Skip to content

Add PDL to cub::DeviceRadixSort#9247

Draft
gonidelis wants to merge 15 commits into
NVIDIA:mainfrom
gonidelis:pdl_radix_sort
Draft

Add PDL to cub::DeviceRadixSort#9247
gonidelis wants to merge 15 commits into
NVIDIA:mainfrom
gonidelis:pdl_radix_sort

Conversation

@gonidelis
Copy link
Copy Markdown
Member

@gonidelis gonidelis commented Jun 4, 2026

do not review

fixes #8765

BEFORE

image

AFTER

image

zoomed in:

image

@copy-pr-bot
Copy link
Copy Markdown
Contributor

copy-pr-bot Bot commented Jun 4, 2026

Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

@cccl-authenticator-app cccl-authenticator-app Bot moved this from Todo to In Progress in CCCL Jun 4, 2026
@gonidelis
Copy link
Copy Markdown
Member Author

gonidelis commented Jun 4, 2026

Full sweep final perf results on B200 here or below:

['main.json', 'pdl_final.json']
# base

## [0] NVIDIA B200

|  T{ct}  |  OffsetT{ct}  |  Elements{io}  |  Entropy  |   Ref Time |   Ref Noise |   Cmp Time |   Cmp Noise |       Diff |   %Diff |  Status  |
|---------|---------------|----------------|-----------|------------|-------------|------------|-------------|------------|---------|----------|
|   I8    |      I32      |      2^16      |     1     |  24.470 us |       1.42% |  20.490 us |       1.12% |  -3.980 us | -16.26% |  🟢 FAST  |
|   I8    |      I32      |      2^20      |     1     |  31.670 us |       1.84% |  27.926 us |       3.37% |  -3.745 us | -11.82% |  🟢 FAST  |
|   I8    |      I32      |      2^24      |     1     | 120.148 us |       0.65% | 120.133 us |       0.64% |  -0.014 us |  -0.01% |  🔵 SAME  |
|   I8    |      I32      |      2^28      |     1     |   1.310 ms |       0.25% |   1.311 ms |       0.26% |   0.588 us |   0.04% |  🔵 SAME  |
|   I8    |      I32      |      2^16      |   0.544   |  23.532 us |       4.30% |  20.389 us |       1.35% |  -3.143 us | -13.36% |  🟢 FAST  |
|   I8    |      I32      |      2^20      |   0.544   |  31.454 us |       2.86% |  28.008 us |       3.01% |  -3.446 us | -10.96% |  🟢 FAST  |
|   I8    |      I32      |      2^24      |   0.544   | 122.896 us |       0.40% | 123.446 us |       0.69% |   0.550 us |   0.45% |  🔴 SLOW  |
|   I8    |      I32      |      2^28      |   0.544   |   1.339 ms |       0.11% |   1.339 ms |       0.11% |   0.376 us |   0.03% |  🔵 SAME  |
|   I8    |      I32      |      2^16      |   0.201   |  23.941 us |       2.21% |  20.041 us |       3.24% |  -3.901 us | -16.29% |  🟢 FAST  |
|   I8    |      I32      |      2^20      |   0.201   |  29.798 us |       2.18% |  25.989 us |       3.65% |  -3.810 us | -12.79% |  🟢 FAST  |
|   I8    |      I32      |      2^24      |   0.201   | 114.650 us |       0.90% | 114.747 us |       0.89% |   0.098 us |   0.09% |  🔵 SAME  |
|   I8    |      I32      |      2^28      |   0.201   |   1.257 ms |       0.27% |   1.257 ms |       0.28% |   0.165 us |   0.01% |  🔵 SAME  |
|   I8    |      I64      |      2^16      |     1     |  24.170 us |       2.69% |  20.412 us |       1.62% |  -3.758 us | -15.55% |  🟢 FAST  |
|   I8    |      I64      |      2^20      |     1     |  31.593 us |       1.17% |  27.142 us |       3.36% |  -4.451 us | -14.09% |  🟢 FAST  |
