutMin = 0.003368695630499405, muG(cOutMin) = 2.7184496504327658E-5, U(cOutMin) = 0.003873078622389717 NIO Buffer results : 53.757 us action statistics: first = 164.068 us, mean = 53.757 us (CI deltas: +-699.433 ns), sd = 501.875 us (CI deltas: -14.870 us, +30.306 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32768 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32768 --block statistics: mean = 1.762 s (CI deltas: +-22.919 ms), sd = 90.849 ms (CI deltas: -2.692 ms, +5.486 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 4: r[1] is -6.61649985984924 sigma below its mean, r[2] is 3.7947685492530487 sigma above its mean, r[3] is -2.508918410068421 sigma below its mean, r[4] is 1.9749501331272297 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 5.937138191484014% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.69836133956234% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32768.0, muB = 1.7615164500166662, sigmaB = 0.0908490521962988, muA = 5.375721588185627E-5, sigmaA = 5.018748505654178E-4, tMin = 0.0, muGMin = 2.6878607940928135E-5, sigmaG = 6.719651985232034E-6, cMax1 = 371, cMax2 = 93, cMax = 93, cOutMin = 93, varOutMin = 0.008228654386447704, muG(cOutMin) = 2.698463193907957E-5, U(cOutMin) = 0.009460146251690765 NIO Transfer results : 52.219 us action statistics: first = 161.988 us, mean = 52.219 us (CI deltas: -1.019 us, +1.255 us), sd = 803.475 us (CI deltas: -130.260 us, +108.998 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32768 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32768 --block statistics: mean = 1.711 s (CI deltas: -33.406 ms, +41.137 ms), sd = 145.444 ms (CI deltas: -23.580 ms, +19.731 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 4: r[1] is -2.632888012355439 sigma below its mean, r[2] is -2.4012045275962386 sigma below its mean, r[7] is 2.1073763179085967 sigma above its mean, r[10] is 2.1412769473039153 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 3.655072909613736% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.8896541076775% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32768.0, muB = 1.711110389616667, sigmaB = 0.1454444576113144, muA = 5.2218944995625826E-5, sigmaA = 8.034747051793731E-4, tMin = 0.0, muGMin = 2.6109472497812913E-5, sigmaG = 6.527368124453228E-6, cMax1 = 137, cMax2 = 34, cMax = 34, cOutMin = 34, varOutMin = 0.021130747580200525, muG(cOutMin) = 2.6325024422785564E-5, U(cOutMin) = 0.02498197176944719 Start benchmark with medium-binary Native Copy results : 28.134 ms action statistics: first = 35.060 ms, mean = 28.134 ms (CI deltas: -383.793 us, +531.689 us), sd = 9.980 ms (CI deltas: -2.215 ms, +4.760 ms) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32 --block statistics: mean = 900.275 ms (CI deltas: -12.281 ms, +17.014 ms), sd = 56.457 ms (CI deltas: -12.530 ms, +26.926 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 878.821 ms, interquantileRange = 67.017 ms --1 are EXTREME (on the high side): #59 = 1.144 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 5: r[1] is 6.185890765968779 sigma above its mean, r[2] is 3.895935418890056 sigma above its mean, r[3] is 2.9642875342506594 sigma above its mean, r[4] is 2.518507268078677 sigma above its mean, r[5] is 2.1761557101981426 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 6.830055019427651% of the measured sd ---------- --action sd values likely INFLATED by outliers --they cause at least 32.907862719767344% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32.0, muB = 0.9002752287166664, sigmaB = 0.056456981532057554, muA = 0.028133600897395826, sigmaA = 0.009980278621660394, tMin = 0.0, muGMin = 0.014066800448697913, sigmaG = 0.0035167001121744782, cMax1 = 28, cMax2 = 21, cMax = 21, cOutMin = 21, varOutMin = 0.0010489021768645887, muG(cOutMin) = 0.01464134854018305, U(cOutMin) = 0.035200971179745376 Naive Streams results : 109.286 ms action statistics: first = 110.798 ms, mean = 109.286 ms (CI deltas: -150.978 us, +176.083 us), sd = 2.570 ms (CI deltas: -429.997 us, +620.899 us) WARNING: execution times have mild outliers, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.749 s (CI deltas: -2.416 ms, +2.817 ms), sd = 10.282 ms (CI deltas: -1.720 ms, +2.484 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.748 s, interquantileRange = 13.493 ms --3 are mild (on the high side): #57 = 1.774 s, #58 = 1.775 s, #59 = 1.779 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is 4.6401401872626105 sigma above its mean, r[2] is 3.0442213864138985 sigma above its mean, r[3] is 2.3647728788763653 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 52.26822143531981% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859374999999996% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.7485785760666668, sigmaB = 0.010281561105049547, muA = 0.10928616100416667, sigmaA = 0.0025703902762623868, tMin = 0.0, muGMin = 0.05464308050208334, sigmaG = 0.0025703902762623868, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 1, varOutMin = 6.19397453653521E-6, muG(cOutMin) = 0.10912024302527971, U(cOutMin) = 0.1117749306874711 Buffered Streams results : 672.608 us action statistics: first = 787.462 us, mean = 672.608 us (CI deltas: -1.442 us, +1.754 us), sd = 284.931 us (CI deltas: -51.873 us, +66.799 us) WARNING: execution times have mild outliers, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 2048 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 2048 --block statistics: mean = 1.378 s (CI deltas: -2.953 ms, +3.593 ms), sd = 12.895 ms (CI deltas: -2.347 ms, +3.023 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.377 s, interquantileRange = 16.976 ms --2 are mild (on the high side): #58 = 1.409 s, #59 = 1.416 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 36.990898482046376% of the measured sd ---------- --action sd values likely INFLATED by outliers --they cause at least 39.460190114489556% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 2048.0, muB = 1.3775013337166666, sigmaB = 0.01289452476590809, muA = 6.726080731038411E-4, sigmaA = 2.84931434442234E-4, tMin = 0.0, muGMin = 3.3630403655192056E-4, sigmaG = 8.407600913798014E-5, cMax1 = 1739, cMax2 = 1210, cMax = 1210, cOutMin = 1210, varOutMin = 6.560997232419966E-5, muG(cOutMin) = 3.3638029256110826E-4, U(cOutMin) = 9.054666516945933E-4 Custom Buffer Streams results : 115.924 us action statistics: first = 234.718 us, mean = 115.924 us (CI deltas: -1.020 us, +2.091 us), sd = 709.272 us (CI deltas: -358.501 us, +477.758 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16384 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16384 --block statistics: mean = 1.899 s (CI deltas: -16.704 ms, +34.261 ms), sd = 90.787 ms (CI deltas: -45.888 ms, +61.153 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.860 s, interquantileRange = 82.129 ms --2 are EXTREME (on the high side): #58 = 2.267 s, #59 = 2.385 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 6.169182765745209% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.29963002288214% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16384.0, muB = 1.8992949577833333, sigmaB = 0.09078681445066508, muA = 1.159237645131429E-4, sigmaA = 7.09271987895821E-4, tMin = 0.0, muGMin = 5.796188225657145E-5, sigmaG = 1.4490470564142863E-5, cMax1 = 426, cMax2 = 108, cMax = 108, cOutMin = 108, varOutMin = 0.008184519463929786, muG(cOutMin) = 5.8159294975303484E-5, U(cOutMin) = 0.008821243266345313 Custom Buffer Buffered Streams results : 122.318 us action statistics: first = 231.175 us, mean = 122.318 us (CI deltas: -289.189 ns, +275.445 ns), sd = 142.402 us (CI deltas: -20.394 us, +26.017 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16384 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16384 --block statistics: mean = 2.004 s (CI deltas: -4.738 ms, +4.513 ms), sd = 18.227 ms (CI deltas: -2.610 ms, +3.330 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[2] is -1.99317713527799 sigma below its mean, r[4] is 2.9165425097658835 sigma above its mean, r[7] is 2.5689803847431127 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 31.56336759661175% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 83.4824583406552% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16384.0, muB = 2.004061636116667, sigmaB = 0.018227470779191988, muA = 1.223182150950114E-4, sigmaA = 1.424021154624374E-4, tMin = 0.0, muGMin = 6.11591075475057E-5, sigmaG = 1.5289776886876427E-5, cMax1 = 6982, cMax2 = 2572, cMax = 2572, cOutMin = 2572, varOutMin = 2.773626964600375E-4, muG(cOutMin) = 6.116730318140701E-5, U(cOutMin) = 4.5070717129668474E-4 NIO Buffer results : 133.045 us action statistics: first = 245.325 us, mean = 133.045 us (CI deltas: -1.780 us, +2.937 us), sd = 803.643 us (CI deltas: -242.193 us, +478.778 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 8192 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 8192 --block statistics: mean = 1.090 s (CI deltas: -14.584 ms, +24.057 ms), sd = 72.737 ms (CI deltas: -21.921 ms, +43.334 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.061 s, interquantileRange = 16.585 ms --11 are EXTREME (on the high side): #49 = 1.131 s, #50 = 1.140 s, #51 = 1.152 s, #52 = 1.217 s, #53 = 1.217 s, #54 = 1.226 s, #55 = 1.226 s, #56 = 1.233 s, #57 = 1.240 s, #58 = 1.241 s, #59 = 1.446 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[7] is 3.072970893253475 sigma above its mean, r[13] is 2.0108723153805155 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 5.832968895133172% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.28636229670413% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 8192.0, muB = 1.08990111995, sigmaB = 0.07273745402044962, muA = 1.3304457030639648E-4, sigmaA = 8.036429216266348E-4, tMin = 0.0, muGMin = 6.652228515319824E-5, sigmaG = 1.663057128829956E-5, cMax1 = 218, cMax2 = 55, cMax = 55, cOutMin = 55, varOutMin = 0.005252980521811513, muG(cOutMin) = 6.698740637264433E-5, U(cOutMin) = 0.009905901714468963 NIO Transfer results : 92.452 us action statistics: first = 249.807 us, mean = 92.452 us (CI deltas: -1.573 us, +2.640 us), sd = 1.013 ms (CI deltas: -333.867 us, +410.088 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16384 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16384 --block statistics: mean = 1.515 s (CI deltas: -25.766 ms, +43.247 ms), sd = 129.621 ms (CI deltas: -42.735 ms, +52.491 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.468 s, interquantileRange = 8.209 ms --8 are EXTREME (on the high side): #52 = 1.501 s, #53 = 1.756 s, #54 = 1.770 s, #55 = 1.806 s, #56 = 1.815 s, #57 = 1.840 s, #58 = 1.999 s, #59 = 2.022 s --1 are mild (on the high side): #51 = 1.487 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 3.858757801729827% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.77951114524674% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16384.0, muB = 1.514729116466667, sigmaB = 0.12962064345103394, muA = 9.245172829996747E-5, sigmaA = 0.0010126612769612027, tMin = 0.0, muGMin = 4.622586414998373E-5, sigmaG = 1.1556466037495933E-5, cMax1 = 135, cMax2 = 34, cMax = 34, cOutMin = 34, varOutMin = 0.016764465749014853, muG(cOutMin) = 4.6275701608021585E-5, U(cOutMin) = 0.02229768809339747 Start benchmark with big-binary Native Copy results : 33.385 ms action statistics: first = 127.257 ms, mean = 33.385 ms (CI deltas: -501.766 us, +562.730 us), sd = 11.850 ms (CI deltas: -1.513 ms, +1.668 ms) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32 --block statistics: mean = 1.068 s (CI deltas: -16.056 ms, +18.007 ms), sd = 67.036 ms (CI deltas: -8.561 ms, +9.438 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 5: r[1] is 7.48572013205971 sigma above its mean, r[2] is 4.2391275472747925 sigma above its mean, r[3] is 3.1787338534955065 sigma above its mean, r[4] is 2.600905568851213 sigma above its mean, r[5] is 2.205656264749953 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 6.266120181878562% of the measured sd ---------- --action sd values likely INFLATED by outliers --they cause at least 32.90962895042175% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32.0, muB = 1.0683243294333333, sigmaB = 0.06703583849523029, muA = 0.03338513529479167, sigmaA = 0.011850373995625885, tMin = 0.0, muGMin = 0.016692567647395833, sigmaG = 0.004173141911848958, cMax1 = 28, cMax2 = 21, cMax = 21, cOutMin = 21, varOutMin = 0.001478894104592391, muG(cOutMin) = 0.01736428725633232, U(cOutMin) = 0.0417770080768418 Buffered Streams results : 66.586 ms action statistics: first = 65.816 ms, mean = 66.586 ms (CI deltas: -740.720 us, +1.830 ms), sd = 16.441 ms (CI deltas: -14.052 ms, +12.615 ms) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.065 s (CI deltas: -11.852 ms, +29.286 ms), sd = 65.764 ms (CI deltas: -56.210 ms, +50.462 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.053 s, interquantileRange = 12.874 ms --2 are EXTREME (on the high side): #58 = 1.395 s, #59 = 1.436 s --1 are mild (on the high side): #57 = 1.087 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 6.3784333694037745% of the measured sd ---------- --action sd values likely INFLATED by outliers --they cause at least 23.39822830074312% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.065371127, sigmaB = 0.06576436710160627, muA = 0.0665856954375, sigmaA = 0.016441091775401567, tMin = 0.0, muGMin = 0.03329284771875, sigmaG = 0.0083232119296875, cMax1 = 15, cMax2 = 12, cMax = 12, cOutMin = 12, varOutMin = 0.001011962138242216, muG(cOutMin) = 0.03903625617493674, U(cOutMin) = 0.07576884185835442 Custom Buffer Streams results : 9.702 ms action statistics: first = 10.220 ms, mean = 9.702 ms (CI deltas: -12.456 us, +16.200 us), sd = 630.878 us (CI deltas: -126.168 us, +144.551 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 128 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 128 --block statistics: mean = 1.242 s (CI deltas: -1.594 ms, +2.074 ms), sd = 7.138 ms (CI deltas: -1.427 ms, +1.635 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.240 s, interquantileRange = 6.146 ms --1 are EXTREME (on the high side): #59 = 1.262 s --6 are mild (on the high side): #53 = 1.254 s, #54 = 1.257 s, #55 = 1.257 s, #56 = 1.257 s, #57 = 1.257 s, #58 = 1.261 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is 5.037597345292224 sigma above its mean, r[2] is 3.1511464982684223 sigma above its mean, r[3] is 2.344782565167017 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 63.45000906571069% of the measured sd Custom Buffer Buffered Streams results : 9.647 ms action statistics: first = 10.131 ms, mean = 9.647 ms (CI deltas: -8.166 us, +11.450 us), sd = 431.571 us (CI deltas: -103.250 us, +134.525 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 128 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 128 --block statistics: mean = 1.235 s (CI deltas: -1.045 ms, +1.466 ms), sd = 4.883 ms (CI deltas: -1.168 ms, +1.522 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.233 s, interquantileRange = 3.915 ms --4 are EXTREME (on the high side): #56 = 1.247 s, #57 = 1.248 s, #58 = 1.248 s, #59 = 1.252 s --3 are mild (on the high side): #53 = 1.241 s, #54 = 1.242 s, #55 = 1.245 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is 3.9440254747672285 sigma above its mean, r[2] is 2.4010124914650333 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 92.49045288607005% of the measured sd NIO Buffer results : 11.013 ms action statistics: first = 11.639 ms, mean = 11.013 ms (CI deltas: -7.984 us, +10.208 us), sd = 404.543 us (CI deltas: -73.632 us, +108.434 us) WARNING: execution times have mild outliers, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 128 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 128 --block statistics: mean = 1.410 s (CI deltas: -1.022 ms, +1.307 ms), sd = 4.577 ms (CI deltas: -833.054 us, +1.227 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.408 s, interquantileRange = 4.638 ms --6 are mild (on the high side): #54 = 1.418 s, #55 = 1.418 s, #56 = 1.419 s, #57 = 1.419 s, #58 = 1.419 s, #59 = 1.425 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is 4.329776592753662 sigma above its mean, r[2] is 1.972477500959068 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 100.0% of the measured sd NIO Transfer results : 7.016 ms action statistics: first = 8.800 ms, mean = 7.016 ms (CI deltas: -4.418 us, +7.980 us), sd = 363.299 us (CI deltas: -136.830 us, +218.441 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 256 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 256 --block statistics: mean = 1.796 s (CI deltas: -1.131 ms, +2.043 ms), sd = 5.813 ms (CI deltas: -2.189 ms, +3.495 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.794 s, interquantileRange = 4.104 ms --3 are EXTREME (on the high side): #57 = 1.810 s, #58 = 1.817 s, #59 = 1.825 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 93.6993085788283% of the measured sd Start benchmark with fat-binary Native Copy results : 94.494 ms action statistics: first = 1.130 s, mean = 94.494 ms (CI deltas: -71.622 us, +60.132 us), sd = 1.035 ms (CI deltas: -186.454 us, +248.240 us) WARNING: execution times have mild outliers, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.512 s (CI deltas: -1.146 ms, +962.107 us), sd = 4.141 ms (CI deltas: -745.817 us, +992.958 us) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.513 s, interquantileRange = 5.083 ms --2 are mild (on the low side): #0 = 1.500 s, #1 = 1.501 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 100.0% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859374999999996% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.5119111416000004, sigmaB = 0.0041412389982742265, muA = 0.09449444635000002, sigmaA = 0.0010353097495685566, tMin = 0.0, muGMin = 0.04724722317500001, sigmaG = 0.0010353097495685566, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 1, varOutMin = 1.004874635204725E-6, muG(cOutMin) = 0.09442761739302927, U(cOutMin) = 0.09549688070456132 Buffered Streams results : 677.607 ms action statistics: first = 666.248 ms, mean = 677.607 ms (CI deltas: -5.577 ms, +9.600 ms), sd = 39.981 ms (CI deltas: -13.700 ms, +18.275 ms) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 2 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 2 --block statistics: mean = 1.355 s (CI deltas: -11.153 ms, +19.200 ms), sd = 56.542 ms (CI deltas: -19.375 ms, +25.845 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.337 s, interquantileRange = 12.930 ms --7 are EXTREME (on the high side): #53 = 1.417 s, #54 = 1.459 s, #55 = 1.473 s, #56 = 1.492 s, #57 = 1.500 s, #58 = 1.562 s, #59 = 1.597 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is 2.6824694648039697 sigma above its mean, r[2] is 3.2998126537344077 sigma above its mean, r[3] is 2.4308339766712996 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 8.367369690594236% of the measured sd Custom Buffer Streams results : 104.376 ms action statistics: first = 106.399 ms, mean = 104.376 ms (CI deltas: -111.982 us, +166.662 us), sd = 2.131 ms (CI deltas: -594.416 us, +727.045 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.670 s (CI deltas: -1.792 ms, +2.667 ms), sd = 8.524 ms (CI deltas: -2.378 ms, +2.908 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.667 s, interquantileRange = 7.110 ms --3 are EXTREME (on the high side): #57 = 1.694 s, #58 = 1.699 s, #59 = 1.699 s --2 are mild (on the high side): #55 = 1.690 s, #56 = 1.691 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 61.61360334065933% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859375% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.6700188163666665, sigmaB = 0.008523898716538806, muA = 0.10437617602291666, sigmaA = 0.0021309746791347015, tMin = 0.0, muGMin = 0.05218808801145833, sigmaG = 0.0021309746791347015, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 15, varOutMin = 4.257237265418666E-6, muG(cOutMin) = 0.09638502097616153, U(cOutMin) = 0.10490891969270033 Custom Buffer Buffered Streams results : 104.014 ms action statistics: first = 109.021 ms, mean = 104.014 ms (CI deltas: -131.593 us, +155.585 us), sd = 2.264 ms (CI deltas: -375.888 us, +530.814 us) WARNING: execution times have mild outliers, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.664 s (CI deltas: -2.105 ms, +2.489 ms), sd = 9.057 ms (CI deltas: -1.504 ms, +2.123 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.662 s, interquantileRange = 11.583 ms --2 are mild (on the high side): #58 = 1.688 s, #59 = 1.691 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 57.888390936795844% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859374999999995% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.6642291910833331, sigmaB = 0.009056685699089446, muA = 0.10401432444270832, sigmaA = 0.0022641714247723615, tMin = 0.0, muGMin = 0.05200716222135416, sigmaG = 0.0022641714247723615, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 1, varOutMin = 4.8060677257084695E-6, muG(cOutMin) = 0.10386817280568995, U(cOutMin) = 0.10620659899798383 NIO Buffer results : 118.624 ms action statistics: first = 118.798 ms, mean = 118.624 ms (CI deltas: -514.232 us, +1.028 ms), sd = 11.045 ms (CI deltas: -4.836 ms, +6.751 ms) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.898 s (CI deltas: -8.228 ms, +16.451 ms), sd = 44.181 ms (CI deltas: -19.346 ms, +27.006 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.885 s, interquantileRange = 12.704 ms --5 are EXTREME (on the high side): #55 = 1.969 s, #56 = 1.974 s, #57 = 2.021 s, #58 = 2.071 s, #59 = 2.122 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is 5.599725700968135 sigma above its mean, r[2] is 2.904396609211641 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 12.67263725632731% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859375% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.8979880862333336, sigmaB = 0.04418083748593523, muA = 0.11862425538958335, sigmaA = 0.011045209371483808, tMin = 0.0, muGMin = 0.059312127694791675, sigmaG = 0.011045209371483808, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 15, varOutMin = 1.1437185943116911E-4, muG(cOutMin) = 0.07720472024651907, U(cOutMin) = 0.1213855577324543 NIO Transfer results : 79.243 ms action statistics: first = 78.922 ms, mean = 79.243 ms (CI deltas: -61.524 us, +131.524 us), sd = 1.363 ms (CI deltas: -589.553 us, +1.122 ms) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.268 s (CI deltas: -984.380 us, +2.104 ms), sd = 5.454 ms (CI deltas: -2.358 ms, +4.489 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.267 s, interquantileRange = 2.766 ms --4 are EXTREME (on the high side): #56 = 1.279 s, #57 = 1.280 s, #58 = 1.284 s, #59 = 1.300 s --1 are mild (on the high side): #55 = 1.273 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 83.90779740930589% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859374999999997% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.2678935639833335, sigmaB = 0.00545372561883194, muA = 0.07924334774895835, sigmaA = 0.001363431404707985, tMin = 0.0, muGMin = 0.03962167387447917, sigmaG = 0.001363431404707985, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 1, varOutMin = 1.7427611206349891E-6, muG(cOutMin) = 0.07915533863022284, U(cOutMin) = 0.0805634845299909 Start benchmark with enormous-binary Native Copy results : 18.917 s action statistics: first = 38.916 s, mean = 18.917 s (CI deltas: -2.679 s, +3.026 s), sd = 11.222 s (CI deltas: -1.565 s, +1.661 s) ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 Custom Buffer Streams results : 23.557 s action statistics: first = 18.933 s, mean = 23.557 s (CI deltas: -3.412 s, +3.545 s), sd = 13.808 s (CI deltas: -1.317 s, +1.439 s) WARNING: execution times may have serial correlation ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is -3.6168272252120492 sigma below its mean, r[10] is -2.0290766402705684 sigma below its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) Custom Buffer Buffered Streams results : 22.543 s action statistics: first = 38.412 s, mean = 22.543 s (CI deltas: -3.401 s, +4.046 s), sd = 14.706 s (CI deltas: -2.353 s, +3.298 s) WARNING: execution times may be inaccurate, execution times may have serial correlation ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES MAY BE TOO SMALL --a final JVM cleanup took an extra 1.3803433713444262% of the total execution time --this time should be included in the measurements ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is -2.9620137632253822 sigma below its mean, r[2] is 2.1621113198237807 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) NIO Buffer results : 22.900 s action statistics: first = 17.706 s, mean = 22.900 s (CI deltas: -3.287 s, +3.294 s), sd = 13.012 s (CI deltas: -1.139 s, +1.182 s) WARNING: execution times may be inaccurate ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES MAY BE TOO SMALL --a final JVM cleanup took an extra 1.5872150510496632% of the total execution time --this time should be included in the measurements NIO Transfer results : 24.197 s action statistics: first = 19.988 s, mean = 24.197 s (CI deltas: -3.036 s, +3.345 s), sd = 12.629 s (CI deltas: -1.894 s, +2.313 s) ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 