Commits

Branch selector

 main

User selector

All users

Datepicker

All time

Commit History

Commits on Apr 1, 2025

• Add one_shot_all_reduce_copy to allow non-symm-mem allocated tensors to be reduced (#150129)

  Show description for 1700599
  

  ngimel

  authored and

  pytorchmergebot

  committed
  1700599

  Copy full SHA for 1700599

• enable out variant of 2-shot reduction (#150153)

  Show description for 414b9ae
  

  ngimel

  authored and

  pytorchmergebot

  committed
  414b9ae

  Copy full SHA for 414b9ae

• Suppress more warnings (#149833)

  Show description for 7e7e569
  

  tugsbayasgalan

  authored and

  pytorchmergebot

  committed
  7e7e569

  Copy full SHA for 7e7e569

• [dynamo] add error message for unsupported LOAD_BUILD_CLASS (#150323)

  Show description for 790d459
  

  williamwen42

  authored and

  pytorchmergebot

  committed
  790d459

  Copy full SHA for 790d459

• [Doc] Update CMAKE_PREFIX_PATH for XPU windows README (#148863)

  Show description for ce52674
  

  2 peopleauthored and

  pytorchmergebot

  committed
  ce52674

  Copy full SHA for ce52674

• [fr] Added protection against missing stack frames in fr cont. (#150133)

  Show description for 31634b8
  

  VieEeEw

  authored and

  pytorchmergebot

  committed
  31634b8

  Copy full SHA for 31634b8

• [CI] Skip test_copy_large_tensor on M2-15 runners (#150377)

  Show description for 827b730
  

  malfet

  authored and

  pytorchmergebot

  committed
  827b730

  Copy full SHA for 827b730

• `torch.backends.mkldnn.flags()` CM should not warn (#150358)

  Show description for 6470b37
  

  malfet

  authored and

  pytorchmergebot

  committed
  6470b37

  Copy full SHA for 6470b37

• [Inductor] optimize the heuristics of parallel reduction (#149614)

  Show description for 5cb5675
  

  jiayisunx

  authored and

  pytorchmergebot

  committed
  5cb5675

  Copy full SHA for 5cb5675

• [Intel gpu] always set deterministic for xpu accuracy test (#149028)

  Show description for 0f12951
  

  2 peopleauthored and

  pytorchmergebot

  committed
  0f12951

  Copy full SHA for 0f12951

• [BE] Get rid of cross-compile and x86 build options for Mac (#150362)

  Show description for 7ab8532
  

  malfet

  authored and

  pytorchmergebot

  committed
  7ab8532

  Copy full SHA for 7ab8532

• Add a warning when a tensor with requires_grad=True is converted to a scalar (#143261)

  Show description for 4ce0b95
  

  3 peopleauthored and

  pytorchmergebot

  committed
  4ce0b95

  Copy full SHA for 4ce0b95

• [ROCm] Enable more inductor UTs (#149513)

  Show description for 49b7d0d
  

  jataylo

  authored and

  pytorchmergebot

  committed
  49b7d0d

  Copy full SHA for 49b7d0d

Commits on Mar 31, 2025

• Fix typo (#150363)

  Show description for c75dac5
  

  malfet

  authored and

  pytorchmergebot

  committed
  c75dac5

  Copy full SHA for c75dac5

• [MPS] Add support for hermite_polynomial_h. (#150279)

  Show description for b48505a
  

  3 peopleauthored and

  pytorchmergebot

  committed
  b48505a

  Copy full SHA for b48505a

• [AOTInductor] Free tensors in test (#150274)

  Show description for a2070e2
  

  muchulee8

  authored and

  pytorchmergebot

  committed
  a2070e2

  Copy full SHA for a2070e2

• [cachinghostallocator] remove the check on cudaHostRegister path (#150070)

  Show description for 982a7f7
  

  842974287

  authored and

  pytorchmergebot

  committed
  982a7f7

  Copy full SHA for 982a7f7

• Merge Triton ScaledMM as epilogue to MM template (#150045)