|   I8    |      I64      |      2^24      |     1     | 120.438 us |       0.80% | 120.312 us |       0.74% |  -0.126 us |  -0.10% |  🔵 SAME  |
|   I8    |      I64      |      2^28      |     1     |   1.327 ms |       0.11% |   1.327 ms |       0.12% |   0.012 us |   0.00% |  🔵 SAME  |
|   I8    |      I64      |      2^16      |   0.544   |  24.046 us |       3.67% |  20.260 us |       2.19% |  -3.787 us | -15.75% |  🟢 FAST  |
|   I8    |      I64      |      2^20      |   0.544   |  31.989 us |       2.97% |  27.746 us |       3.26% |  -4.243 us | -13.26% |  🟢 FAST  |
|   I8    |      I64      |      2^24      |   0.544   | 122.254 us |       0.68% | 122.166 us |       0.61% |  -0.089 us |  -0.07% |  🔵 SAME  |
|   I8    |      I64      |      2^28      |   0.544   |   1.339 ms |       0.24% |   1.339 ms |       0.25% |   0.047 us |   0.00% |  🔵 SAME  |
|   I8    |      I64      |      2^16      |   0.201   |  23.570 us |       1.05% |  20.060 us |       4.08% |  -3.510 us | -14.89% |  🟢 FAST  |
|   I8    |      I64      |      2^20      |   0.201   |  29.678 us |       0.27% |  25.735 us |       2.83% |  -3.943 us | -13.29% |  🟢 FAST  |
|   I8    |      I64      |      2^24      |   0.201   | 112.713 us |       0.55% | 112.600 us |       0.35% |  -0.114 us |  -0.10% |  🔵 SAME  |
|   I8    |      I64      |      2^28      |   0.201   |   1.271 ms |       0.12% |   1.271 ms |       0.12% |   0.048 us |   0.00% |  🔵 SAME  |
|   I16   |      I32      |      2^16      |     1     |  35.083 us |       1.86% |  29.884 us |       1.90% |  -5.200 us | -14.82% |  🟢 FAST  |
|   I16   |      I32      |      2^20      |     1     |  46.326 us |       1.48% |  41.028 us |       2.48% |  -5.298 us | -11.44% |  🟢 FAST  |
|   I16   |      I32      |      2^24      |     1     | 201.302 us |       0.59% | 201.490 us |       0.59% |   0.188 us |   0.09% |  🔵 SAME  |
|   I16   |      I32      |      2^28      |     1     |   2.535 ms |       0.10% |   2.538 ms |       0.10% |   3.033 us |   0.12% |  🔴 SLOW  |
|   I16   |      I32      |      2^16      |   0.544   |  35.363 us |       2.34% |  30.131 us |       3.01% |  -5.232 us | -14.80% |  🟢 FAST  |
|   I16   |      I32      |      2^20      |   0.544   |  47.517 us |       1.74% |  41.986 us |       1.14% |  -5.532 us | -11.64% |  🟢 FAST  |
|   I16   |      I32      |      2^24      |   0.544   | 206.325 us |       0.54% | 206.578 us |       0.53% |   0.253 us |   0.12% |  🔵 SAME  |
|   I16   |      I32      |      2^28      |   0.544   |   2.590 ms |       0.11% |   2.593 ms |       0.12% |   3.309 us |   0.13% |  🔴 SLOW  |
|   I16   |      I32      |      2^16      |   0.201   |  34.282 us |       2.57% |  29.142 us |       2.96% |  -5.140 us | -14.99% |  🟢 FAST  |
|   I16   |      I32      |      2^20      |   0.201   |  44.916 us |       2.03% |  38.927 us |       0.47% |  -5.989 us | -13.33% |  🟢 FAST  |
|   I16   |      I32      |      2^24      |   0.201   | 192.663 us |       0.59% | 192.581 us |       0.60% |  -0.082 us |  -0.04% |  🔵 SAME  |
|   I16   |      I32      |      2^28      |   0.201   |   2.419 ms |       0.12% |   2.423 ms |       0.13% |   3.264 us |   0.13% |  🔴 SLOW  |
|   I16   |      I64      |      2^16      |     1     |  34.525 us |       1.67% |  28.691 us |       1.28% |  -5.834 us | -16.90% |  🟢 FAST  |