Bench between two disks Start benchmark with little-text Native Copy results : 23.314 ms action statistics: first = 99.895 ms, mean = 23.314 ms (CI deltas: -13.250 us, +12.998 us), sd = 414.328 us (CI deltas: -63.509 us, +86.217 us) WARNING: execution times have mild outliers, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 64 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 64 --block statistics: mean = 1.492 s (CI deltas: -848.013 us, +831.878 us), sd = 3.315 ms (CI deltas: -508.074 us, +689.738 us) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.492 s, interquantileRange = 3.859 ms --1 are mild (on the low side): #0 = 1.483 s --1 are mild (on the high side): #59 = 1.500 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 4: r[1] is 3.465028430779666 sigma above its mean, r[2] is 2.9455646510323517 sigma above its mean, r[3] is 2.601645365581389 sigma above its mean, r[4] is 2.2235371252300093 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 100.0% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 1.5380859375% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 64.0, muB = 1.4920867005333334, sigmaB = 0.003314621148499998, muA = 0.023313854695833334, sigmaA = 4.1432764356249977E-4, tMin = 0.0, muGMin = 0.011656927347916667, sigmaG = 4.1432764356249977E-4, cMax1 = 63, cMax2 = 63, cMax = 63, cOutMin = 63, varOutMin = 1.689850931541155E-7, muG(cOutMin) = 0.02005102450277865, U(cOutMin) = 0.023365645651278646 Naive Streams results : 10.697 ms action statistics: first = 10.670 ms, mean = 10.697 ms (CI deltas: -15.077 us, +22.098 us), sd = 807.740 us (CI deltas: -223.486 us, +353.110 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 128 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 128 --block statistics: mean = 1.369 s (CI deltas: -1.930 ms, +2.829 ms), sd = 9.139 ms (CI deltas: -2.528 ms, +3.995 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.368 s, interquantileRange = 7.936 ms --1 are EXTREME (on the high side): #59 = 1.408 s --4 are mild (on the high side): #55 = 1.388 s, #56 = 1.388 s, #57 = 1.394 s, #58 = 1.394 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 52.03615589794386% of the measured sd Naive Readers results : 478.026 us action statistics: first = 857.627 us, mean = 478.026 us (CI deltas: -483.944 ns, +503.025 ns), sd = 124.763 us (CI deltas: -18.431 us, +22.377 us) WARNING: SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 4096 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 4096 --block statistics: mean = 1.958 s (CI deltas: -1.982 ms, +2.060 ms), sd = 7.985 ms (CI deltas: -1.180 ms, +1.432 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 71.2189148324978% of the measured sd ---------- --action sd values likely INFLATED by outliers --they cause at least 19.6536637574315% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 4096.0, muB = 1.9579941504166667, sigmaB = 0.007984809324656568, muA = 4.78025915629069E-4, sigmaA = 1.2476264569775887E-4, tMin = 0.0, muGMin = 2.390129578145345E-4, sigmaG = 5.975323945363363E-5, cMax1 = 3838, cMax2 = 3251, cMax = 3251, cOutMin = 3251, varOutMin = 1.2530621768814141E-5, muG(cOutMin) = 2.3916885345029145E-4, U(cOutMin) = 5.401096491083268E-4 Buffered Streams results : 105.939 us action statistics: first = 202.067 us, mean = 105.939 us (CI deltas: -331.273 ns, +314.931 ns), sd = 163.702 us (CI deltas: -15.817 us, +17.237 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16384 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16384 --block statistics: mean = 1.736 s (CI deltas: -5.428 ms, +5.160 ms), sd = 20.954 ms (CI deltas: -2.025 ms, +2.206 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 6: r[1] is -4.542565967750287 sigma below its mean, r[2] is 2.2365999569983153 sigma above its mean, r[5] is 2.5879222294767414 sigma above its mean, r[6] is -2.751364806205907 sigma below its mean, r[7] is 2.7310950470955624 sigma above its mean, r[12] is 2.0862470593543825 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 25.552083521410484% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 89.93679896550015% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16384.0, muB = 1.7356996562500002, sigmaB = 0.020953914965012227, muA = 1.0593869972229005E-4, sigmaA = 1.6370246066415802E-4, tMin = 0.0, muGMin = 5.2969349861145026E-5, sigmaG = 1.3242337465286257E-5, cMax1 = 4851, cMax2 = 1561, cMax = 1561, cOutMin = 1561, varOutMin = 3.94882402521632E-4, muG(cOutMin) = 5.297244886502265E-5, U(cOutMin) = 6.088975315334846E-4 Buffered Readers results : 139.948 us action statistics: first = 249.192 us, mean = 139.948 us (CI deltas: -294.547 ns, +262.417 ns), sd = 98.931 us (CI deltas: -15.356 us, +17.492 us) WARNING: execution times have mild outliers, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 8192 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 8192 --block statistics: mean = 1.146 s (CI deltas: -2.413 ms, +2.150 ms), sd = 8.954 ms (CI deltas: -1.390 ms, +1.583 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.147 s, interquantileRange = 10.478 ms --1 are mild (on the low side): #0 = 1.126 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 48.596694074238016% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 64.82850533871286% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 8192.0, muB = 1.1464562479000002, sigmaB = 0.008954188305697313, muA = 1.399482724487305E-4, sigmaA = 9.893073860906019E-5, tMin = 0.0, muGMin = 6.997413622436525E-5, sigmaG = 1.7493534056091312E-5, cMax1 = 5481, cMax2 = 2769, cMax = 2769, cOutMin = 2769, varOutMin = 5.1977867227185294E-5, muG(cOutMin) = 6.99912846907414E-5, U(cOutMin) = 2.7695684760639563E-4 Custom Buffer Streams results : 47.052 us action statistics: first = 140.366 us, mean = 47.052 us (CI deltas: -407.568 ns, +445.724 ns), sd = 304.624 us (CI deltas: -25.540 us, +15.739 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32768 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32768 --block statistics: mean = 1.542 s (CI deltas: -13.355 ms, +14.605 ms), sd = 55.143 ms (CI deltas: -4.623 ms, +2.849 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 6: r[1] is -4.467277130093971 sigma below its mean, r[3] is 2.1450913545070587 sigma above its mean, r[4] is -3.088491410644747 sigma below its mean, r[5] is 2.927735119334014 sigma above its mean, r[8] is 2.0619774051074424 sigma above its mean, r[13] is 2.157849615697315 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 9.151199495883104% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.37112188350449% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32768.0, muB = 1.5417906424833339, sigmaB = 0.05514280912191591, muA = 4.7051716384379085E-5, sigmaA = 3.0462386143579803E-4, tMin = 0.0, muGMin = 2.3525858192189543E-5, sigmaG = 5.881464548047386E-6, cMax1 = 763, cMax2 = 194, cMax = 194, cOutMin = 194, varOutMin = 0.0030216069160910897, muG(cOutMin) = 2.3547338764002625E-5, U(cOutMin) = 0.003993606338065528 Custom Buffer Readers results : 62.772 us action statistics: first = 168.706 us, mean = 62.772 us (CI deltas: -224.188 ns, +193.403 ns), sd = 105.970 us (CI deltas: -15.608 us, +15.959 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16384 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16384 --block statistics: mean = 1.028 s (CI deltas: -3.673 ms, +3.169 ms), sd = 13.564 ms (CI deltas: -1.998 ms, +2.043 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[2] is -2.908238424565585 sigma below its mean, r[4] is 3.9426574842450597 sigma above its mean, r[8] is 2.0163446172477437 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 30.384699046412866% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 91.43758647306689% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16384.0, muB = 1.0284607298333333, sigmaB = 0.01356416688401582, muA = 6.277226134236653E-5, sigmaA = 1.0597005378137359E-4, tMin = 0.0, muGMin = 3.1386130671183266E-5, sigmaG = 7.846532667795816E-6, cMax1 = 4268, cMax2 = 1327, cMax = 1327, cOutMin = 1327, varOutMin = 1.6823292773988964E-4, muG(cOutMin) = 3.139229217606318E-5, U(cOutMin) = 4.188296808879804E-4 Custom Buffer Buffered Streams results : 53.655 us action statistics: first = 155.451 us, mean = 53.655 us (CI deltas: -141.217 ns, +155.566 ns), sd = 105.885 us (CI deltas: -13.447 us, +17.910 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32768 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32768 --block statistics: mean = 1.758 s (CI deltas: -4.627 ms, +5.098 ms), sd = 19.167 ms (CI deltas: -2.434 ms, +3.242 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 5: r[1] is -3.3366043920263477 sigma below its mean, r[4] is -2.1119577337943474 sigma below its mean, r[5] is 3.6058763689535582 sigma above its mean, r[6] is -2.597055862102113 sigma below its mean, r[11] is -2.171047498242307 sigma below its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 28.114139814699122% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 93.59424400870002% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32768.0, muB = 1.7581706950000002, sigmaB = 0.01916725518770999, muA = 5.3655111541748054E-5, sigmaA = 1.058851259372091E-4, tMin = 0.0, muGMin = 2.6827555770874027E-5, sigmaG = 6.706888942718507E-6, cMax1 = 6711, cMax2 = 1983, cMax = 1983, cOutMin = 1983, varOutMin = 3.438499698870595E-4, muG(cOutMin) = 2.6832152002323302E-5, U(cOutMin) = 4.7006701745258567E-4 Custom Buffer Buffered Readers results : 80.187 us action statistics: first = 200.522 us, mean = 80.187 us (CI deltas: -161.173 ns, +148.753 ns), sd = 78.226 us (CI deltas: -11.367 us, +13.146 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16384 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16384 --block statistics: mean = 1.314 s (CI deltas: -2.641 ms, +2.437 ms), sd = 10.013 ms (CI deltas: -1.455 ms, +1.683 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[4] is 3.722991596598249 sigma above its mean, r[8] is 2.4366612857469083 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 46.521882001685015% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 77.96129981227892% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16384.0, muB = 1.3137823124500005, sigmaB = 0.010012871361154103, muA = 8.01869087188721E-5, sigmaA = 7.822555750901643E-5, tMin = 0.0, muGMin = 4.009345435943605E-5, sigmaG = 1.0023363589859013E-5, cMax1 = 8427, cMax2 = 3443, cMax = 3443, cOutMin = 3443, varOutMin = 7.81621225814606E-5, muG(cOutMin) = 4.0100343430955934E-5, U(cOutMin) = 2.30857905347081E-4 NIO Buffer results : 53.469 us action statistics: first = 226.375 us, mean = 53.469 us (CI deltas: -723.085 ns, +727.634 ns), sd = 519.542 us (CI deltas: -17.112 us, +25.566 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32768 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32768 --block statistics: mean = 1.752 s (CI deltas: -23.694 ms, +23.843 ms), sd = 94.047 ms (CI deltas: -3.098 ms, +4.628 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is -6.160114442662101 sigma below its mean, r[2] is 3.6564609614422046 sigma above its mean, r[3] is -2.5315624017263163 sigma below its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 5.719845652829488% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.72108622996669% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32768.0, muB = 1.7520669566999998, sigmaB = 0.09404706496499837, muA = 5.3468840231323236E-5, sigmaA = 5.195415420893916E-4, tMin = 0.0, muGMin = 2.6734420115661618E-5, sigmaG = 6.6836050289154045E-6, cMax1 = 343, cMax2 = 86, cMax = 86, cOutMin = 86, varOutMin = 0.008820180922746602, muG(cOutMin) = 2.6819936571861195E-5, U(cOutMin) = 0.010180671972772469 NIO Transfer results : 52.285 us action statistics: first = 166.936 us, mean = 52.285 us (CI deltas: -1.033 us, +1.319 us), sd = 832.219 us (CI deltas: -142.655 us, +145.442 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32768 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32768 --block statistics: mean = 1.713 s (CI deltas: -33.833 ms, +43.206 ms), sd = 150.648 ms (CI deltas: -25.823 ms, +26.328 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.630 s, interquantileRange = 102.979 ms --2 are EXTREME (on the high side): #58 = 2.052 s, #59 = 2.125 s --13 are mild (on the high side): #45 = 1.926 s, #46 = 1.928 s, #47 = 1.935 s, #48 = 1.939 s, #49 = 1.942 s, #50 = 1.945 s, #51 = 1.946 s, #52 = 1.946 s, #53 = 1.948 s, #54 = 1.950 s, #55 = 1.961 s, #56 = 1.979 s, #57 = 2.024 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is -2.437894375530916 sigma below its mean, r[2] is -2.048273213969651 sigma