calculate_k             time:   [20.454 ns 20.459 ns 20.467 ns]
                        change: [-0.0045% +0.0847% +0.2410%] (p = 0.25 > 0.05)
                        No change in performance detected.
Found 4 outliers among 100 measurements (4.00%)
  1 (1.00%) low mild
  1 (1.00%) high mild
  2 (2.00%) high severe

data_generation/synthesize_displacement/200
                        time:   [2.4539 µs 2.4548 µs 2.4558 µs]
                        thrpt:  [81.438 Melem/s 81.474 Melem/s 81.504 Melem/s]
                 change:
                        time:   [+0.3766% +0.4303% +0.4814%] (p = 0.00 < 0.05)
                        thrpt:  [-0.4790% -0.4285% -0.3752%]
                        Change within noise threshold.
Found 1 outliers among 100 measurements (1.00%)
  1 (1.00%) high severe
data_generation/synthesize_displacement/1000
                        time:   [11.818 µs 11.819 µs 11.820 µs]
                        thrpt:  [84.601 Melem/s 84.610 Melem/s 84.618 Melem/s]
                 change:
                        time:   [-0.1583% -0.1453% -0.1340%] (p = 0.00 < 0.05)
                        thrpt:  [+0.1342% +0.1456% +0.1586%]
                        Change within noise threshold.
Found 13 outliers among 100 measurements (13.00%)
  2 (2.00%) low mild
  3 (3.00%) high mild
  8 (8.00%) high severe
data_generation/synthesize_displacement/10000
                        time:   [115.03 µs 115.05 µs 115.09 µs]
                        thrpt:  [86.891 Melem/s 86.918 Melem/s 86.935 Melem/s]
                 change:
                        time:   [+0.0242% +0.0662% +0.0959%] (p = 0.00 < 0.05)
                        thrpt:  [-0.0958% -0.0661% -0.0242%]
                        Change within noise threshold.
Found 7 outliers among 100 measurements (7.00%)
  1 (1.00%) low mild
  3 (3.00%) high mild
  3 (3.00%) high severe
data_generation/synthesize_displacement/100000
                        time:   [1.1594 ms 1.1599 ms 1.1606 ms]
                        thrpt:  [86.160 Melem/s 86.216 Melem/s 86.250 Melem/s]
                 change:
                        time:   [+0.0245% +0.0605% +0.0994%] (p = 0.00 < 0.05)
                        thrpt:  [-0.0993% -0.0605% -0.0245%]
                        Change within noise threshold.
Found 16 outliers among 100 measurements (16.00%)
  1 (1.00%) low severe
  7 (7.00%) low mild
  5 (5.00%) high mild
  3 (3.00%) high severe

forward_pass/lffn_mlp/200
                        time:   [1.0039 ms 1.0047 ms 1.0061 ms]
                        thrpt:  [198.78 Kelem/s 199.06 Kelem/s 199.22 Kelem/s]
                 change:
                        time:   [-27.235% -27.185% -27.130%] (p = 0.00 < 0.05)
                        thrpt:  [+37.231% +37.334% +37.429%]
                        Performance has improved.
Found 3 outliers among 50 measurements (6.00%)
  2 (4.00%) high mild
  1 (2.00%) high severe
forward_pass/lffn_mlp/1000
                        time:   [7.9961 ms 7.9981 ms 8.0003 ms]
                        thrpt:  [125.00 Kelem/s 125.03 Kelem/s 125.06 Kelem/s]
                 change:
                        time:   [+56.096% +56.255% +56.395%] (p = 0.00 < 0.05)
                        thrpt:  [-36.059% -36.002% -35.937%]
                        Performance has regressed.
Found 5 outliers among 50 measurements (10.00%)
  3 (6.00%) low mild
  1 (2.00%) high mild
  1 (2.00%) high severe
forward_pass/lffn_mlp/10000
                        time:   [83.357 ms 83.377 ms 83.398 ms]
                        thrpt:  [119.91 Kelem/s 119.94 Kelem/s 119.97 Kelem/s]
                 change:
                        time:   [+74.586% +74.896% +75.179%] (p = 0.00 < 0.05)
                        thrpt:  [-42.916% -42.823% -42.722%]
                        Performance has regressed.
Found 4 outliers among 50 measurements (8.00%)
  3 (6.00%) low mild
  1 (2.00%) high mild