|   I16   |      I64      |      2^20      |     1     |  47.076 us |       1.85% |  40.859 us |       1.19% |  -6.217 us | -13.21% |  🟢 FAST  |
|   I16   |      I64      |      2^24      |     1     | 200.712 us |       0.59% | 200.729 us |       0.52% |   0.017 us |   0.01% |  🔵 SAME  |
|   I16   |      I64      |      2^28      |     1     |   2.482 ms |       0.12% |   2.481 ms |       0.12% |  -0.783 us |  -0.03% |  🔵 SAME  |
|   I16   |      I64      |      2^16      |   0.544   |  34.807 us |       0.30% |  28.669 us |       0.27% |  -6.138 us | -17.63% |  🟢 FAST  |
|   I16   |      I64      |      2^20      |   0.544   |  47.345 us |       1.67% |  41.963 us |       0.67% |  -5.382 us | -11.37% |  🟢 FAST  |
|   I16   |      I64      |      2^24      |   0.544   | 203.470 us |       0.50% | 203.102 us |       0.62% |  -0.368 us |  -0.18% |  🔵 SAME  |
|   I16   |      I64      |      2^28      |   0.544   |   2.541 ms |       0.12% |   2.540 ms |       0.12% |  -0.538 us |  -0.02% |  🔵 SAME  |
|   I16   |      I64      |      2^16      |   0.201   |  33.856 us |       2.61% |  28.597 us |       1.18% |  -5.260 us | -15.54% |  🟢 FAST  |
|   I16   |      I64      |      2^20      |   0.201   |  45.026 us |       2.18% |  39.322 us |       2.28% |  -5.704 us | -12.67% |  🟢 FAST  |
|   I16   |      I64      |      2^24      |   0.201   | 191.861 us |       0.67% | 192.502 us |       0.66% |   0.641 us |   0.33% |  🔵 SAME  |
|   I16   |      I64      |      2^28      |   0.201   |   2.379 ms |       0.13% |   2.378 ms |       0.12% |  -0.747 us |  -0.03% |  🔵 SAME  |
|   I32   |      I32      |      2^16      |     1     |  54.705 us |       1.15% |  46.829 us |       2.06% |  -7.876 us | -14.40% |  🟢 FAST  |
|   I32   |      I32      |      2^20      |     1     |  76.069 us |       1.27% |  68.633 us |       0.85% |  -7.437 us |  -9.78% |  🟢 FAST  |
|   I32   |      I32      |      2^24      |     1     | 372.268 us |       0.29% | 372.443 us |       0.28% |   0.174 us |   0.05% |  🔵 SAME  |
|   I32   |      I32      |      2^28      |     1     |   4.929 ms |       0.06% |   4.929 ms |       0.06% |  -0.391 us |  -0.01% |  🔵 SAME  |
|   I32   |      I32      |      2^16      |   0.544   |  55.053 us |       1.66% |  47.265 us |       1.88% |  -7.788 us | -14.15% |  🟢 FAST  |
|   I32   |      I32      |      2^20      |   0.544   |  77.516 us |       0.92% |  69.610 us |       1.18% |  -7.906 us | -10.20% |  🟢 FAST  |
|   I32   |      I32      |      2^24      |   0.544   | 381.053 us |       0.25% | 380.514 us |       0.27% |  -0.539 us |  -0.14% |  🔵 SAME  |
|   I32   |      I32      |      2^28      |   0.544   |   5.081 ms |       0.13% |   5.080 ms |       0.12% |  -0.654 us |  -0.01% |  🔵 SAME  |
|   I32   |      I32      |      2^16      |   0.201   |  54.069 us |       1.40% |  46.044 us |       0.73% |  -8.024 us | -14.84% |  🟢 FAST  |
|   I32   |      I32      |      2^20      |   0.201   |  74.387 us |       1.02% |  66.381 us |       1.16% |  -8.006 us | -10.76% |  🟢 FAST  |
|   I32   |      I32      |      2^24      |   0.201   | 359.566 us |       0.23% | 359.762 us |       0.25% |   0.197 us |   0.05% |  🔵 SAME  |