below its mean, r[7] is 2.0081640019682805 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 3.5310584589991416% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.89618845112436% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32768.0, muB = 1.7132719252833333, sigmaB = 0.15064768014692814, muA = 5.2284909829203286E-5, sigmaA = 8.322187203274602E-4, tMin = 0.0, muGMin = 2.6142454914601643E-5, sigmaG = 6.535613728650411E-6, cMax1 = 128, cMax2 = 32, cMax = 32, cOutMin = 32, varOutMin = 0.02267116378963784, muG(cOutMin) = 2.6266168446169748E-5, U(cOutMin) = 0.026669457344672513 Start benchmark with medium-text Native Copy results : 28.067 ms action statistics: first = 78.281 ms, mean = 28.067 ms (CI deltas: -492.021 us, +586.039 us), sd = 12.028 ms (CI deltas: -1.820 ms, +2.039 ms) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32 --block statistics: mean = 898.130 ms (CI deltas: -15.745 ms, +18.753 ms), sd = 68.038 ms (CI deltas: -10.296 ms, +11.534 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 5: r[1] is 7.4225551406254064 sigma above its mean, r[2] is 4.179782972885591 sigma above its mean, r[3] is 3.1169437671263998 sigma above its mean, r[4] is 2.536245455231278 sigma above its mean, r[5] is 2.1557700158943787 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 5.660694994596901% of the measured sd ---------- --action sd values likely INFLATED by outliers --they cause at least 42.121466772196136% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32.0, muB = 0.8981303342500001, sigmaB = 0.06803840656313945, muA = 0.028066572945312503, sigmaA = 0.0120276046654808, tMin = 0.0, muGMin = 0.014033286472656252, sigmaG = 0.003508321618164063, cMax1 = 27, cMax2 = 18, cMax = 18, cOutMin = 18, varOutMin = 0.0019498973723164137, muG(cOutMin) = 0.014684786530124451, U(cOutMin) = 0.03847462904601432 Naive Streams results : 108.138 ms action statistics: first = 108.414 ms, mean = 108.138 ms (CI deltas: -151.748 us, +149.741 us), sd = 2.394 ms (CI deltas: -323.348 us, +434.796 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.730 s (CI deltas: -2.428 ms, +2.396 ms), sd = 9.576 ms (CI deltas: -1.293 ms, +1.739 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is 5.351240608534129 sigma above its mean, r[2] is 2.890702081068802 sigma above its mean, r[3] is 2.4542859076186443 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 55.82253921650526% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859374999999999% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.7302149513000002, sigmaB = 0.009576231791343709, muA = 0.10813843445625002, sigmaA = 0.002394057947835927, tMin = 0.0, muGMin = 0.05406921722812501, sigmaG = 0.002394057947835927, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 15, varOutMin = 5.373293866496597E-6, muG(cOutMin) = 0.0991607171518653, U(cOutMin) = 0.108736948943209 Naive Readers results : 4.436 ms action statistics: first = 7.250 ms, mean = 4.436 ms (CI deltas: -4.551 us, +5.557 us), sd = 318.004 us (CI deltas: -55.138 us, +76.505 us) WARNING: execution times have mild outliers, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 256 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 256 --block statistics: mean = 1.136 s (CI deltas: -1.165 ms, +1.422 ms), sd = 5.088 ms (CI deltas: -882.205 us, +1.224 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.135 s, interquantileRange = 6.041 ms --2 are mild (on the high side): #58 = 1.150 s, #59 = 1.150 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 85.1141460843614% of the measured sd Buffered Streams results : 668.937 us action statistics: first = 781.404 us, mean = 668.937 us (CI deltas: -1.254 us, +1.866 us), sd = 269.734 us (CI deltas: -68.766 us, +94.267 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 2048 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 2048 --block statistics: mean = 1.370 s (CI deltas: -2.567 ms, +3.822 ms), sd = 12.207 ms (CI deltas: -3.112 ms, +4.266 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.365 s, interquantileRange = 4.021 ms --11 are EXTREME (on the high side): #49 = 1.380 s, #50 = 1.382 s, #51 = 1.383 s, #52 = 1.385 s, #53 = 1.391 s, #54 = 1.391 s, #55 = 1.393 s, #56 = 1.396 s, #57 = 1.401 s, #58 = 1.408 s, #59 = 1.416 s --2 are mild (on the high side): #47 = 1.377 s, #48 = 1.378 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is 2.5618797583492214 sigma above its mean, r[14] is 2.0508313081137155 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 38.96823872992087% of the measured sd ---------- --action sd values likely INFLATED by outliers --they cause at least 37.09908498463318% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 2048.0, muB = 1.3699833319333334, sigmaB = 0.01220677870090627, muA = 6.68937173795573E-4, sigmaA = 2.6973424987044805E-4, tMin = 0.0, muGMin = 3.344685868977865E-4, sigmaG = 8.361714672444662E-5, cMax1 = 1764, cMax2 = 1259, cMax = 1259, cOutMin = 1259, varOutMin = 5.52796571370949E-5, muG(cOutMin) = 3.345739281808262E-4, U(cOutMin) = 8.784785564723285E-4 Buffered Readers results : 851.767 us action statistics: first = 979.949 us, mean = 851.767 us (CI deltas: -1.040 us, +1.082 us), sd = 190.267 us (CI deltas: -19.342 us, +23.561 us) WARNING: SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 2048 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 2048 --block statistics: mean = 1.744 s (CI deltas: -2.131 ms, +2.215 ms), sd = 8.610 ms (CI deltas: -875.339 us, +1.066 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 62.33778246244223% of the measured sd ---------- --action sd values likely INFLATED by outliers --they cause at least 15.168488147666201% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 2048.0, muB = 1.7444194408000004, sigmaB = 0.008610498921994891, muA = 8.517673050781252E-4, sigmaA = 1.902669430356889E-4, tMin = 0.0, muGMin = 4.258836525390626E-4, sigmaG = 1.0647091313476565E-4, cMax1 = 1952, cMax2 = 1721, cMax = 1721, cOutMin = 1721, varOutMin = 1.1246022030939382E-5, muG(cOutMin) = 4.263235556031002E-4, U(cOutMin) = 9.326040895512996E-4 Custom Buffer Streams results : 113.295 us action statistics: first = 228.352 us, mean = 113.295 us (CI deltas: -685.770 ns, +946.737 ns), sd = 403.076 us (CI deltas: -89.664 us, +87.262 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16384 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16384 --block statistics: mean = 1.856 s (CI deltas: -11.236 ms, +15.511 ms), sd = 51.594 ms (CI deltas: -11.477 ms, +11.170 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.831 s, interquantileRange = 19.854 ms --11 are EXTREME (on the high side): #49 = 1.933 s, #50 = 1.937 s, #51 = 1.941 s, #52 = 1.948 s, #53 = 1.950 s, #54 = 1.959 s, #55 = 1.967 s, #56 = 1.970 s, #57 = 1.983 s, #58 = 1.988 s, #59 = 2.001 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[5] is 3.4066358409623887 sigma above its mean, r[11] is 2.130761264862964 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 10.731831997035322% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 97.94647162263628% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16384.0, muB = 1.8562332542333333, sigmaB = 0.051593693079421125, muA = 1.1329548670857747E-4, sigmaA = 4.0307572718297754E-4, tMin = 0.0, muGMin = 5.6647743354288736E-5, sigmaG = 1.4161935838572184E-5, cMax1 = 1200, cMax2 = 317, cMax = 317, cOutMin = 317, varOutMin = 0.0026072461054788094, muG(cOutMin) = 5.6712502168373855E-5, U(cOutMin) = 0.0029811781763219893 Custom Buffer Readers results : 202.264 us action statistics: first = 321.241 us, mean = 202.264 us (CI deltas: -480.180 ns, +494.731 ns), sd = 173.553 us (CI deltas: -18.650 us, +23.713 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 8192 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 8192 --block statistics: mean = 1.657 s (CI deltas: -3.934 ms, +4.053 ms), sd = 15.708 ms (CI deltas: -1.688 ms, +2.146 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 4: r[1] is -4.339268170826608 sigma below its mean, r[5] is 2.9665980324840096 sigma above its mean, r[6] is -2.380520313050812 sigma below its mean, r[7] is 2.4204751377564855 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 33.30285655842797% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 73.17978037038263% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 8192.0, muB = 1.6569426937, sigmaB = 0.015708201518917796, muA = 2.0226351241455078E-4, sigmaA = 1.7355274709799373E-4, tMin = 0.0, muGMin = 1.0113175620727539E-4, sigmaG = 2.5282939051818848E-5, cMax1 = 4736, cMax2 = 2101, cMax = 2101, cOutMin = 2101, varOutMin = 1.805693480601473E-4, muG(cOutMin) = 1.0114132819362619E-4, U(cOutMin) = 4.954263987018671E-4 Custom Buffer Buffered Streams results : 121.764 us action statistics: first = 239.138 us, mean = 121.764 us (CI deltas: -264.820 ns, +232.579 ns), sd = 125.577 us (CI deltas: -17.454 us, +17.181 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16384 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16384 --block statistics: mean = 1.995 s (CI deltas: -4.339 ms, +3.811 ms), sd = 16.074 ms (CI deltas: -2.234 ms, +2.199 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[2] is -2.5083788716016047 sigma below its mean, r[3] is 3.1165824450940662 sigma above its mean, r[7] is 2.4378665111810975 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 35.71110475676899% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 79.82737170710257% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16384.0, muB = 1.9949799002333335, sigmaB = 0.01607386006985032, muA = 1.2176390992635092E-4, sigmaA = 1.2557703179570562E-4, tMin = 0.0, muGMin = 6.088195496317546E-5, sigmaG = 1.5220488740793865E-5, cMax1 = 7970, cMax2 = 3148, cMax = 3148, cOutMin = 3148, varOutMin = 2.062491640807901E-4, muG(cOutMin) = 6.0886424992285084E-5, U(cOutMin) = 3.7772782053222616E-4 Custom Buffer Buffered Readers results : 232.448 us action statistics: first = 358.961 us, mean = 232.448 us (CI deltas: -351.777 ns, +378.771 ns), sd = 129.736 us (CI deltas: -16.359 us, +20.960 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 8192 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 8192 --block statistics: mean = 1.904 s (CI deltas: -2.882 ms, +3.103 ms), sd = 11.742 ms (CI deltas: -1.481 ms, +1.897 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is 3.766359418012371 sigma above its mean, r[2] is 2.399620807226476 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 47.75920994050609% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 53.291205778405484% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 8192.0, muB = 1.9042112625333332, sigmaB = 0.011742339632967417, muA = 2.3244766388346352E-4, sigmaA = 1.29735749710419E-4, tMin = 0.0, muGMin = 1.1622383194173176E-4, sigmaG = 2.905595798543294E-5, cMax1 = 6263, cMax2 = 3703, cMax = 3703, cOutMin = 3703, varOutMin = 7.34792681537126E-5, muG(cOutMin) = 1.1624671841987245E-4, U(cOutMin) = 3.733134603150218E-4 NIO Buffer results : 131.442 us action statistics: first = 261.94 us, mean = 131.442 us (CI deltas: -1.517 us, +2.293 us), sd = 663.659 us (CI deltas: -182.211 us, +171.143 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 8192 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 8192 --block statistics: mean = 1.077 s (CI deltas: -12.429 ms, +18.788 ms), sd = 60.068 ms (CI deltas: -16.492 ms, +15.490 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.051 s, interquantileRange = 18.082 ms --8 are EXTREME (on the high side): #52 = 1.214 s, #53 = 1.215 s, #54 = 1.225 s, #55 = 1.226 s, #56 = 1.228 s, #57 = 1.237 s, #58 = 1.239 s, #59 = 1.242 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 7.020627217315627% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 98.97552993613668% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 8192.0, muB = 1.0767710386500002, sigmaB = 0.06006755799937111, muA = 1.314417771789551E-4, sigmaA = 6.63659024854243E-4, tMin = 0.0, muGMin = 6.572088858947755E-5, sigmaG = 1.643022214736939E-5, cMax1 = 309, cMax2 = 79, cMax = 79, cOutMin = 79, varOutMin = 0.003571147501573549, muG(cOutMin) = 6.597273850477304E-5, U(cOutMin) = 0.006854863432414893 NIO Transfer results : 91.548 us action statistics: first = 234.659 us, mean = 91.548 us (CI deltas: -1.611 us, +2.763 us), sd = 1.048 ms (CI deltas: -352.407 us, +455.355 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16384 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16384 --block statistics: mean = 1.500 s (CI deltas: -26.398 ms, +45.270 ms), sd = 134.133 ms (CI deltas: -45.108 ms, +58.285 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.451 