forward_pass_optimized/lffn_mlp_workspace/200
                        time:   [165.62 µs 165.95 µs 166.23 µs]
                        thrpt:  [1.2032 Melem/s 1.2052 Melem/s 1.2076 Melem/s]
                 change:
                        time:   [-45.790% -45.583% -45.413%] (p = 0.00 < 0.05)
                        thrpt:  [+83.193% +83.765% +84.468%]
                        Performance has improved.
forward_pass_optimized/lffn_mlp_workspace/1000
                        time:   [864.18 µs 866.52 µs 870.61 µs]
                        thrpt:  [1.1486 Melem/s 1.1540 Melem/s 1.1572 Melem/s]
                 change:
                        time:   [+182.49% +183.29% +184.42%] (p = 0.00 < 0.05)
                        thrpt:  [-64.840% -64.701% -64.600%]
                        Performance has regressed.
Found 4 outliers among 50 measurements (8.00%)
  3 (6.00%) low mild
  1 (2.00%) high severe
forward_pass_optimized/lffn_mlp_workspace/10000
                        time:   [13.047 ms 13.486 ms 13.931 ms]
                        thrpt:  [717.84 Kelem/s 741.49 Kelem/s 766.48 Kelem/s]
                 change:
                        time:   [+3995.2% +4125.8% +4266.8%] (p = 0.00 < 0.05)
                        thrpt:  [-97.710% -97.634% -97.558%]
                        Performance has regressed.
Found 1 outliers among 50 measurements (2.00%)
  1 (2.00%) high mild
forward_pass_optimized/lffn_mlp_workspace/100000
                        time:   [147.69 ms 149.19 ms 150.75 ms]
                        thrpt:  [663.34 Kelem/s 670.30 Kelem/s 677.08 Kelem/s]
                 change:
                        time:   [+20908% +21105% +21319%] (p = 0.00 < 0.05)
                        thrpt:  [-99.533% -99.528% -99.524%]
                        Performance has regressed.
Found 1 outliers among 50 measurements (2.00%)
  1 (2.00%) high mild

pde_residual_analytical/helmholtz_cpu/200
                        time:   [5.2198 µs 5.2210 µs 5.2221 µs]
                        thrpt:  [38.299 Melem/s 38.307 Melem/s 38.316 Melem/s]
                 change:
                        time:   [-0.4450% -0.4153% -0.3850%] (p = 0.00 < 0.05)
                        thrpt:  [+0.3865% +0.4171% +0.4470%]
                        Change within noise threshold.
Found 3 outliers among 50 measurements (6.00%)
  1 (2.00%) low mild
  2 (4.00%) high mild
pde_residual_analytical/helmholtz_cpu/1000
                        time:   [25.163 µs 25.167 µs 25.173 µs]
                        thrpt:  [39.726 Melem/s 39.734 Melem/s 39.740 Melem/s]
                 change:
                        time:   [+0.0934% +0.1150% +0.1422%] (p = 0.00 < 0.05)
                        thrpt:  [-0.1420% -0.1148% -0.0933%]
                        Change within noise threshold.
Found 2 outliers among 50 measurements (4.00%)
  1 (2.00%) high mild
  1 (2.00%) high severe
pde_residual_analytical/helmholtz_cpu/10000
                        time:   [244.57 µs 244.59 µs 244.60 µs]
                        thrpt:  [40.883 Melem/s 40.885 Melem/s 40.887 Melem/s]
                 change:
                        time:   [+0.0740% +0.0828% +0.0919%] (p = 0.00 < 0.05)
                        thrpt:  [-0.0918% -0.0828% -0.0739%]
                        Change within noise threshold.
Found 7 outliers among 50 measurements (14.00%)
  1 (2.00%) low mild
  2 (4.00%) high mild
  4 (8.00%) high severe