|   I32   |      I32      |      2^28      |   0.201   |   4.732 ms |       0.09% |   4.731 ms |       0.09% |  -0.376 us |  -0.01% |  🔵 SAME  |
|   I32   |      I64      |      2^16      |     1     |  54.325 us |       0.87% |  47.078 us |       2.03% |  -7.247 us | -13.34% |  🟢 FAST  |
|   I32   |      I64      |      2^20      |     1     |  76.001 us |       1.16% |  68.927 us |       0.84% |  -7.074 us |  -9.31% |  🟢 FAST  |
|   I32   |      I64      |      2^24      |     1     | 378.250 us |       0.29% | 378.293 us |       0.31% |   0.043 us |   0.01% |  🔵 SAME  |
|   I32   |      I64      |      2^28      |     1     |   5.080 ms |       0.05% |   5.079 ms |       0.05% |  -1.095 us |  -0.02% |  🔵 SAME  |
|   I32   |      I64      |      2^16      |   0.544   |  54.686 us |       1.49% |  47.763 us |       1.57% |  -6.922 us | -12.66% |  🟢 FAST  |
|   I32   |      I64      |      2^20      |   0.544   |  76.806 us |       0.36% |  69.245 us |       0.86% |  -7.561 us |  -9.84% |  🟢 FAST  |
|   I32   |      I64      |      2^24      |   0.544   | 384.241 us |       0.24% | 383.871 us |       0.24% |  -0.370 us |  -0.10% |  🔵 SAME  |
|   I32   |      I64      |      2^28      |   0.544   |   5.224 ms |       0.06% |   5.223 ms |       0.06% |  -1.048 us |  -0.02% |  🔵 SAME  |
|   I32   |      I64      |      2^16      |   0.201   |  54.221 us |       0.87% |  46.080 us |       0.00% |  -8.141 us | -15.02% |  🟢 FAST  |
|   I32   |      I64      |      2^20      |   0.201   |  73.408 us |       1.23% |  66.647 us |       0.72% |  -6.761 us |  -9.21% |  🟢 FAST  |
|   I32   |      I64      |      2^24      |   0.201   | 363.703 us |       0.35% | 363.556 us |       0.34% |  -0.148 us |  -0.04% |  🔵 SAME  |
|   I32   |      I64      |      2^28      |   0.201   |   4.876 ms |       0.06% |   4.876 ms |       0.06% |  -0.315 us |  -0.01% |  🔵 SAME  |
|   I64   |      I32      |      2^16      |     1     |  90.076 us |       1.13% |  78.653 us |       0.94% | -11.423 us | -12.68% |  🟢 FAST  |
|   I64   |      I32      |      2^20      |     1     | 158.963 us |       0.38% | 147.786 us |       0.41% | -11.177 us |  -7.03% |  🟢 FAST  |
|   I64   |      I32      |      2^24      |     1     |   1.412 ms |       0.15% |   1.412 ms |       0.15% |  -0.150 us |  -0.01% |  🔵 SAME  |
|   I64   |      I32      |      2^28      |     1     |  21.218 ms |       0.26% |  21.752 ms |       0.04% | 534.063 us |   2.52% |  🔴 SLOW  |
|   I64   |      I32      |      2^16      |   0.544   |  91.178 us |       0.52% |  79.203 us |       0.74% | -11.975 us | -13.13% |  🟢 FAST  |
|   I64   |      I32      |      2^20      |   0.544   | 158.479 us |       0.55% | 146.142 us |       0.46% | -12.337 us |  -7.78% |  🟢 FAST  |
|   I64   |      I32      |      2^24      |   0.544   |   1.410 ms |       0.14% |   1.410 ms |       0.15% |  -0.371 us |  -0.03% |  🔵 SAME  |
|   I64   |      I32      |      2^28      |   0.544   |  21.162 ms |       0.24% |  21.162 ms |       0.23% |  -0.318 us |  -0.00% |  🔵 SAME  |
|   I64   |      I32      |      2^16      |   0.201   |  87.034 us |       0.16% |  75.519 us |       0.72% | -11.515 us | -13.23% |  🟢 FAST  |
|   I64   |      I32      |      2^20      |   0.201   | 152.515 us |       0.23% | 141.271 us |       0.35% | -11.244 us |  -7.37% |  🟢 FAST  |
