Commits

Branch selector

 main

User selector

eqy

Datepicker

All time

Commit History

Commits on Jun 28, 2023

• [cuDNN][cuDNN V8 API] Improve safety of ParamsHash keys (pytorch#104122)

  Show description for 2504af5
  

  eqy

  authored and

  pytorchmergebot

  committed
  2504af5

  Copy full SHA for 2504af5

Commits on Jun 23, 2023

• [cuDNN][cuDNN V8 API] Thread safety fixes for cuDNN V8 API (pytorch#103939)

  Show description for ef3d1cf
  

  eqy

  authored and

  pytorchmergebot

  committed
  ef3d1cf

  Copy full SHA for ef3d1cf

Commits on May 26, 2023

• [CUDA][DLPack] Try ~~bumping sleep interval~~ running on explicit side-stream for Windows dlpack test (pytorch#102283)

  Show description for 54f3838
  

  eqy

  authored and

  pytorchmergebot

  committed
  54f3838

  Copy full SHA for 54f3838

Commits on May 24, 2023

• [CUDA][DLPack] Handle legacy default streams for DLPack conversion (pytorch#101318)

  Show description for 66f6e0e
  

  eqy

  authored and

  pytorchmergebot

  committed
  66f6e0e

  Copy full SHA for 66f6e0e

Commits on May 11, 2023

• [cuBLAS][cuBLASLt] Allow user-specified cuBLASLt workspace size via CUBLASLT_WORKSPACE_SIZE (pytorch#101145)

  Show description for c25fdc2
  

  eqy

  authored and

  pytorchmergebot

  committed
  c25fdc2

  Copy full SHA for c25fdc2

Commits on May 10, 2023

• [CUDA][CUBLAS] Fix BF16 reduced precision reduction note in docs (pytorch#101044)

  Show description for 33f3dca
  

  eqy

  authored and

  pytorchmergebot

  committed
  33f3dca

  Copy full SHA for 33f3dca

• [CUDA][CUBLAS] Add cuBLAS workspace allocation behavior to docs (pytorch#100919)

  Show description for 6e2efd1
  

  eqy

  authored and

  pytorchmergebot

  committed
  6e2efd1

  Copy full SHA for 6e2efd1

Commits on May 4, 2023

• [cuDNN][CUDA] Fix for install_cudnn.sh following 12.1 CI update (pytorch#100501)

  Show description for 3c1dd0a
  

  eqy

  authored and

  pytorchmergebot

  committed
  3c1dd0a

  Copy full SHA for 3c1dd0a

Commits on May 3, 2023

• [cuDNN][cuDNN V8 frontend API] Clean up time_sorted_plan workaround for cuDNN v8 API (pytorch#100287)

  Show description for 8f6951c
  

  eqy

  authored and

  pytorchmergebot

  committed
  8f6951c

  Copy full SHA for 8f6951c

Commits on Apr 28, 2023

• [CUDA] Switch to at::empty_like in adaptive_avg_pool3d_backward_cuda (pytorch#100202)

  Show description for 43dea76
  

  eqy

  authored and

  pytorchmergebot

  committed
  43dea76

  Copy full SHA for 43dea76

• Bump cuDNN frontend submodule to 0.9 (pytorch#99674)

  Show description for 5022143
  

  eqy

  authored and

  pytorchmergebot

  committed
  5022143

  Copy full SHA for 5022143

• [CUDA] Do accumulation for Adaptive Average Pooling in opmath_t (pytorch#99378)

  Show description for 3f656ad
  

  eqy

  authored and

  pytorchmergebot

  committed
  3f656ad

  Copy full SHA for 3f656ad

Commits on Apr 27, 2023

• [CUDA] Switch to at::empty in max_pool3d_with_indices_backward_cuda (pytorch#100138)

  Show description for 925a378
  

  eqy

  authored and

  pytorchmergebot

  committed
  925a378

  Copy full SHA for 925a378

Commits on Apr 18, 2023

• [CUBLAS] Specify alignment for cuBlasLt addmm (pytorch#98975)

  Show description for 5c5ad53
  

  eqy

  authored and

  pytorchmergebot

  committed
  5c5ad53

  Copy full SHA for 5c5ad53

Commits on Apr 14, 2023

• [NCCL] Add experimental Nonblocking NCCL Fault Tolerance/Checking (pytorch#95715)

  Show description for c4f81cb
  

  eqy

  authored and

  pytorchmergebot

  committed
  c4f81cb

  Copy full SHA for c4f81cb

Commits on Apr 12, 2023

• [NCCL] Add experimental Nonblocking NCCL Fault Tolerance/Checking (pytorch#95715)

  Show description for a33eac3
  

  eqy

  authored and

  pytorchmergebot

  committed
  a33eac3

  Copy full SHA for a33eac3

Commits on Mar 30, 2023

• [CUDA][CUDA 11] Remove more CUDA 11 version checks (pytorch#92934)

  Show description for 2fddcf0
  

  eqy

  authored and

  pytorchmergebot

  committed
  2fddcf0

  Copy full SHA for 2fddcf0

Commits on Mar 29, 2023

• [cuDNN][cuDNN V8 API] Fix incorrect use of emplace in the benchmark cache (pytorch#97838)