s, interquantileRange = 13.001 ms --7 are EXTREME (on the high side): #53 = 1.752 s, #54 = 1.760 s, #55 = 1.769 s, #56 = 1.809 s, #57 = 1.824 s, #58 = 2.004 s, #59 = 2.050 s --1 are mild (on the high side): #52 = 1.485 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 3.710688954017908% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.79891077136634% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16384.0, muB = 1.499925190333333, sigmaB = 0.13413263453926863, muA = 9.154816835530597E-5, sigmaA = 0.0010479112073380362, tMin = 0.0, muGMin = 4.5774084177652986E-5, sigmaG = 1.1443521044413247E-5, cMax1 = 124, cMax2 = 31, cMax = 31, cOutMin = 31, varOutMin = 0.01795538455188521, muG(cOutMin) = 4.592549126686541E-5, U(cOutMin) = 0.02415824618213816 Start benchmark with big-text Native Copy results : 33.210 ms action statistics: first = 146.955 ms, mean = 33.210 ms (CI deltas: -582.574 us, +682.536 us), sd = 14.091 ms (CI deltas: -2.054 ms, +2.260 ms) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32 --block statistics: mean = 1.063 s (CI deltas: -18.642 ms, +21.841 ms), sd = 79.711 ms (CI deltas: -11.620 ms, +12.785 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 5: r[1] is 7.456193472792529 sigma above its mean, r[2] is 4.198609918579263 sigma above its mean, r[3] is 3.137213723354128 sigma above its mean, r[4] is 2.5601513181785456 sigma above its mean, r[5] is 2.1623577335268065 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 5.25588791207467% of the measured sd ---------- --action sd values likely INFLATED by outliers --they cause at least 42.0887615445945% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32.0, muB = 1.0627277899333334, sigmaB = 0.07971116025095822, muA = 0.03321024343541667, sigmaA = 0.014091075487425032, tMin = 0.0, muGMin = 0.016605121717708334, sigmaG = 0.0041512804294270836, cMax1 = 27, cMax2 = 18, cMax = 18, cOutMin = 18, varOutMin = 0.0026742648011194156, muG(cOutMin) = 0.017538748706400584, U(cOutMin) = 0.04539918378020696 Naive Readers results : 438.845 ms action statistics: first = 531.396 ms, mean = 438.845 ms (CI deltas: -516.254 us, +689.778 us), sd = 4.682 ms (CI deltas: -933.710 us, +1.842 ms) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 4 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 4 --block statistics: mean = 1.755 s (CI deltas: -2.065 ms, +2.759 ms), sd = 9.364 ms (CI deltas: -1.867 ms, +3.683 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.753 s, interquantileRange = 9.647 ms --1 are EXTREME (on the high side): #59 = 1.792 s --2 are mild (on the high side): #57 = 1.775 s, #58 = 1.779 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 57.50200811811445% of the measured sd Buffered Streams results : 65.382 ms action statistics: first = 65.839 ms, mean = 65.382 ms (CI deltas: -89.804 us, +107.754 us), sd = 1.548 ms (CI deltas: -225.620 us, +237.395 us) WARNING: SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.046 s (CI deltas: -1.437 ms, +1.724 ms), sd = 6.192 ms (CI deltas: -902.481 us, +949.582 us) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 67.12942392064181% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859375% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.0461132040333334, sigmaB = 0.006191995448062722, muA = 0.06538207525208334, sigmaA = 0.0015479988620156805, tMin = 0.0, muGMin = 0.03269103762604167, sigmaG = 0.0015479988620156805, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 15, varOutMin = 2.2465316970017268E-6, muG(cOutMin) = 0.05957707951952454, U(cOutMin) = 0.06576907496758726 Buffered Readers results : 82.065 ms action statistics: first = 81.933 ms, mean = 82.065 ms (CI deltas: -156.066 us, +174.442 us), sd = 2.614 ms (CI deltas: -301.353 us, +488.476 us) WARNING: SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.313 s (CI deltas: -2.497 ms, +2.791 ms), sd = 10.456 ms (CI deltas: -1.205 ms, +1.954 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 44.535725705270515% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859375% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.313043499, sigmaB = 0.01045647323550246, muA = 0.0820652186875, sigmaA = 0.002614118308875615, tMin = 0.0, muGMin = 0.04103260934375, sigmaG = 0.002614118308875615, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 15, varOutMin = 6.406513624498787E-6, muG(cOutMin) = 0.07226227502921644, U(cOutMin) = 0.0827187482647189 Custom Buffer Streams results : 9.626 ms action statistics: first = 10.075 ms, mean = 9.626 ms (CI deltas: -5.930 us, +7.729 us), sd = 300.779 us (CI deltas: -63.906 us, +92.925 us) WARNING: execution times have mild outliers, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 128 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 128 --block statistics: mean = 1.232 s (CI deltas: -758.999 us, +989.368 us), sd = 3.403 ms (CI deltas: -723.013 us, +1.051 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.232 s, interquantileRange = 4.087 ms --3 are mild (on the high side): #57 = 1.241 s, #58 = 1.242 s, #59 = 1.244 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 100.0% of the measured sd Custom Buffer Readers results : 17.602 ms action statistics: first = 18.354 ms, mean = 17.602 ms (CI deltas: -16.351 us, +22.395 us), sd = 599.960 us (CI deltas: -135.107 us, +196.958 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 64 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 64 --block statistics: mean = 1.127 s (CI deltas: -1.046 ms, +1.433 ms), sd = 4.800 ms (CI deltas: -1.081 ms, +1.576 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.125 s, interquantileRange = 4.214 ms --2 are EXTREME (on the high side): #58 = 1.143 s, #59 = 1.143 s --2 are mild (on the high side): #56 = 1.136 s, #57 = 1.137 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 89.86921382383406% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 1.5380859374999933% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 64.0, muB = 1.1265172256, sigmaB = 0.0047996788868813955, muA = 0.01760183165, sigmaA = 5.999598608601744E-4, tMin = 0.0, muGMin = 0.008800915825, sigmaG = 5.999598608601744E-4, cMax1 = 63, cMax2 = 63, cMax = 63, cOutMin = 1, varOutMin = 3.543275872270558E-7, muG(cOutMin) = 0.017592383170306803, U(cOutMin) = 0.01819708587067135 Custom Buffer Buffered Streams results : 9.581 ms action statistics: first = 10.062 ms, mean = 9.581 ms (CI deltas: -5.659 us, +5.704 us), sd = 253.945 us (CI deltas: -36.192 us, +56.255 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 128 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 128 --block statistics: mean = 1.226 s (CI deltas: -724.386 us, +730.104 us), sd = 2.873 ms (CI deltas: -409.471 us, +636.450 us) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is 4.753653991632496 sigma above its mean, r[2] is 3.2056196482195354 sigma above its mean, r[3] is 3.002486857613842 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 100.0% of the measured sd Custom Buffer Buffered Readers results : 19.053 ms action statistics: first = 19.741 ms, mean = 19.053 ms (CI deltas: -14.835 us, +18.674 us), sd = 523.391 us (CI deltas: -97.323 us, +166.400 us) WARNING: execution times have mild outliers, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 64 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 64 --block statistics: mean = 1.219 s (CI deltas: -949.422 us, +1.195 ms), sd = 4.187 ms (CI deltas: -778.586 us, +1.331 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.218 s, interquantileRange = 5.403 ms --1 are mild (on the high side): #59 = 1.234 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is 4.087230249400516 sigma above its mean, r[2] is 2.8172837154417305 sigma above its mean, r[3] is 2.3418813365790854 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 100.0% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 1.5380859374999933% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 64.0, muB = 1.2193957438333338, sigmaB = 0.004187129033815596, muA = 0.01905305849739584, sigmaA = 5.233911292269495E-4, tMin = 0.0, muGMin = 0.00952652924869792, sigmaG = 5.233911292269495E-4, cMax1 = 63, cMax2 = 63, cMax = 63, cOutMin = 1, varOutMin = 2.696579886198123E-7, muG(cOutMin) = 0.019044815861881843, U(cOutMin) = 0.019572344534777755 NIO Buffer results : 11.118 ms action statistics: first = 11.705 ms, mean = 11.118 ms (CI deltas: -16.527 us, +18.361 us), sd = 779.892 us (CI deltas: -113.719 us, +160.462 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 128 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 128 --block statistics: mean = 1.423 s (CI deltas: -2.116 ms, +2.350 ms), sd = 8.823 ms (CI deltas: -1.287 ms, +1.815 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is 5.3821789530725885 sigma above its mean, r[2] is 2.160353087371279 sigma above its mean, r[5] is 2.279240864830156 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 54.945778363360944% of the measured sd NIO Transfer results : 7.022 ms action statistics: first = 7.467 ms, mean = 7.022 ms (CI deltas: -5.836 us, +8.371 us), sd = 434.741 us (CI deltas: -114.715 us, +140.498 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 256 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 256 --block statistics: mean = 1.798 s (CI deltas: -1.494 ms, +2.143 ms), sd = 6.956 ms (CI deltas: -1.835 ms, +2.248 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.796 s, interquantileRange = 6.258 ms --2 are EXTREME (on the high side): #58 = 1.820 s, #59 = 1.821 s --4 are mild (on the high side): #54 = 1.811 s, #55 = 1.811 s, #56 = 1.814 s, #57 = 1.817 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 78.33284427372833% of the measured sd Start benchmark with fat-text Native Copy results : 106.501 ms action statistics: first = 960.810 ms, mean = 106.501 ms (CI deltas: -88.540 us, +77.038 us), sd = 1.309 ms (CI deltas: -204.978 us, +272.479 us) WARNING: SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.704 s (CI deltas: -1.417 ms, +1.233 ms), sd = 5.237 ms (CI deltas: -819.912 us, +1.090 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 100.0% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859374999999998% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.7040181301333333, sigmaB = 0.005236857748656106, muA = 0.10650113313333333, sigmaA = 0.0013092144371640265, tMin = 0.0, muGMin = 0.053250566566666664, sigmaG = 0.0013092144371640265, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 1, varOutMin = 1.6069147898237982E-6, muG(cOutMin) = 0.10641662370480415, U(cOutMin) = 0.10776877456127097 Buffered Streams results : 662.442 ms action statistics: first = 686.819 ms, mean = 662.442 ms (CI deltas: -1.879 ms, +4.607 ms), sd = 15.447 ms (CI deltas: -7.709 ms, +15.251 ms) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 2 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 2 --block statistics: mean = 1.325 s (CI deltas: -3.758 ms, +9.213 ms), sd = 21.845 ms (CI deltas: -10.903 ms, +21.568 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.319 s, interquantileRange = 6.049 ms --5 are EXTREME (on the high side): #55 = 1.357 s, #56 = 1.362 s, #57 = 1.368 s, #58 = 1.375 s, #59 = 1.463 s --2 are mild (on the high side): #53 = 1.334 s, #54 = 1.338 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 21.41367604867854% of the measured sd Buffered Readers results : 824.793 ms action statistics: first = 835.017 ms, mean = 824.793 ms (CI deltas: -759.662 us, +970.879 us), sd = 4.782 ms (CI deltas: -926.088 us, +954.940 us) WARNING: execution times have mild outliers, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 2 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 2 --block statistics: mean = 1.650 s (CI deltas: -1.519 ms, +1.942 ms), sd = 6.762 ms (CI deltas: -1.310 ms, +1.350 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.647 s, interquantileRange = 6.119 ms --7 are mild (on the high side): #53 = 1.661 s, #54 = 1.662 s, #55 = 1.664 s, #56 = 1.664 s, #57 = 1.666 s, #58 = 1.666 s, #59 = 1.667 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 77.18944735452091% of the measured sd Custom Buffer Streams results : 100.402 ms action statistics: first = 102.405 ms, mean = 100.402 ms (CI deltas: -82.802 us, +136.521 us), sd = 1.648 ms (CI deltas: -547.474 us, +818.768 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.606 s (CI deltas: -1.325 ms, +2.184 ms), sd = 6.593 ms (CI deltas: -2.190 ms, +3.275 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.605 s, interquantileRange = 5.061 ms --2 are EXTREME (on the high side): #58 = 1.633 s, #59 = 1.635 s --2 are mild (on