pde_residual_tensor/helmholtz_tensor/200
                        time:   [8.9912 µs 8.9927 µs 8.9941 µs]
                        thrpt:  [22.237 Melem/s 22.240 Melem/s 22.244 Melem/s]
                 change:
                        time:   [-94.901% -94.898% -94.896%] (p = 0.00 < 0.05)
                        thrpt:  [+1859.2% +1860.2% +1861.1%]
                        Performance has improved.
Found 1 outliers among 50 measurements (2.00%)
  1 (2.00%) low mild
pde_residual_tensor/helmholtz_tensor/1000
                        time:   [32.475 µs 32.480 µs 32.486 µs]
                        thrpt:  [30.783 Melem/s 30.788 Melem/s 30.793 Melem/s]
                 change:
                        time:   [-86.501% -86.489% -86.474%] (p = 0.00 < 0.05)
                        thrpt:  [+639.30% +640.12% +640.79%]
                        Performance has improved.
Found 2 outliers among 50 measurements (4.00%)
  1 (2.00%) high mild
  1 (2.00%) high severe
pde_residual_tensor/helmholtz_tensor/10000
                        time:   [286.30 µs 286.33 µs 286.36 µs]
                        thrpt:  [34.921 Melem/s 34.925 Melem/s 34.928 Melem/s]
                 change:
                        time:   [-52.384% -52.367% -52.349%] (p = 0.00 < 0.05)
                        thrpt:  [+109.86% +109.94% +110.02%]
                        Performance has improved.
Found 2 outliers among 50 measurements (4.00%)
  1 (2.00%) high mild
  1 (2.00%) high severe

mse_loss/mse_computation/200
                        time:   [1.2442 µs 1.2445 µs 1.2448 µs]
                        thrpt:  [321.34 Melem/s 321.42 Melem/s 321.49 Melem/s]
                 change:
                        time:   [-97.569% -97.568% -97.567%] (p = 0.00 < 0.05)
                        thrpt:  [+4009.6% +4012.0% +4014.3%]
                        Performance has improved.
mse_loss/mse_computation/1000
                        time:   [2.5931 µs 2.5950 µs 2.5963 µs]
                        thrpt:  [770.32 Melem/s 770.72 Melem/s 771.28 Melem/s]
                 change:
                        time:   [-95.777% -95.772% -95.767%] (p = 0.00 < 0.05)
                        thrpt:  [+2262.6% +2265.2% +2267.9%]
                        Performance has improved.
mse_loss/mse_computation/10000
                        time:   [20.134 µs 20.139 µs 20.144 µs]
                        thrpt:  [992.83 Melem/s 993.09 Melem/s 993.34 Melem/s]
                 change:
                        time:   [-86.724% -86.714% -86.702%] (p = 0.00 < 0.05)
                        thrpt:  [+652.02% +652.66% +653.24%]
                        Performance has improved.
Found 4 outliers among 50 measurements (8.00%)
  2 (4.00%) low mild
  2 (4.00%) high mild

training_step/single_step/200
                        time:   [1.0369 ms 1.0374 ms 1.0382 ms]
                        thrpt:  [192.65 Kelem/s 192.79 Kelem/s 192.88 Kelem/s]
                 change:
                        time:   [-35.586% -35.519% -35.445%] (p = 0.00 < 0.05)
                        thrpt:  [+54.907% +55.086% +55.246%]
                        Performance has improved.
training_step/single_step/1000
                        time:   [8.0317 ms 8.0406 ms 8.0539 ms]
                        thrpt:  [124.16 Kelem/s 124.37 Kelem/s 124.51 Kelem/s]
                 change:
                        time:   [+48.329% +48.726% +49.214%] (p = 0.00 < 0.05)
                        thrpt:  [-32.982% -32.762% -32.582%]
                        Performance has regressed.
Found 7 outliers among 30 measurements (23.33%)
  3 (10.00%) low severe
  1 (3.33%) low mild
  1 (3.33%) high mild
  2 (6.67%) high severe