|   I64   |      I32      |      2^24      |   0.201   |   1.360 ms |       0.13% |   1.360 ms |       0.13% |   0.555 us |   0.04% |  🔵 SAME  |
|   I64   |      I32      |      2^28      |   0.201   |  20.410 ms |       0.22% |  20.409 ms |       0.23% |  -1.461 us |  -0.01% |  🔵 SAME  |
|   I64   |      I64      |      2^16      |     1     |  90.883 us |       0.61% |  78.886 us |       0.34% | -11.997 us | -13.20% |  🟢 FAST  |
|   I64   |      I64      |      2^20      |     1     | 159.396 us |       0.46% | 148.083 us |       0.39% | -11.313 us |  -7.10% |  🟢 FAST  |
|   I64   |      I64      |      2^24      |     1     |   1.429 ms |       0.12% |   1.429 ms |       0.14% |  -0.062 us |  -0.00% |  🔵 SAME  |
|   I64   |      I64      |      2^28      |     1     |  21.378 ms |       0.26% |  21.386 ms |       0.25% |   8.195 us |   0.04% |  🔵 SAME  |
|   I64   |      I64      |      2^16      |   0.544   |  91.210 us |       0.43% |  79.484 us |       0.97% | -11.726 us | -12.86% |  🟢 FAST  |
|   I64   |      I64      |      2^20      |   0.544   | 158.677 us |       0.24% | 147.065 us |       0.59% | -11.612 us |  -7.32% |  🟢 FAST  |
|   I64   |      I64      |      2^24      |   0.544   |   1.428 ms |       0.12% |   1.427 ms |       0.12% |  -0.710 us |  -0.05% |  🔵 SAME  |
|   I64   |      I64      |      2^28      |   0.544   |  21.341 ms |       0.22% |  21.343 ms |       0.23% |   2.216 us |   0.01% |  🔵 SAME  |
|   I64   |      I64      |      2^16      |   0.201   |  87.940 us |       0.95% |  76.529 us |       0.84% | -11.411 us | -12.98% |  🟢 FAST  |
|   I64   |      I64      |      2^20      |   0.201   | 152.811 us |       0.42% | 141.461 us |       0.35% | -11.350 us |  -7.43% |  🟢 FAST  |
|   I64   |      I64      |      2^24      |   0.201   |   1.371 ms |       0.13% |   1.370 ms |       0.12% |  -0.217 us |  -0.02% |  🔵 SAME  |
|   I64   |      I64      |      2^28      |   0.201   |  20.561 ms |       0.23% |  20.561 ms |       0.22% |  -0.366 us |  -0.00% |  🔵 SAME  |
|  I128   |      I32      |      2^16      |     1     | 149.840 us |       0.52% | 130.056 us |       0.26% | -19.784 us | -13.20% |  🟢 FAST  |
|  I128   |      I32      |      2^20      |     1     | 339.795 us |       0.36% | 322.952 us |       0.36% | -16.843 us |  -4.96% |  🟢 FAST  |
|  I128   |      I32      |      2^24      |     1     |   3.469 ms |       0.07% |   3.469 ms |       0.07% |   0.010 us |   0.00% |  🔵 SAME  |
|  I128   |      I32      |      2^28      |     1     |  53.036 ms |       0.05% |  53.033 ms |       0.05% |  -2.701 us |  -0.01% |  🔵 SAME  |
|  I128   |      I32      |      2^16      |   0.544   | 148.759 us |       0.47% | 128.477 us |       0.67% | -20.282 us | -13.63% |  🟢 FAST  |
|  I128   |      I32      |      2^20      |   0.544   | 337.840 us |       0.33% | 320.703 us |       0.39% | -17.137 us |  -5.07% |  🟢 FAST  |
|  I128   |      I32      |      2^24      |   0.544   |   3.449 ms |       0.07% |   3.448 ms |       0.08% |  -1.093 us |  -0.03% |  🔵 SAME  |
|  I128   |      I32      |      2^28      |   0.544   |  53.403 ms |       0.02% |  53.399 ms |       0.02% |  -3.940 us |  -0.01% |  🔵 SAME  |