the high side): #56 = 1.618 s, #57 = 1.621 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 78.12201489128581% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859374999999996% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.6064279653833333, sigmaB = 0.006593429984960762, muA = 0.10040174783645833, sigmaA = 0.0016483574962401905, tMin = 0.0, muGMin = 0.050200873918229166, sigmaG = 0.0016483574962401905, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 1, varOutMin = 2.5472647831980257E-6, muG(cOutMin) = 0.10029534681759943, U(cOutMin) = 0.10199776311934193 Custom Buffer Readers results : 181.292 ms action statistics: first = 185.144 ms, mean = 181.292 ms (CI deltas: -290.523 us, +294.976 us), sd = 3.275 ms (CI deltas: -332.642 us, +413.902 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 8 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 8 --block statistics: mean = 1.450 s (CI deltas: -2.324 ms, +2.360 ms), sd = 9.264 ms (CI deltas: -940.854 us, +1.171 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is 5.284446133706902 sigma above its mean, r[2] is 2.8283499045043006 sigma above its mean, r[3] is 2.3509688643382645 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 52.831391309567636% of the measured sd Custom Buffer Buffered Streams results : 99.959 ms action statistics: first = 101.399 ms, mean = 99.959 ms (CI deltas: -130.529 us, +179.005 us), sd = 2.400 ms (CI deltas: -546.051 us, +739.647 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.599 s (CI deltas: -2.088 ms, +2.864 ms), sd = 9.601 ms (CI deltas: -2.184 ms, +2.959 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.596 s, interquantileRange = 8.876 ms --1 are EXTREME (on the high side): #59 = 1.632 s --5 are mild (on the high side): #54 = 1.616 s, #55 = 1.618 s, #56 = 1.619 s, #57 = 1.623 s, #58 = 1.628 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 53.53172268296043% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859375% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.5993389330666665, sigmaB = 0.009600928646395699, muA = 0.09995868331666666, sigmaA = 0.0024002321615989248, tMin = 0.0, muGMin = 0.04997934165833333, sigmaG = 0.0024002321615989248, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 15, varOutMin = 5.401044777725481E-6, muG(cOutMin) = 0.09095781271067069, U(cOutMin) = 0.10055874135706638 Custom Buffer Buffered Readers results : 195.502 ms action statistics: first = 196.749 ms, mean = 195.502 ms (CI deltas: -256.554 us, +342.972 us), sd = 3.290 ms (CI deltas: -717.349 us, +1.158 ms) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 8 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 8 --block statistics: mean = 1.564 s (CI deltas: -2.052 ms, +2.744 ms), sd = 9.305 ms (CI deltas: -2.029 ms, +3.276 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.561 s, interquantileRange = 9.890 ms --1 are EXTREME (on the high side): #59 = 1.599 s --3 are mild (on the high side): #56 = 1.583 s, #57 = 1.586 s, #58 = 1.589 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 54.62059969142062% of the measured sd NIO Buffer results : 114.323 ms action statistics: first = 115.285 ms, mean = 114.323 ms (CI deltas: -146.682 us, +179.229 us), sd = 2.562 ms (CI deltas: -464.107 us, +638.974 us) WARNING: execution times have mild outliers, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.829 s (CI deltas: -2.347 ms, +2.868 ms), sd = 10.248 ms (CI deltas: -1.856 ms, +2.556 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.827 s, interquantileRange = 12.465 ms --3 are mild (on the high side): #57 = 1.856 s, #58 = 1.857 s, #59 = 1.859 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 53.63284036081512% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859374999999994% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.8291638015333336, sigmaB = 0.010248249932668749, muA = 0.11432273759583335, sigmaA = 0.002562062483167187, tMin = 0.0, muGMin = 0.057161368797916674, sigmaG = 0.002562062483167187, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 1, varOutMin = 6.153903907174506E-6, muG(cOutMin) = 0.1141573571736792, U(cOutMin) = 0.11680344392814568 NIO Transfer results : 75.064 ms action statistics: first = 75.362 ms, mean = 75.064 ms (CI deltas: -22.552 us, +27.359 us), sd = 392.113 us (CI deltas: -72.448 us, +95.255 us) WARNING: execution times have mild outliers, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.201 s (CI deltas: -360.832 us, +437.742 us), sd = 1.568 ms (CI deltas: -289.793 us, +381.021 us) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.201 s, interquantileRange = 1.775 ms --4 are mild (on the high side): #56 = 1.204 s, #57 = 1.205 s, #58 = 1.205 s, #59 = 1.206 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is 3.427817528232673 sigma above its mean, r[2] is 2.5988838571426136 sigma above its mean, r[3] is 2.1855981540682805 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 100.0% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859375% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.2010254033000003, sigmaB = 0.0015684518586091976, muA = 0.07506408770625002, sigmaA = 3.921129646522994E-4, tMin = 0.0, muGMin = 0.03753204385312501, sigmaG = 3.921129646522994E-4, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 15, varOutMin = 1.4414304098288944E-7, muG(cOutMin) = 0.0735936640888039, U(cOutMin) = 0.0751621159474131 Start benchmark with little-binary Native Copy results : 39.159 ms action statistics: first = 70.874 ms, mean = 39.159 ms (CI deltas: -23.577 us, +23.047 us), sd = 523.501 us (CI deltas: -64.651 us, +86.562 us) WARNING: SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32 --block statistics: mean = 1.253 s (CI deltas: -754.468 us, +737.505 us), sd = 2.961 ms (CI deltas: -365.720 us, +489.669 us) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 100.0% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 3.0273437500000004% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32.0, muB = 1.2531021965166669, sigmaB = 0.0029613700026179563, muA = 0.03915944364114584, sigmaA = 5.235012026133951E-4, tMin = 0.0, muGMin = 0.01957972182057292, sigmaG = 5.235012026133951E-4, cMax1 = 31, cMax2 = 31, cMax = 31, cOutMin = 31, varOutMin = 2.654893369771189E-7, muG(cOutMin) = 0.036290616451109696, U(cOutMin) = 0.03925198645372765 Naive Streams results : 10.781 ms action statistics: first = 11.001 ms, mean = 10.781 ms (CI deltas: -11.919 us, +16.704 us), sd = 624.244 us (CI deltas: -175.833 us, +317.346 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 128 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 128 --block statistics: mean = 1.380 s (CI deltas: -1.526 ms, +2.138 ms), sd = 7.063 ms (CI deltas: -1.989 ms, +3.590 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.380 s, interquantileRange = 7.780 ms --1 are EXTREME (on the high side): #59 = 1.411 s --1 are mild (on the high side): #58 = 1.403 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 67.59721087159456% of the measured sd Buffered Streams results : 106.257 us action statistics: first = 210.560 us, mean = 106.257 us (CI deltas: -342.916 ns, +324.730 ns), sd = 168.836 us (CI deltas: -16.590 us, +17.461 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16384 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16384 --block statistics: mean = 1.741 s (CI deltas: -5.618 ms, +5.320 ms), sd = 21.611 ms (CI deltas: -2.124 ms, +2.235 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 6: r[1] is -4.405916242454898 sigma below its mean, r[2] is 2.178665885576643 sigma above its mean, r[5] is 2.6595304649981575 sigma above its mean, r[6] is -2.7812321786805208 sigma below its mean, r[7] is 2.247820817543802 sigma above its mean, r[12] is 2.0987808290572385 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 24.81247279152789% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 90.43656622191692% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16384.0, muB = 1.7409226173000003, sigmaB = 0.02161095169642323, muA = 1.0625748396606447E-4, sigmaA = 1.688355601283065E-4, tMin = 0.0, muGMin = 5.3128741983032236E-5, sigmaG = 1.3282185495758059E-5, cMax1 = 4663, cMax2 = 1483, cMax = 1483, cOutMin = 1483, varOutMin = 4.223688192440118E-4, muG(cOutMin) = 5.314445301742191E-5, U(cOutMin) = 6.399306290542121E-4 Custom Buffer Streams results : 47.247 us action statistics: first = 148.233 us, mean = 47.247 us (CI deltas: -420.579 ns, +461.433 ns), sd = 315.324 us (CI deltas: -26.219 us, +16.833 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32768 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32768 --block statistics: mean = 1.548 s (CI deltas: -13.782 ms, +15.120 ms), sd = 57.080 ms (CI deltas: -4.746 ms, +3.047 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 4: r[1] is -4.6204044301815115 sigma below its mean, r[4] is -3.078488179592689 sigma below its mean, r[5] is 3.3999736364240056 sigma above its mean, r[8] is 1.9968422204708922 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 8.858961289812292% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.4098894798738% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32768.0, muB = 1.548182904433333, sigmaB = 0.057079812327745304, muA = 4.724679273783365E-5, sigmaA = 3.153243934828453E-4, tMin = 0.0, muGMin = 2.3623396368916825E-5, sigmaG = 5.905849092229206E-6, cMax1 = 719, cMax2 = 182, cMax = 182, cOutMin = 182, varOutMin = 0.0032388785551542075, muG(cOutMin) = 2.3685344968254552E-5, U(cOutMin) = 0.004265781611526321 Custom Buffer Buffered Streams results : 53.768 us action statistics: first = 153.797 us, mean = 53.768 us (CI deltas: -135.663 ns, +146.228 ns), sd = 101.188 us (CI deltas: -11.147 us, +12.748 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32768 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32768 --block statistics: mean = 1.762 s (CI deltas: -4.445 ms, +4.792 ms), sd = 18.317 ms (CI deltas: -2.018 ms, +2.308 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 5: r[1] is -3.855601296756049 sigma below its mean, r[4] is -2.0366468651601632 sigma below its mean, r[5] is 2.8073057431420283 sigma above its mean, r[6] is -2.400033501626497 sigma below its mean, r[7] is 1.9778139880087637 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 29.450176362305246% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 92.99905985769595% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32768.0, muB = 1.7618733824166668, sigmaB = 0.018316970843033256, muA = 5.376810859425863E-5, sigmaA = 1.0118792417113349E-4, tMin = 0.0, muGMin = 2.6884054297129315E-5, sigmaG = 6.721013574282329E-6, cMax1 = 7234, cMax2 = 2168, cMax = 2168, cOutMin = 2168, varOutMin = 3.1202246711921084E-4, muG(cOutMin) = 2.6889848903043173E-5, U(cOutMin) = 4.3313837914370185E-4 NIO Buffer results : 53.528 us action statistics: first = 160.399 us, mean = 53.528 us (CI deltas: -716.702 ns, +707.273 ns), sd = 510.512 us (CI deltas: -15.020 us, +18.421 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32768 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32768 --block statistics: mean = 1.754 s (CI deltas: -23.485 ms, +23.176 ms), sd = 92.413 ms (CI deltas: -2.719 ms, +3.335 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is -6.361273796199412 sigma below its mean, r[2] is 3.611535528252364 sigma above its mean, r[3] is -2.5015051235836134 sigma below its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 5.824226994876313% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.71130890061775% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32768.0, muB = 1.754002788383334, sigmaB = 0.09241257114399323, muA = 5.352791712595624E-5, sigmaA = 5.105121540843896E-4, tMin = 0.0, muGMin = 2.676395856297812E-5, sigmaG = 6.69098964074453E-6, cMax1 = 356, cMax2 = 89, cMax = 89, cOutMin = 89, varOutMin = 0.008515428845060966, muG(cOutMin) = 2.688820624613756E-5, U(cOutMin) = 0.009835090971525896 NIO Transfer results : 52.292 us action statistics: first = 159.495 us, mean = 52.292 us (CI deltas: -1.040 us, +1.341 us), sd = 845.120 us (CI deltas: -148.329 us, +184.397 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32768 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32768 --block statistics: mean = 1.713 s (CI deltas: -34.083 ms, +43.946 ms), sd = 152.983 ms (CI deltas: -26.850 ms, +33.380 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.630 s, interquantileRange = 96.216 ms --2 are EXTREME (on the high side): #58 = 2.148 s, #59 = 2.150 s --13 are mild (on the high side): #45 = 1.925 s, #46 = 1.931 s, #47 = 1.934 s, #48 = 1.938 s, #49 = 1.939 s, #50 = 1.942 s, #51 = 1.943 s, #52 = 1.946 s, #53 = 1.946 s, #54 = 1.948 s, #55 = 1.955 s, #56 = 1.955 s, #57 = 1.967 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is -2.338868307522368 sigma below its mean, r[2] is -2.234685748323801 sigma below its mean, r[7] is 2.8033482148973516 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 3.4773841547186928% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.89942476698532% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32768.0, muB = 1.7134968642166668, sigmaB = 0.15298300694235198, muA = 5.2291774420674646E-5, sigmaA = 8.451197001191074E-4, tMin = 0.0, muGMin = 2.6145887210337323E-5, sigmaG = 6.536471802584331E-6, cMax1 = 124, cMax2 = 31, cMax = 31, cOutMin = 31, varOutMin = 0.023380261986323925, muG(cOutMin) = 2.628618777834026E-5, U(cOutMin) = 0.027515030158617476 Start benchmark with medium-binary Native Copy results : 29.354 ms action statistics: first = 61.828 ms, mean = 29.354 ms (CI deltas: -449.307 us, +523.458 us), sd = 10.890 ms (CI deltas: -1.687 ms, +1.824 ms) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32 --block statistics: mean = 939.316 ms (CI deltas: -14.378 ms, +16.751 ms), sd = 61.601 ms (CI deltas: -9.544 ms, +10.315 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 5: r[1] is 7.416307246935743 sigma above its mean, r[2] is 4.182379257839522 sigma above its mean, r[3] is 3.11849288920482 sigma above its mean, r[4] is 2.5454149261915124 sigma above its mean, r[5] is 2.14464826309498 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 6.394024815668237% of the measured sd ---------- --action sd values likely INFLATED by outliers --they cause at least 35.90344286740956% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32.0, muB = 0.9393158696666664, sigmaB = 0.06160071019571341, muA = 0.029353620927083324, sigmaA = 0.010889569976324062, tMin = 0.0, muGMin = 0.014676810463541662, sigmaG = 0.0036692026158854155, cMax1 = 28, cMax2 = 20, cMax = 20, cOutMin = 20, varOutMin = 0.0013624090959672094, muG(cOutMin) = 0.015597768152682118, U(cOutMin) = 0.037607132591724046 Naive Streams results : 108.139 ms action statistics: first = 113.597 ms, mean = 108.139 ms (CI deltas: -140.320 us, +160.829 us), sd = 2.367 ms (CI deltas: -343.929 us, +495.913 us) WARNING: execution times have mild outliers, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.730 s (CI deltas: -2.245 ms, +2.573 ms), sd = 9.469 ms (CI deltas: -1.376 ms, +1.984 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.730 s, interquantileRange = 13.251 ms --1 are mild (on the high side): #59 = 1.757 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is 5.2271341755069605 sigma above its mean, r[2] is 2.2608444612628853 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 56.45414090293591% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859374999999993% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.730231820616667, sigmaB = 0.009469140299468024, muA = 0.10813948878854168, sigmaA = 0.002367285074867006, tMin = 0.0, muGMin = 0.05406974439427084, sigmaG = 0.002367285074867006, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 1, varOutMin = 5.2537862115825745E-6, muG(cOutMin) = 0.10798668119403027, U(cOutMin) = 0.11043160270621301 Buffered Streams results : 670.848 us action statistics: first = 783.129 us, mean = 670.848 us (CI deltas: -1.167 us, +1.470 us), sd = 233.215 us (CI deltas: -43.936 us, +66.309 us) WARNING: execution times have mild outliers, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 2048 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 2048 --block statistics: mean = 1.374 s (CI deltas: -2.391 ms, +3.010 ms), sd = 10.554 ms (CI deltas: -1.988 ms, +3.001 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.370 s, interquantileRange = 13.938 ms --2 are mild (on the high side): #58 = 1.405 s, #59 = 1.408 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 45.13461464007903% of the measured sd ---------- --action sd values likely INFLATED by outliers --they cause at least 30.391124976256368% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 2048.0, muB = 1.3738959593166664, sigmaB = 0.010554104527324353, muA = 6.70847636385091E-4, sigmaA = 2.332149650194591E-4, tMin = 0.0, muGMin = 3.354238181925455E-4, sigmaG = 8.385595454813638E-5, cMax1 = 1829, cMax2 = 1399, cMax = 1399, cOutMin = 1399, varOutMin = 3.385240739054278E-5, muG(cOutMin) = 3.355280361067821E-4, U(cOutMin) = 8.264033337264938E-4 Custom Buffer Streams results : 113.764 us action statistics: first = 246.859 us, mean = 113.764 us (CI deltas: -572.467 ns, +721.225 ns), sd = 324.447 us (CI deltas: -59.992 us, +54.605 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16384 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16384 --block statistics: mean = 1.864 s (CI deltas: -9.379 ms, +11.817 ms), sd = 41.529 ms (CI deltas: -7.679 ms, +6.989 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.843 s, interquantileRange = 31.349 ms --1 are EXTREME (on the high side): #59 = 1.963 s --12 are mild (on the high side): #47 = 1.920 s, #48 = 1.922 s, #49 = 1.930 s, #50 = 1.930 s, #51 = 1.931 s, #52 = 1.932 s, #53 = 1.935 s, #54 = 1.939 s, #55 = 1.950 s, #56 = 1.951 s, #57 = 1.952 s, #58 = 1.961 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[2] is -1.9824811870119907 sigma below its mean, r[4] is 2.433260182739096 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 13.360178969418643% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 96.83457113945686% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16384.0, muB = 1.8639075289666667, sigmaB = 0.04152925231219978, muA = 1.137638872660319E-4, sigmaA = 3.244472836890608E-4, tMin = 0.0, muGMin = 5.688194363301595E-5, sigmaG = 1.4220485908253988E-5, cMax1 = 1796, cMax2 = 489, cMax = 489, cOutMin = 489, varOutMin = 0.0016700853171991244, muG(cOutMin) = 5.690959911900919E-5, U(cOutMin) = 0.001961818918139091 Custom Buffer Buffered Streams results : 121.967 us action statistics: first = 240.803 us, mean = 121.967 us (CI deltas: -308.323 ns, +276.309 ns), sd = 147.461 us (CI deltas: -19.798 us, +21.071 us) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16384 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16384 --block statistics: mean = 1.998 s (CI deltas: -5.052 ms, +4.527 ms), sd = 18.875 ms (CI deltas: -2.534 ms, +2.697 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 4: r[2] is -2.742097536063787 sigma below its mean, r[3] is 2.013267722708259 sigma above its mean, r[5] is -2.184529085305334 sigma below its mean, r[7] is 3.336649659102178 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 30.43673517234369% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 84.50694403383503% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16384.0, muB = 1.9983054244000003, sigmaB = 0.018875004978292732, muA = 1.2196688381347658E-4, sigmaA = 1.4746097639291197E-4, tMin = 0.0, muGMin = 6.098344190673829E-5, sigmaG = 1.5245860476684572E-5, cMax1 = 6680, cMax2 = 2411, cMax = 2411, cOutMin = 2411, varOutMin = 3.010693511449288E-4, muG(cOutMin) = 6.099323017338826E-5, U(cOutMin) = 4.7534094532859653E-4 NIO Buffer results : 131.176 us action statistics: first = 260.509 us, mean = 131.176 us (CI deltas: -1.442 us, +2.192 us), sd = 632.337 us (CI deltas: -175.204 us, +165.915 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 8192 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 8192 --block statistics: mean = 1.075 s (CI deltas: -11.813 ms, +17.959 ms), sd = 57.233 ms (CI deltas: -15.858 ms, +15.017 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.050 s, interquantileRange = 16.887 ms --8 are EXTREME (on the high side): #52 = 1.206 s, #53 = 1.211 s, #54 = 1.213 s, #55 = 1.216 s, #56 = 1.217 s, #57 = 1.222 s, #58 = 1.227 s, #59 = 1.241 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 7.36092559858002% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 98.87216861490047% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 8192.0, muB = 1.0745905087000003, sigmaB = 0.057232574051795715, muA = 1.31175599206543E-4, sigmaA = 6.323365815122811E-4, tMin = 0.0, muGMin = 6.55877996032715E-5, sigmaG = 1.6396949900817875E-5, cMax1 = 338, cMax2 = 87, cMax = 87, cOutMin = 87, varOutMin = 0.003238624653921552, muG(cOutMin) = 6.56838008606975E-5, U(cOutMin) = 0.0062324517554488154 NIO Transfer results : 91.910 us action statistics: first = 246.714 us, mean = 91.910 us (CI deltas: -1.579 us, +2.609 us), sd = 1.002 ms (CI deltas: -314.450 us, +389.207 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16384 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16384 --block statistics: mean = 1.506 s (CI deltas: -25.865 ms, +42.751 ms), sd = 128.314 ms (CI deltas: -40.250 ms, +49.818 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.460 s, interquantileRange = 14.500 ms --7 are EXTREME (on the high side): #53 = 1.766 s, #54 = 1.779 s, #55 = 1.781 s, #56 = 1.790 s, #57 = 1.903 s, #58 = 1.905 s, #59 = 2.016 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 3.886608418512459% of the measured sd ---------- --action sd values ALMOST CERTAINLY GROSSLY INFLATED by outliers --they cause at least 99.77940036791519% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16384.0, muB = 1.5058517199666668, sigmaB = 0.12831414203150598, muA = 9.190989501749675E-5, sigmaA = 0.0010024542346211405, tMin = 0.0, muGMin = 4.5954947508748376E-5, sigmaG = 1.1488736877187094E-5, cMax1 = 136, cMax2 = 34, cMax = 34, cOutMin = 34, varOutMin = 0.016428198376843065, muG(cOutMin) = 4.6199321449890295E-5, U(cOutMin) = 0.022073318066498836 Start benchmark with big-binary Native Copy results : 33.403 ms action statistics: first = 213.311 ms, mean = 33.403 ms (CI deltas: -539.926 us, +624.927 us), sd = 12.998 ms (CI deltas: -1.889 ms, +2.092 ms) WARNING: execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 32 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 32 --block statistics: mean = 1.069 s (CI deltas: -17.278 ms, +19.998 ms), sd = 73.530 ms (CI deltas: -10.686 ms, +11.832 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 5: r[1] is 7.448547699215833 sigma above its mean, r[2] is 4.198798513278502 sigma above its mean, r[3] is 3.1359801029265584 sigma above its mean, r[4] is 2.5567994239969423 sigma above its mean, r[5] is 2.1586903043420484 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 5.714205826597479% of the measured sd ---------- --action sd values likely INFLATED by outliers --they cause at least 36.048984228899776% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 32.0, muB = 1.0688978507333333, sigmaB = 0.07353031542679475, muA = 0.033403057835416666, sigmaA = 0.012998446165268093, tMin = 0.0, muGMin = 0.016701528917708333, sigmaG = 0.004175382229427083, cMax1 = 27, cMax2 = 20, cMax = 20, cOutMin = 20, varOutMin = 0.0019490630571083044, muG(cOutMin) = 0.016949997257634995, U(cOutMin) = 0.04327489418208567 Buffered Streams results : 65.835 ms action statistics: first = 65.689 ms, mean = 65.835 ms (CI deltas: -128.441 us, +185.782 us), sd = 2.413 ms (CI deltas: -586.834 us, +1.028 ms) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.053 s (CI deltas: -2.055 ms, +2.973 ms), sd = 9.653 ms (CI deltas: -2.347 ms, +4.112 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.049 s, interquantileRange = 10.501 ms --1 are EXTREME (on the high side): #59 = 1.093 s --2 are mild (on the high side): #57 = 1.077 s, #58 = 1.082 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 43.209976368245336% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859375% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.0533531094166668, sigmaB = 0.009652885811993882, muA = 0.06583456933854168, sigmaA = 0.0024132214529984704, tMin = 0.0, muGMin = 0.03291728466927084, sigmaG = 0.0024132214529984704, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 15, varOutMin = 5.4596604198862956E-6, muG(cOutMin) = 0.05678498888979741, U(cOutMin) = 0.0664378747017913 Custom Buffer Streams results : 9.827 ms action statistics: first = 10.689 ms, mean = 9.827 ms (CI deltas: -8.929 us, +10.423 us), sd = 428.720 us (CI deltas: -71.126 us, +134.727 us) WARNING: execution times have mild outliers, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 128 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 128 --block statistics: mean = 1.258 s (CI deltas: -1.143 ms, +1.334 ms), sd = 4.850 ms (CI deltas: -804.704 us, +1.524 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.258 s, interquantileRange = 6.564 ms --1 are mild (on the high side): #59 = 1.275 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 