  Show description for aa4ea6e
  

  eqy

  authored and

  pytorchmergebot

  committed
  aa4ea6e

  Copy full SHA for aa4ea6e

Commits on Mar 25, 2023

• [NCCL] Cleanup NCCL-no-record streams, move to TORCH_NCCL_AVOID_RECORD_STREAMS (pytorch#97053)

  Show description for 0781188
  

  eqy

  authored and

  pytorchmergebot

  committed
  0781188

  Copy full SHA for 0781188

Commits on Mar 14, 2023

• [CUDA] Abate spurious resize warnings in MultiMarginLoss backward (pytorch#96382)

  Show description for 70090b4
  

  eqy

  authored and

  pytorchmergebot

  committed
  70090b4

  Copy full SHA for 70090b4

• [CUDA][cuFFT] Minor fix for cuFFT plan cache docs (pytorch#96373)

  Show description for 6e3e22d
  

  eqy

  authored and

  pytorchmergebot

  committed
  6e3e22d

  Copy full SHA for 6e3e22d

Commits on Mar 2, 2023

• [CUDA] Split out compute capability 8.7 and 7.2 from others (pytorch#95803)

  Show description for db8e91e
  

  eqy

  authored and

  pytorchmergebot

  committed
  db8e91e

  Copy full SHA for db8e91e

Commits on Feb 27, 2023

• [NCCL] (re-open) Optionally avoid recordStream calls in ProcessGroupNCCL (pytorch#89880)

  Show description for 29f9a70
  

  eqy

  authored and

  pytorchmergebot

  committed
  29f9a70

  Copy full SHA for 29f9a70

Commits on Feb 24, 2023

• [CUDA][CUBLAS] Explicitly link against cuBLASLt (pytorch#95094)

  Show description for cc39cd6
  

  eqy

  authored and

  pytorchmergebot

  committed
  cc39cd6

  Copy full SHA for cc39cd6

Commits on Feb 17, 2023

• [CUDA] sm_87 / Jetson Orin support (pytorch#95008)

  Show description for 13ebffe
  

  eqy

  authored and

  pytorchmergebot

  committed
  13ebffe

  Copy full SHA for 13ebffe

Commits on Jan 24, 2023

• [CUDA] Drop CUDA < 11.0 test flags (pytorch#92605)

  Show description for 0bf7506
  

  eqy

  authored and

  pytorchmergebot

  committed
  0bf7506

  Copy full SHA for 0bf7506

Commits on Jan 21, 2023

• [cublas][cublasLt] Fall back to unfused addmm for 2-byte-aligned inputs (pytorch#92201)

  Show description for 1af40d5
  

  eqy

  authored and

  pytorchmergebot

  committed
  1af40d5

  Copy full SHA for 1af40d5

Commits on Jan 20, 2023

• [CUDA][CUBLAS][BFloat16] Tenatively disable reduced precision reductions for some matmul tests (pytorch#92599)

  Show description for 8c0289a
  

  eqy

  authored and

  pytorchmergebot

  committed
  8c0289a

  Copy full SHA for 8c0289a

Commits on Jan 13, 2023

• [cuDNN][cuDNN V8 API] Always build assuming cuDNN >= 8.0 (pytorch#91527)

  Show description for 4d07ad7
  

  eqy

  authored and

  pytorchmergebot

  committed
  4d07ad7

  Copy full SHA for 4d07ad7

Commits on Jan 12, 2023

• [BC-Breaking] Separate stream_id, device_index, and device_type in pack and unpack for Streams (pytorch#81596)

  Show description for e096d2d
  

  eqy

  authored and

  pytorchmergebot

  committed
  e096d2d

  Copy full SHA for e096d2d

• [cuBLAS][TF32] Fix TF32 get/set test when TORCH_ALLOW_TF32_CUBLAS_OVERRIDE is set (pytorch#92052)

  Show description for fb38b9f
  

  eqy

  authored and

  pytorchmergebot

  committed
  fb38b9f

  Copy full SHA for fb38b9f

• Fix CUDA_MAX_THREADS_PER_SM for sm_89 (pytorch#91972)

  Show description for f802641
  

  eqy

  authored and

  pytorchmergebot

  committed
  f802641

  Copy full SHA for f802641

Commits on Jan 11, 2023

• [CUBLAS][TF32] Change cuBLAS TF32 environment variable to be initialization only (pytorch#85859)

  Show description for 19723d7
  

  eqy

  authored and

  pytorchmergebot

  committed
  19723d7

  Copy full SHA for 19723d7

Commits on Jan 5, 2023

• [CUDA] Drop CUDA 10 support (pytorch#89582)

  Show description for bac33ea
  

  eqy

  authored and

  pytorchmergebot

  committed
  bac33ea

  Copy full SHA for bac33ea

Commits on Jan 3, 2023

• [cuBLAS] (re-open) Fix default cuBLAS workspace size and parsing for multiple workspaces (pytorch#91564)

  Show description for 97ff20d
  

  eqy

  authored and

  pytorchmergebot

  committed
  97ff20d

  Copy full SHA for 97ff20d

Pagination

Previous

Next