|  I128   |      I32      |      2^16      |   0.201   | 147.245 us |       0.68% | 127.980 us |       0.32% | -19.265 us | -13.08% |  🟢 FAST  |
|  I128   |      I32      |      2^20      |   0.201   | 333.038 us |       0.34% | 316.622 us |       0.39% | -16.417 us |  -4.93% |  🟢 FAST  |
|  I128   |      I32      |      2^24      |   0.201   |   3.404 ms |       0.08% |   3.404 ms |       0.07% |   0.371 us |   0.01% |  🔵 SAME  |
|  I128   |      I32      |      2^28      |   0.201   |  52.567 ms |       0.01% |  52.567 ms |       0.02% |  -0.165 us |  -0.00% |  🔵 SAME  |
|  I128   |      I64      |      2^16      |     1     | 150.992 us |       0.59% | 131.833 us |       0.49% | -19.159 us | -12.69% |  🟢 FAST  |
|  I128   |      I64      |      2^20      |     1     | 341.711 us |       0.35% | 324.675 us |       0.38% | -17.036 us |  -4.99% |  🟢 FAST  |
|  I128   |      I64      |      2^24      |     1     |   3.483 ms |       0.07% |   3.484 ms |       0.06% |   1.286 us |   0.04% |  🔵 SAME  |
|  I128   |      I64      |      2^28      |     1     |  53.840 ms |       0.02% |  53.847 ms |       0.01% |   7.642 us |   0.01% |  🔵 SAME  |
|  I128   |      I64      |      2^16      |   0.544   | 150.362 us |       0.39% | 130.196 us |       0.45% | -20.166 us | -13.41% |  🟢 FAST  |
|  I128   |      I64      |      2^20      |   0.544   | 339.335 us |       0.35% | 322.322 us |       0.37% | -17.013 us |  -5.01% |  🟢 FAST  |
|  I128   |      I64      |      2^24      |   0.544   |   3.464 ms |       0.06% |   3.465 ms |       0.06% |   0.158 us |   0.00% |  🔵 SAME  |
|  I128   |      I64      |      2^28      |   0.544   |  52.832 ms |       0.05% |  52.836 ms |       0.05% |   4.200 us |   0.01% |  🔵 SAME  |
|  I128   |      I64      |      2^16      |   0.201   | 149.118 us |       0.66% | 129.597 us |       0.52% | -19.521 us | -13.09% |  🟢 FAST  |
|  I128   |      I64      |      2^20      |   0.201   | 335.076 us |       0.35% | 318.236 us |       0.41% | -16.840 us |  -5.03% |  🟢 FAST  |
|  I128   |      I64      |      2^24      |   0.201   |   3.418 ms |       0.06% |   3.418 ms |       0.07% |  -0.270 us |  -0.01% |  🔵 SAME  |
|  I128   |      I64      |      2^28      |   0.201   |  52.034 ms |       0.05% |  52.036 ms |       0.05% |   2.157 us |   0.00% |  🔵 SAME  |
|   F32   |      I32      |      2^16      |     1     |  55.666 us |       0.96% |  47.824 us |       1.47% |  -7.842 us | -14.09% |  🟢 FAST  |
|   F32   |      I32      |      2^20      |     1     |  76.503 us |       1.11% |  68.807 us |       1.23% |  -7.696 us | -10.06% |  🟢 FAST  |
|   F32   |      I32      |      2^24      |     1     | 433.350 us |       0.22% | 433.219 us |       0.22% |  -0.132 us |  -0.03% |  🔵 SAME  |
|   F32   |      I32      |      2^28      |     1     |   6.083 ms |       0.07% |   6.080 ms |       0.06% |  -2.483 us |  -0.04% |  🔵 SAME  |
|   F32   |      I32      |      2^16      |   0.544   |  55.337 us |       0.55% |  48.104 us |       0.41% |  -7.233 us | -13.07% |  🟢 FAST  |
|   F32   |      I32      |      2^20      |   0.544   |  76.991 us |       1.17% |  68.716 us |       0.71% |  -8.274 us | -10.75% |  🟢 FAST  |
