Commits

Breadcrumbs

History for

onmain

User selector

All users

Datepicker

All time

Commit History

Commits on May 3, 2025

• Revert "[ROCm] Upgrade ROCm CI to ROCm6.4 (#151368)"

  Show description for cc28b43
  pytorchmergebot

  committed
  cc28b43

  Copy full SHA for cc28b43

• [c10d] Fix extra CUDA context created by barrier (#149144)

  Show description for 457fa82
  

  kwen2501

  authored and

  pytorchmergebot

  committed
  457fa82

  Copy full SHA for 457fa82

• Remove incorrect assertion (#152653)

  Show description for 1cd68c5
  

  wconstab

  authored and

  pytorchmergebot

  committed
  1cd68c5

  Copy full SHA for 1cd68c5

• [Inductor] Add decomposeK as an autotuning choice for mm (#150654)

  Show description for 84aa098
  

  PaulZhang12

  authored and

  pytorchmergebot

  committed
  84aa098

  Copy full SHA for 84aa098

• [inductor][invoke_subgraph] Free the buffers before the subgraph call (#152494)

  Show description for 73b6b1d
  

  anijain2305

  authored and

  pytorchmergebot

  committed
  73b6b1d

  Copy full SHA for 73b6b1d

• Inductor respects exact strides on custom ops by default (#150511)

  Show description for 3d777ba
  

  zou3519

  authored and

  pytorchmergebot

  committed
  3d777ba

  Copy full SHA for 3d777ba

• Refactor layout constraint selection logic (#148104)

  Show description for 2b37a72
  

  zou3519

  authored and

  pytorchmergebot

  committed
  2b37a72

  Copy full SHA for 2b37a72

Commits on May 2, 2025

• [SymmMem] Use cub's BlockScan instead of in-house impl for offset calculation (#151993)

  Show description for 0e59b59
  

  kwen2501

  authored and

  pytorchmergebot

  committed
  0e59b59

  Copy full SHA for 0e59b59

• [FlexAttention] explicilty create grad_q w/ strides (#152641)

  Show description for a6ea63a
  

  drisspg

  authored and

  pytorchmergebot

  committed
  a6ea63a

  Copy full SHA for a6ea63a

• [associative_scan] Refactoring of input checking and dynamo invocation (#148657)

  Show description for ea12a38
  

  bohnstingl

  authored and

  pytorchmergebot

  committed
  ea12a38

  Copy full SHA for ea12a38

• [Inductor] Fix kernel argument ordering when using dynamic shapes with workspace (#152660)

  Show description for 8afe40b
  

  NikhilAPatel

  authored and

  pytorchmergebot

  committed
  8afe40b

  Copy full SHA for 8afe40b

• [Inductor] Introduce Wrapper IR line for symbolic call args (#152587)

  Show description for add4702
  

  blaine-rister

  authored and

  pytorchmergebot

  committed
  add4702

  Copy full SHA for add4702

• [inductor] Realize bucketize/searchsorted output (#152644)

  Show description for 9ae4906
  

  davidberard98

  authored and

  pytorchmergebot

  committed
  9ae4906

  Copy full SHA for 9ae4906

• [DCP] Add 30min timeout for IPC communications in async checkpointing (#152629)

  Show description for a6dd1c2
  

  MeetVadakkanchery

  authored and

  pytorchmergebot

  committed
  a6dd1c2

  Copy full SHA for a6dd1c2

• Optimize `Sequential` methods description (#147304)

  Show description for d457b44
  

  zeshengzong

  authored and

  pytorchmergebot

  committed
  d457b44

  Copy full SHA for d457b44

• [CUDA][complex] skip `test_reference_numerics_large_jiterator_unary_cuda_complex64` on CUDA (#148024)

  Show description for 216d81d
  

  eqy

  authored and

  pytorchmergebot

  committed
  216d81d

  Copy full SHA for 216d81d

• [AOTAutogradCache][Easy] Move `"einops.einops.rearrange"` to `SAFE_NON_TORCH_FUNCTIONS` (#152640)

  Show description for 16153a0
  

  StrongerXi

  authored and

  pytorchmergebot

  committed
  16153a0

  Copy full SHA for 16153a0

• [ROCm] Add support for SymmetricMemory (#150580)