  Show description for 9810488
  

  PaulZhang12

  authored and

  pytorchmergebot

  committed
  9810488

  Copy full SHA for 9810488

• Update gloo submodule (#150320)

  Show description for 91666ee
  

  malfet

  authored and

  pytorchmergebot

  committed
  91666ee

  Copy full SHA for 91666ee

• Revert "Add a warning when a tensor with requires_grad=True is converted to a scalar (#143261)"

  Show description for 1526ff9
  pytorchmergebot

  committed
  1526ff9

  Copy full SHA for 1526ff9

• [ROCm] cmake 4 workaround for hiprtc (#150324)

  Show description for 423e4a4
  

  amdfaa

  authored and

  pytorchmergebot

  committed
  423e4a4

  Copy full SHA for 423e4a4

• [ROCm][CI] Increase wheel build timeout from 210 to 240 (#150221)

  Show description for 4e2997d
  

  ethanwee1

  authored and

  pytorchmergebot

  committed
  4e2997d

  Copy full SHA for 4e2997d

• [export] min/max ranges for dim hints (#149590)

  Show description for 925fd4a
  

  pianpwk

  authored and

  pytorchmergebot

  committed
  925fd4a

  Copy full SHA for 925fd4a

• cd: Fix naming for windows arm64 libtorch builds (#150310)

  Show description for dfcd98e
  

  seemethere

  authored and

  pytorchmergebot

  committed
  dfcd98e

  Copy full SHA for dfcd98e

• Adjust TestInductorOpInfo to depend on backend, not device (#146911)

  Show description for 80b7f6b
  

  matthewhagraphcore

  authored and

  pytorchmergebot

  committed
  80b7f6b

  Copy full SHA for 80b7f6b

• Make PyTorch buildable by CMake-4.x on s390x (#150294)

  Show description for ab342d3
  

  AlekseiNikiforovIBM

  authored and

  pytorchmergebot

  committed
  ab342d3

  Copy full SHA for ab342d3

• [invoke_subgraph] Support unbacked (#149298)

  Show description for 5e34758
  

  angelayi

  authored and

  pytorchmergebot

  committed
  5e34758

  Copy full SHA for 5e34758

• [dynamic shapes] C++ bindings for guard_or_false/true (#150148)

  Show description for 284b766
  

  pianpwk

  authored and

  pytorchmergebot

  committed
  284b766

  Copy full SHA for 284b766

• [ROCm] Enable several fsdp related UTs (#149369)

  Show description for 47cdad2
  

  pragupta

  authored and

  pytorchmergebot

  committed
  47cdad2

  Copy full SHA for 47cdad2

• Revert "Enable TMA persistent GEMM Template by default (#149427)"

  Show description for 7c85806
  pytorchmergebot

  committed
  7c85806

  Copy full SHA for 7c85806

• Revert "enable out variant of 2-shot reduction (#150153)"

  Show description for 57fa99c
  pytorchmergebot

  committed
  57fa99c

  Copy full SHA for 57fa99c

• Revert "Add one_shot_all_reduce_copy to allow non-symm-mem allocated tensors to be reduced (#150129)"

  Show description for e57fa18
  pytorchmergebot

  committed
  e57fa18

  Copy full SHA for e57fa18

• Update torch-xpu-ops commit pin to 3ee2bd2 (#150300)

  Show description for f74d5d5
  

  chuanqi129

  authored and

  pytorchmergebot

  committed
  f74d5d5

  Copy full SHA for f74d5d5

• Unify use of `enableCollectiveHashDebug_` and trivial updates (#142865)

  Show description for bbb9b24
  

  oraluben

  authored and

  pytorchmergebot

  committed
  bbb9b24

  Copy full SHA for bbb9b24

• [ROCm] use correct workspace for hipblaslt, silence warning (#150227)

  Show description for c158eac
  

  2 peopleauthored and

  pytorchmergebot

  committed
  c158eac

  Copy full SHA for c158eac

Pagination

Previous

Next