OSSForks/pytorch
1
0
Fork 0
mirror of https://github.com/zebrajr/pytorch.git synced 2025-12-07 12:21:27 +01:00
Code Issues Actions 38 Packages Projects Releases Wiki Activity
b51b7e28ee
pytorch/third_party
History
Matthew Sterrett 7e65060410 Adds support for accelerated sorting with x86-simd-sort (#127936) 
Adds x86-simd-sort as a submodule to accelerate sorting for 32-bit and 64-bit datatypes when AVX2 or AVX512 are available.

For contiguous data, this can be over a 10x speedup for large arrays. For discontiguous data, it can give over a 4x speedup with larger arrays. These benchmarks were gathered on a Skylake system (7900x), limited to 8 threads.

<details>
<summary><b>Contiguous Benchmarks</b></summary>

```
float32, normally distributed (in microseconds)
size           Default        AVX2           AVX512         Default/AVX2   Default/AVX512
16             7.150844336    6.886271477    7.132277489    1.038420335    1.002603214
128            9.208030939    8.478154898    7.846915245    1.086089019    1.173458697
1024           37.79037627    23.60707456    16.44122627    1.600807257    2.298513241
10000          714.7355628    203.9921844    105.5683001    3.503739934    6.770361577
100000         8383.074408    721.6333354    465.3709247    11.61680593    18.01374766
1000000        97124.31945    5632.054572    3920.148401    17.24491803    24.77567416
10000000       1161974.907    86070.48988    71533.82301    13.50027063    16.24371323

int32_t, uniformly distributed (in microseconds)
size           Default        AVX2           AVX512         Default/AVX2   Default/AVX512
16             7.203208685    6.92212224     7.014458179    1.040606975    1.026908779
128            8.972388983    8.195516348    7.592543125    1.094792396    1.18173698
1024           32.77489477    23.6874548     15.36617105    1.383639359    2.132925285
10000          607.8824128    193.3402024    99.25090471    3.144107667    6.124703997
100000         523.9384684    608.1836536    442.3166784    0.861480682    1.184532472
1000000        5211.348627    5271.598405    3518.861883    0.988570871    1.480975611
10000000       133853.6263    81463.05084    67852.97394    1.643120714    1.972700952
```

</details>

Note that the int32_t sort is accelerated by FBGEMM's radix sort for larger arrays, but this only handles contiguous data and in one sorting direction.

<details>
<summary><b>Discontiguous Benchmarks</b></summary>

```
float, normal distributed, discontiguous in sorted dimension (in microseconds)
size           Default        AVX2           AVX512         Default/AVX2   Default/AVX512
16             3.836543679    4.011214256    3.84376061     0.956454439    0.99812243
128            5.755310194    5.755723127    4.820394962    0.999928257    1.193949923
1024           49.46946019    24.78790785    15.47874362    1.995709379    3.195960952
10000          665.2505291    236.6165959    143.9490662    2.811512551    4.621429974
100000         4328.002203    1329.001212    818.3516414    3.256582586    5.288682743
1000000        47651.5018     16693.72045    11827.39551    2.854456677    4.028909133
10000000       556655.1288    236252.6258    184215.9828    2.356185998    3.021752621

int32_t, uniformly distributed, discontiguous in sorted dimension  (in microseconds)
size           Default        AVX2           AVX512         Default/AVX2   Default/AVX512
16             3.817994356    3.878117442    3.770039797    0.984496837    1.012719908
128            5.578731397    5.577152082    4.716770534    1.000283176    1.182743862
1024           43.3412619     23.61275801    14.55446819    1.835501887    2.977866408
10000          634.3997478    224.4322851    133.9518324    2.826686667    4.736028889
100000         4084.358152    1292.363303    781.7867576    3.16037924     5.22438902
1000000        46262.20465    16608.35284    11367.51817    2.785478192    4.06968381
10000000       541231.9104    235185.1861    180249.9294    2.301301028    3.002674742
```