training_step_optimized/single_step_workspace/200
                        time:   [171.20 µs 172.39 µs 173.31 µs]
                        thrpt:  [1.1540 Melem/s 1.1602 Melem/s 1.1683 Melem/s]
                 change:
                        time:   [-68.604% -68.497% -68.378%] (p = 0.00 < 0.05)
                        thrpt:  [+216.23% +217.43% +218.51%]
                        Performance has improved.
training_step_optimized/single_step_workspace/1000
                        time:   [898.25 µs 899.97 µs 901.74 µs]
                        thrpt:  [1.1090 Melem/s 1.1111 Melem/s 1.1133 Melem/s]
                 change:
                        time:   [+43.873% +44.644% +45.800%] (p = 0.00 < 0.05)
                        thrpt:  [-31.413% -30.865% -30.494%]
                        Performance has regressed.
Found 3 outliers among 30 measurements (10.00%)
  1 (3.33%) low mild
  1 (3.33%) high mild
  1 (3.33%) high severe

training_step_cached/single_step_cached/200
                        time:   [1.0186 ms 1.0190 ms 1.0193 ms]
                        thrpt:  [196.21 Kelem/s 196.28 Kelem/s 196.34 Kelem/s]
                 change:
                        time:   [-35.771% -35.716% -35.650%] (p = 0.00 < 0.05)
                        thrpt:  [+55.400% +55.559% +55.693%]
                        Performance has improved.
Found 1 outliers among 30 measurements (3.33%)
  1 (3.33%) high severe
training_step_cached/single_step_cached/1000
                        time:   [8.0022 ms 8.0072 ms 8.0140 ms]
                        thrpt:  [124.78 Kelem/s 124.89 Kelem/s 124.97 Kelem/s]
                 change:
                        time:   [+48.982% +49.142% +49.315%] (p = 0.00 < 0.05)
                        thrpt:  [-33.028% -32.950% -32.878%]
                        Performance has regressed.
Found 3 outliers among 30 measurements (10.00%)
  1 (3.33%) high mild
  2 (6.67%) high severe

training_step_fully_optimized/single_step_fully_opt/200
                        time:   [162.95 µs 163.98 µs 165.10 µs]
                        thrpt:  [1.2114 Melem/s 1.2196 Melem/s 1.2273 Melem/s]
                 change:
                        time:   [-68.479% -68.370% -68.284%] (p = 0.00 < 0.05)
                        thrpt:  [+215.30% +216.16% +217.25%]
                        Performance has improved.
Found 4 outliers among 30 measurements (13.33%)
  4 (13.33%) low severe
training_step_fully_optimized/single_step_fully_opt/1000
                        time:   [866.81 µs 867.35 µs 867.96 µs]
                        thrpt:  [1.1521 Melem/s 1.1529 Melem/s 1.1537 Melem/s]
                 change:
                        time:   [+50.825% +51.115% +51.365%] (p = 0.00 < 0.05)
                        thrpt:  [-33.934% -33.825% -33.698%]
                        Performance has regressed.
Found 12 outliers among 30 measurements (40.00%)
  7 (23.33%) low severe
  5 (16.67%) high severe

training_100_epochs/train_100/200
                        time:   [104.17 ms 104.43 ms 104.62 ms]
                        thrpt:  [955.83  elem/s 957.58  elem/s 959.94  elem/s]
                 change:
                        time:   [-35.478% -35.362% -35.250%] (p = 0.00 < 0.05)
                        thrpt:  [+54.439% +54.709% +54.986%]
                        Performance has improved.

training_100_epochs_cached/train_100_cached/200
                        time:   [102.61 ms 102.84 ms 103.17 ms]
                        thrpt:  [969.24  elem/s 972.41  elem/s 974.61  elem/s]
                 change:
                        time:   [-35.437% -35.276% -35.114%] (p = 0.00 < 0.05)
                        thrpt:  [+54.115% +54.502% +54.888%]
                        Performance has improved.

training_100_epochs_fully_optimized/train_100_fully_opt/200
                        time:   [16.746 ms 16.894 ms 17.136 ms]
                        thrpt:  [5.8357 Kelem/s 5.9193 Kelem/s 5.9715 Kelem/s]
                 change:
                        time:   [-67.842% -67.374% -66.941%] (p = 0.00 < 0.05)
                        thrpt:  [+202.49% +206.50% +210.97%]
                        Performance has improved.