  Show description for 1ea2731
  

  2 peopleauthored and

  pytorchmergebot

  committed
  1ea2731

  Copy full SHA for 1ea2731

• consolidate guard_or_x and definitely_x (#152463)

  Show description for 376529c
  

  laithsakka

  authored and

  pytorchmergebot

  committed
  376529c

  Copy full SHA for 376529c

• [StaticCudaLauncher] Ensure cuda context exists before launching kernels (#152667)

  Show description for 3baa85c
  

  jamesjwu

  authored and

  pytorchmergebot

  committed
  3baa85c

  Copy full SHA for 3baa85c

• Avoid triggering ignored requires_grad warning in our code (#152686)

  Show description for f51bee1
  

  albanD

  authored and

  pytorchmergebot

  committed
  f51bee1

  Copy full SHA for f51bee1

• [ROCm] Upgrade ROCm CI to ROCm6.4 (#151368)

  Show description for 844842d
  

  2 peopleauthored and

  pytorchmergebot

  committed
  844842d

  Copy full SHA for 844842d

• Make PGO code state not sensitive to file path by hashing file content when the file is available. (#152628)

  Show description for f65fb0a
  

  laithsakka

  authored and

  pytorchmergebot

  committed
  f65fb0a

  Copy full SHA for f65fb0a

• [invoke_subgraph] Simplify output code for subgraph output node (#152490)

  Show description for ea4b7e0
  

  anijain2305

  authored and

  pytorchmergebot

  committed
  ea4b7e0

  Copy full SHA for ea4b7e0

• [inductor][subgraph] Simplify the resulting output code for subgraph (#152383)

  Show description for f6761f2
  

  anijain2305

  authored and

  pytorchmergebot

  committed
  f6761f2

  Copy full SHA for f6761f2

• Revert "[cutlass backend] Minor lru_cache to slightly speed up filtering ops (#152577)"

  Show description for 50d4698
  pytorchmergebot

  committed
  50d4698

  Copy full SHA for 50d4698

• Fix: Consider input defined unbacked during inductor codegen for runtime asserts (#152231)

  Show description for 38a9a8b
  

  laithsakka

  authored and

  pytorchmergebot

  committed
  38a9a8b

  Copy full SHA for 38a9a8b

• [SymmMem] Add all_to_all_vdev (#151819)

  Show description for 829752b
  

  kwen2501

  authored and

  pytorchmergebot

  committed
  829752b

  Copy full SHA for 829752b

• [inductor][invoke_subgraph] Remove assertion checks for outputs of invoke_subgraph (#152384)

  Show description for 3731b70
  

  anijain2305

  authored and

  pytorchmergebot

  committed
  3731b70

  Copy full SHA for 3731b70

• [invoke_subgraph] rename identifiers to prevent python mangling (#152581)

  Show description for 9e3fc41
  

  anijain2305

  authored and

  pytorchmergebot

  committed
  9e3fc41

  Copy full SHA for 9e3fc41

• Revert "Use swap_tensors path in nn.Module.to for all subclasses that override __torch_dispatch__ (#152539)"

  Show description for 4f9f1ab
  pytorchmergebot

  committed
  4f9f1ab

  Copy full SHA for 4f9f1ab

• [SymmMem] Add all-to-all (#151498)

  Show description for d7961a1
  

  kwen2501

  authored and

  pytorchmergebot

  committed
  d7961a1

  Copy full SHA for d7961a1

• Revert "[Inductor] Add decomposeK as an autotuning choice for mm (#150654)"

  Show description for 7c3e679
  pytorchmergebot

  committed
  7c3e679

  Copy full SHA for 7c3e679

• [invoke_subgraph] Unpacked operands (#152547)

  Show description for 4649fd1
  

  anijain2305

  authored and

  pytorchmergebot

  committed
  4649fd1

  Copy full SHA for 4649fd1

• Remove unnecessary __STDC_FORMAT_MACROS macro (#152513)

  Show description for ac5de6d
  

  FFFrog

  authored and

  pytorchmergebot

  committed
  ac5de6d

  Copy full SHA for ac5de6d

Pagination

Previous

Next