</details>

Pull Request resolved: https://github.com/pytorch/pytorch/pull/127936
Approved by: https://github.com/jgong5, https://github.com/peterbell10, https://github.com/sanchitintel
2024-11-02 02:14:01 +00:00
..
benchmark@0d98dba29d
composable_kernel@cedccd59c9 [ROCm][CK] Explicit cast values to half (#138751) 2024-10-28 22:00:26 +00:00
cpp-httplib@3b6597bba9
cpuinfo@1e83a2fdd3 Update cpuinfo submodule (#138351) 2024-10-19 01:12:29 +00:00
cudnn_frontend@936021bfed [BE]: Update cudnn_frontend submodule to 1.8.0 (#138709) 2024-10-26 01:55:33 +00:00
cutlass@bbe579a9e3 Revert "[CUDA][CUTLASS][submodule] Fixes for CUTLASS upgrade (#131493)" 2024-08-16 18:09:33 +00:00
eigen@3147391d94
fbgemm@dbc3157bf2
flatbuffers@01834de25e
fmt@0c9fce2ffe [BE][Ez]: Update fmtlib submodule to 11.0.2 (#132036) 2024-07-29 15:50:00 +00:00
FP16@4dfe081cf6
FXdiv@b408327ac2
gemmlowp
gloo@5354032ea0
googletest@e2239ee604
ideep@41d636c2bb Update submodule ideep to include aarch64 change (#134897) 2024-09-06 16:40:26 +00:00
ittapi@5b8a7d7422
kineto@ed052ea024 [Profiler] Add Test for Clear on Fork (#137511) 2024-10-14 23:20:33 +00:00
mimalloc@b66e3214d8
miniz-2.1.0 reimport pr137735 due to merging check issues (#138959) 2024-10-27 16:31:34 +00:00
nccl [BE]: Update NCCL submodule to 2.21.5 (#124014) 2024-07-02 14:39:33 +00:00
nlohmann@87cda1d664
NNPACK@c07e3a0400
NVTX@e170594ac7 [Reland2] Update NVTX to NVTX3 (#109843) 2024-08-20 16:33:26 +00:00
onnx@b8baa84466 [Submodule] update submodule onnx==1.17.0 (#139128) 2024-10-31 02:50:00 +00:00
opentelemetry-cpp@a799f4aed9
pocketfft@9d3ab05a7f
protobuf@d1eca4e4b4
psimd@072586a71b
pthreadpool@4fe0e1e183
pybind11@a2e59f0e70 [BE][Ez]: Update pybind11 to 2.13.6. Exposes new conduit cross-compat API (#136087) 2024-09-14 20:48:44 +00:00
python-peachpy@f45429b087
sleef@60e76d2bce
tensorflow_cuda_bazel_build/cuda
tensorpipe@52791a2fd2
valgrind-headers
VulkanMemoryAllocator@a6bfc23725
x86-simd-sort@f99c392904 Adds support for accelerated sorting with x86-simd-sort (#127936) 2024-11-02 02:14:01 +00:00
XNNPACK@87ee0b46b8 Update PyTorch for XNNPACK 87ee0b4 (#134518) 2024-08-28 19:24:04 +00:00
BUCK.oss
BUILD
build_bundled.py Fix manual licensing (#128630) 2024-06-14 00:12:09 +00:00
cpp-httplib.BUILD Reapply "distributed debug handlers (#126601)" (#127805) 2024-06-04 19:44:30 +00:00
cuda.BUILD [Reland] Add wrappers for synchronous GPUDirect Storage APIs (#133489) 2024-08-15 17:11:52 +00:00
cudnn_frontend.BUILD
cudnn.BUILD
cutlass.BUILD [BE][CUDA][Bugfix]: Enable extended MMA shapes in CUTLASS. (#133686) 2024-09-28 21:11:15 +00:00
eigen.BUILD
fmt.BUILD
generate-cpuinfo-wrappers.py
generate-xnnpack-wrappers.py Update generate-xnnpack-wrappers.py parsing to handle build identifier (#134724) 2024-09-04 08:45:46 +00:00
glog.buck.bzl
gloo.BUILD
ideep.BUILD
kineto.buck.bzl [lint] Remove unnecessary BUCKRESTRICTEDSYNTAX suppressions 2024-07-19 07:19:11 -07:00
kineto.BUILD
LICENSES_BUNDLED.txt Fix manual licensing (#128630) 2024-06-14 00:12:09 +00:00
METADATA.bzl
mkl_headers.BUILD
mkl-dnn.BUILD Add oneDNN BRGEMM support on CPU (#131878) 2024-09-07 13:22:30 +00:00
mkl.BUILD [Submodule] Remove deprecated USE_TBB option and TBB submodule (#127051) 2024-05-31 01:20:45 +00:00
nlohmann.BUILD [aoti] Add initial custom op support (#127034) 2024-07-24 20:29:55 +00:00
onnx.BUILD
opentelemetry-cpp.BUILD
README.md
sleef.BUILD
sleef.bzl
substitution.bzl
tensorpipe.BUILD
xnnpack_src_defs.bzl Update PyTorch for XNNPACK 87ee0b4 (#134518) 2024-08-28 19:24:04 +00:00
xnnpack_wrapper_defs.bzl Update PyTorch for XNNPACK 87ee0b4 (#134518) 2024-08-28 19:24:04 +00:00
xnnpack.buck.bzl [xplat][XNNPACK] don't prefer static linkage in xplat for main target (#135529) 2024-09-09 22:47:01 +00:00
xpu.txt Update torch-xpu-ops commit pin (#139041) 2024-10-31 05:06:06 +00:00

README.md

This folder contains vendored copies of third-party libraries that we use.