|   F32   |      I32      |      2^24      |   0.544   | 439.117 us |       0.38% | 438.765 us |       0.36% |  -0.352 us |  -0.08% |  🔵 SAME  |
|   F32   |      I32      |      2^28      |   0.544   |   6.151 ms |       0.11% |   6.149 ms |       0.10% |  -2.513 us |  -0.04% |  🔵 SAME  |
|   F32   |      I32      |      2^16      |   0.201   |  55.202 us |       1.17% |  47.495 us |       1.23% |  -7.707 us | -13.96% |  🟢 FAST  |
|   F32   |      I32      |      2^20      |   0.201   |  74.127 us |       1.04% |  66.667 us |       0.71% |  -7.461 us | -10.06% |  🟢 FAST  |
|   F32   |      I32      |      2^24      |   0.201   | 424.965 us |       0.26% | 424.872 us |       0.25% |  -0.094 us |  -0.02% |  🔵 SAME  |
|   F32   |      I32      |      2^28      |   0.201   |   5.923 ms |       0.13% |   5.922 ms |       0.13% |  -1.246 us |  -0.02% |  🔵 SAME  |
|   F32   |      I64      |      2^16      |     1     |  54.190 us |       0.71% |  46.626 us |       1.22% |  -7.565 us | -13.96% |  🟢 FAST  |
|   F32   |      I64      |      2^20      |     1     |  77.952 us |       0.91% |  70.679 us |       0.39% |  -7.273 us |  -9.33% |  🟢 FAST  |
|   F32   |      I64      |      2^24      |     1     | 393.466 us |       0.27% | 393.244 us |       0.28% |  -0.223 us |  -0.06% |  🔵 SAME  |
|   F32   |      I64      |      2^28      |     1     |   5.293 ms |       0.05% |   5.291 ms |       0.05% |  -1.399 us |  -0.03% |  🔵 SAME  |
|   F32   |      I64      |      2^16      |   0.544   |  54.238 us |       0.78% |  46.537 us |       1.52% |  -7.701 us | -14.20% |  🟢 FAST  |
|   F32   |      I64      |      2^20      |   0.544   |  76.881 us |       0.53% |  70.126 us |       1.27% |  -6.755 us |  -8.79% |  🟢 FAST  |
|   F32   |      I64      |      2^24      |   0.544   | 398.899 us |       0.25% | 398.591 us |       0.23% |  -0.307 us |  -0.08% |  🔵 SAME  |
|   F32   |      I64      |      2^28      |   0.544   |   5.378 ms |       0.05% |   5.376 ms |       0.05% |  -1.672 us |  -0.03% |  🔵 SAME  |
|   F32   |      I64      |      2^16      |   0.201   |  53.044 us |       1.90% |  45.904 us |       1.12% |  -7.139 us | -13.46% |  🟢 FAST  |
|   F32   |      I64      |      2^20      |   0.201   |  74.775 us |       0.56% |  67.827 us |       1.44% |  -6.948 us |  -9.29% |  🟢 FAST  |
|   F32   |      I64      |      2^24      |   0.201   | 383.314 us |       0.30% | 382.994 us |       0.29% |  -0.320 us |  -0.08% |  🔵 SAME  |
|   F32   |      I64      |      2^28      |   0.201   |   5.144 ms |       0.05% |   5.142 ms |       0.04% |  -1.541 us |  -0.03% |  🔵 SAME  |
|   F64   |      I32      |      2^16      |     1     |  95.176 us |       0.38% |  82.810 us |       0.70% | -12.365 us | -12.99% |  🟢 FAST  |
|   F64   |      I32      |      2^20      |     1     | 147.506 us |       0.60% | 135.783 us |       0.53% | -11.723 us |  -7.95% |  🟢 FAST  |
|   F64   |      I32      |      2^24      |     1     |   1.053 ms |       0.16% |   1.053 ms |       0.16% |  -0.037 us |  -0.00% |  🔵 SAME  |
|   F64   |      I32      |      2^28      |     1     |  15.328 ms |       0.03% |  15.335 ms |       0.03% |   6.902 us |   0.05% |  🔴 SLOW  |
|   F64   |      I32      |      2^16      |   0.544   |  95.445 us |       0.65% |  83.444 us |       0.79% | -12.001 us | -12.57% |  🟢 FAST  |