93.971002144911% of the measured sd Custom Buffer Buffered Streams results : 9.798 ms action statistics: first = 10.336 ms, mean = 9.798 ms (CI deltas: -10.031 us, +12.165 us), sd = 490.230 us (CI deltas: -87.337 us, +128.198 us) WARNING: execution times have mild outliers, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 128 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 128 --block statistics: mean = 1.254 s (CI deltas: -1.284 ms, +1.557 ms), sd = 5.546 ms (CI deltas: -988.111 us, +1.450 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.253 s, interquantileRange = 6.760 ms --2 are mild (on the high side): #58 = 1.269 s, #59 = 1.271 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 82.05811126117052% of the measured sd NIO Buffer results : 11.156 ms action statistics: first = 11.757 ms, mean = 11.156 ms (CI deltas: -14.110 us, +17.025 us), sd = 690.016 us (CI deltas: -123.937 us, +219.472 us) WARNING: execution times have mild outliers, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 128 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 128 --block statistics: mean = 1.428 s (CI deltas: -1.806 ms, +2.179 ms), sd = 7.807 ms (CI deltas: -1.402 ms, +2.483 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.426 s, interquantileRange = 11.356 ms --1 are mild (on the high side): #59 = 1.455 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is 3.7700961587505972 sigma above its mean, r[2] is 2.3085027069122854 sigma above its mean, r[3] is 2.1575484888320444 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 62.20733054260778% of the measured sd NIO Transfer results : 7.038 ms action statistics: first = 7.599 ms, mean = 7.038 ms (CI deltas: -5.579 us, +9.713 us), sd = 452.785 us (CI deltas: -160.121 us, +252.427 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 256 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 256 --block statistics: mean = 1.802 s (CI deltas: -1.428 ms, +2.487 ms), sd = 7.245 ms (CI deltas: -2.562 ms, +4.039 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.800 s, interquantileRange = 3.761 ms --5 are EXTREME (on the high side): #55 = 1.816 s, #56 = 1.817 s, #57 = 1.823 s, #58 = 1.825 s, #59 = 1.837 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is 5.636797569942608 sigma above its mean, r[2] is 3.1655969188774926 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 75.3007747517682% of the measured sd Start benchmark with fat-binary Native Copy results : 95.855 ms action statistics: first = 1.254 s, mean = 95.855 ms (CI deltas: -73.097 us, +62.967 us), sd = 1.071 ms (CI deltas: -184.165 us, +348.713 us) WARNING: execution times have mild outliers, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.534 s (CI deltas: -1.170 ms, +1.007 ms), sd = 4.285 ms (CI deltas: -736.659 us, +1.395 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.534 s, interquantileRange = 4.713 ms --2 are mild (on the low side): #0 = 1.519 s, #1 = 1.524 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 100.0% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859375% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.533680255116667, sigmaB = 0.0042849560873029525, muA = 0.09585501594479169, sigmaA = 0.0010712390218257381, tMin = 0.0, muGMin = 0.047927507972395846, sigmaG = 0.0010712390218257381, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 15, varOutMin = 1.075830976764529E-6, muG(cOutMin) = 0.09183786961294517, U(cOutMin) = 0.09612282570024813 Buffered Streams results : 668.669 ms action statistics: first = 676.932 ms, mean = 668.669 ms (CI deltas: -1.573 ms, +4.991 ms), sd = 14.129 ms (CI deltas: -8.900 ms, +16.862 ms) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 2 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 2 --block statistics: mean = 1.337 s (CI deltas: -3.146 ms, +9.983 ms), sd = 19.981 ms (CI deltas: -12.586 ms, +23.847 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.333 s, interquantileRange = 9.572 ms --1 are EXTREME (on the high side): #59 = 1.479 s --1 are mild (on the high side): #58 = 1.359 s ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 23.521172170985942% of the measured sd Custom Buffer Streams results : 105.466 ms action statistics: first = 106.795 ms, mean = 105.466 ms (CI deltas: -123.630 us, +169.146 us), sd = 2.272 ms (CI deltas: -480.840 us, +676.761 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.687 s (CI deltas: -1.978 ms, +2.706 ms), sd = 9.089 ms (CI deltas: -1.923 ms, +2.707 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.684 s, interquantileRange = 6.956 ms --2 are EXTREME (on the high side): #58 = 1.713 s, #59 = 1.719 s --6 are mild (on the high side): #52 = 1.701 s, #53 = 1.701 s, #54 = 1.704 s, #55 = 1.705 s, #56 = 1.706 s, #57 = 1.707 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is 3.574619200239526 sigma above its mean, r[2] is 3.2012360160122046 sigma above its mean, r[8] is 1.9918656398834382 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 58.08522022048355% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859374999999996% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.6874524766333334, sigmaB = 0.009088753855429094, muA = 0.10546577978958334, sigmaA = 0.0022721884638572734, tMin = 0.0, muGMin = 0.05273288989479167, sigmaG = 0.0022721884638572734, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 1, varOutMin = 4.840162889330692E-6, muG(cOutMin) = 0.10531911065491728, U(cOutMin) = 0.1076658168095743 Custom Buffer Buffered Streams results : 105.023 ms action statistics: first = 118.522 ms, mean = 105.023 ms (CI deltas: -84.613 us, +128.718 us), sd = 1.628 ms (CI deltas: -461.210 us, +691.039 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.680 s (CI deltas: -1.354 ms, +2.059 ms), sd = 6.510 ms (CI deltas: -1.845 ms, +2.764 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.678 s, interquantileRange = 4.759 ms --3 are EXTREME (on the high side): #57 = 1.698 s, #58 = 1.702 s, #59 = 1.707 s --3 are mild (on the high side): #54 = 1.690 s, #55 = 1.691 s, #56 = 1.691 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is 4.1369771938623074 sigma above its mean, r[2] is 2.6919343815601 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 80.9204372741285% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859375% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.6803674689499999, sigmaB = 0.006510256779119285, muA = 0.10502296680937499, sigmaA = 0.0016275641947798211, tMin = 0.0, muGMin = 0.052511483404687496, sigmaG = 0.0016275641947798211, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 15, varOutMin = 2.483404882621207E-6, muG(cOutMin) = 0.09891960107895066, U(cOutMin) = 0.10542985785806995 NIO Buffer results : 118.303 ms action statistics: first = 120.916 ms, mean = 118.303 ms (CI deltas: -122.830 us, +158.531 us), sd = 2.194 ms (CI deltas: -422.155 us, +668.042 us) WARNING: execution times have mild outliers, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.893 s (CI deltas: -1.965 ms, +2.536 ms), sd = 8.775 ms (CI deltas: -1.689 ms, +2.672 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.889 s, interquantileRange = 9.328 ms --4 are mild (on the high side): #56 = 1.911 s, #57 = 1.912 s, #58 = 1.914 s, #59 = 1.924 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[2] is 2.7156578643075444 sigma above its mean, r[7] is 2.056695378055656 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 63.715098374833204% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859374999999994% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.8928435431666661, sigmaB = 0.008775446194712636, muA = 0.11830272144791663, sigmaA = 0.002193861548678159, tMin = 0.0, muGMin = 0.05915136072395832, sigmaG = 0.002193861548678159, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 1, varOutMin = 4.512214213845493E-6, muG(cOutMin) = 0.11816110829388471, U(cOutMin) = 0.12042691875839547 NIO Transfer results : 81.198 ms action statistics: first = 83.026 ms, mean = 81.198 ms (CI deltas: -55.601 us, +92.176 us), sd = 1.111 ms (CI deltas: -361.533 us, +462.829 us) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS, execution times may have serial correlation, SD VALUES MAY BE INACCURATE ---------- --the action statistics were calculated from block statistics --each block measured 16 task executions --the user says that task internally performs m = 1 actions --then the number of actions per block measurement is a = 16 --block statistics: mean = 1.299 s (CI deltas: -889.617 us, +1.475 ms), sd = 4.444 ms (CI deltas: -1.446 ms, +1.851 ms) --the forumla used to convert block statistics to action statistics (mean scales as 1/a, sd scales as 1/sqrt(a)) assumes that the action execution times are iid ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 1.298 s, interquantileRange = 2.173 ms --5 are EXTREME (on the high side): #55 = 1.307 s, #56 = 1.312 s, #57 = 1.313 s, #58 = 1.314 s, #59 = 1.317 s --1 are mild (on the high side): #54 = 1.306 s ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is 5.434005079784809 sigma above its mean, r[2] is 2.2686752880387764 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) ---------- --block sd values MAY NOT REFLECT TASK'S INTRINSIC VARIATION --guesstimate: environmental noise explains at least 100.0% of the measured sd ---------- --action sd values might be somewhat inflated by outliers --they cause at least 5.859374999999996% of the measured VARIANCE according to a equi-valued outlier model --model quantities: a = 16.0, muB = 1.2991731050333335, sigmaB = 0.0044440357830957235, muA = 0.08119831906458334, sigmaA = 0.0011110089457739309, tMin = 0.0, muGMin = 0.04059915953229167, sigmaG = 0.0011110089457739309, cMax1 = 15, cMax2 = 15, cMax = 15, cOutMin = 1, varOutMin = 1.157194572740344E-6, muG(cOutMin) = 0.08112660374550884, U(cOutMin) = 0.08227404885070082 Start benchmark with enormous-binary Native Copy results : 6.293 s action statistics: first = 9.371 s, mean = 6.293 s (CI deltas: -737.601 ms, +796.337 ms), sd = 3.018 s (CI deltas: -195.325 ms, +260.180 ms) WARNING: execution times may have serial correlation ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is -5.007435175764749 sigma below its mean, r[2] is 2.709415063162774 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) Custom Buffer Streams results : 9.971 s action statistics: first = 4.018 s, mean = 9.971 s (CI deltas: -1.299 s, +1.455 s), sd = 5.430 s (CI deltas: -579.159 ms, +793.407 ms) WARNING: execution times may have serial correlation ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 3: r[1] is -4.445998136299752 sigma below its mean, r[2] is 2.636157588837478 sigma above its mean, r[4] is 2.4502399001189477 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) Custom Buffer Buffered Streams results : 13.461 s action statistics: first = 14.498 s, mean = 13.461 s (CI deltas: -1.768 s, +1.812 s), sd = 7.093 s (CI deltas: -711.063 ms, +859.129 ms) WARNING: execution times may have serial correlation ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES HAVE SERIAL CORRELATION --1 of the 15 autocorrelation function coefficients (r[k]) that were computed are expected to fall outside their 95% CI --but found these 2: r[1] is -3.695655037333083 sigma below its mean, r[2] is 1.990073358429063 sigma above its mean --the 95% CI for the r[k] was calculated as mean +- 1.96 sigma (i.e. a Gaussian distribution was assumed) NIO Buffer results : 16.654 s action statistics: first = 19.098 s, mean = 16.654 s (CI deltas: -2.096 s, +2.306 s), sd = 8.682 s (CI deltas: -1.263 s, +1.569 s) WARNING: execution times have mild outliers ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 16.762 s, interquantileRange = 10.819 s --2 are mild (on the high side): #58 = 37.021 s, #59 = 37.339 s NIO Transfer results : 5.556 s action statistics: first = 12.793 s, mean = 5.556 s (CI deltas: -384.033 ms, +728.367 ms), sd = 2.008 s (CI deltas: -878.516 ms, +1.019 s) WARNING: EXECUTION TIMES HAVE EXTREME OUTLIERS ---------- --each confidence interval (CI) is reported as either +- deltas from the point estimate, or as a closed interval ([x, y]) --each confidence interval has confidence level = 0.95 ---------- --EXECUTION TIMES APPEAR TO HAVE OUTLIERS --this was determined using the boxplot algorithm with median = 5.038 s, interquantileRange = 126.530 ms --1 are EXTREME (on the low side): #0 = 4.148 s --1 are mild (on the low side): #1 = 4.768 s --4 are EXTREME (on the high side): #56 = 10.510 s, #57 = 13.534 s, #58 = 13.819 s, #59 = 13.826 s --3 are mild (on the high side): #53 = 5.311 s, #54 = 5.366 s, #55 = 5.406 s