|   F64   |      I32      |      2^20      |   0.544   | 147.682 us |       0.56% | 135.873 us |       0.46% | -11.809 us |  -8.00% |  🟢 FAST  |
|   F64   |      I32      |      2^24      |   0.544   |   1.071 ms |       0.14% |   1.071 ms |       0.14% |  -0.214 us |  -0.02% |  🔵 SAME  |
|   F64   |      I32      |      2^28      |   0.544   |  15.567 ms |       0.03% |  15.567 ms |       0.03% |   0.498 us |   0.00% |  🔵 SAME  |
|   F64   |      I32      |      2^16      |   0.201   |  94.843 us |       0.81% |  82.650 us |       1.00% | -12.193 us | -12.86% |  🟢 FAST  |
|   F64   |      I32      |      2^20      |   0.201   | 144.343 us |       0.25% | 132.518 us |       0.56% | -11.825 us |  -8.19% |  🟢 FAST  |
|   F64   |      I32      |      2^24      |   0.201   |   1.037 ms |       0.15% |   1.037 ms |       0.15% |   0.183 us |   0.02% |  🔵 SAME  |
|   F64   |      I32      |      2^28      |   0.201   |  15.026 ms |       0.03% |  15.027 ms |       0.03% |   0.571 us |   0.00% |  🔵 SAME  |
|   F64   |      I64      |      2^16      |     1     |  95.325 us |       0.51% |  83.689 us |       0.64% | -11.636 us | -12.21% |  🟢 FAST  |
|   F64   |      I64      |      2^20      |     1     | 147.956 us |       0.43% | 136.437 us |       0.51% | -11.519 us |  -7.79% |  🟢 FAST  |
|   F64   |      I64      |      2^24      |     1     |   1.048 ms |       0.15% |   1.048 ms |       0.14% |   0.261 us |   0.02% |  🔵 SAME  |
|   F64   |      I64      |      2^28      |     1     |  15.259 ms |       0.03% |  15.264 ms |       0.03% |   5.380 us |   0.04% |  🔴 SLOW  |
|   F64   |      I64      |      2^16      |   0.544   |  95.158 us |       0.45% |  83.515 us |       0.94% | -11.643 us | -12.24% |  🟢 FAST  |
|   F64   |      I64      |      2^20      |   0.544   | 147.271 us |       0.57% | 135.626 us |       0.57% | -11.645 us |  -7.91% |  🟢 FAST  |
|   F64   |      I64      |      2^24      |   0.544   |   1.056 ms |       0.15% |   1.057 ms |       0.14% |   0.233 us |   0.02% |  🔵 SAME  |
|   F64   |      I64      |      2^28      |   0.544   |  15.490 ms |       0.03% |  15.495 ms |       0.03% |   5.296 us |   0.03% |  🔴 SLOW  |
|   F64   |      I64      |      2^16      |   0.201   |  95.195 us |       0.42% |  83.021 us |       0.60% | -12.175 us | -12.79% |  🟢 FAST  |
|   F64   |      I64      |      2^20      |   0.201   | 144.181 us |       0.51% | 132.892 us |       0.69% | -11.289 us |  -7.83% |  🟢 FAST  |
|   F64   |      I64      |      2^24      |   0.201   |   1.023 ms |       0.14% |   1.023 ms |       0.13% |  -0.064 us |  -0.01% |  🔵 SAME  |
|   F64   |      I64      |      2^28      |   0.201   |  14.979 ms |       0.03% |  14.983 ms |       0.03% |   4.474 us |   0.03% |  🔴 SLOW  |

# Summary

- Total Matches: 168
  - Pass    (diff <= min_noise): 75
  - Unknown (infinite noise):    0
  - Failure (diff > min_noise):  93

@gonidelis
Copy link
Copy Markdown
Member Author

codex suggests that this breaks C Parallel API :'| I will ping c parallel folk to keep an eye

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

Status: In Progress

Development

Successfully merging this pull request may close these issues.

Use PDL in DeviceRadixSort

1 participant