Commits

Branch selector

 develop-upstream

User selector

All users

All time

Commit History

Commits on May 21, 2026

• Merge pull request #3202 from ROCm/develop-upstream-sync-20260505

  Show description for 3b5928f
  hsharsha

  authored
  3b5928f

  Copy full SHA for 3b5928f

Commits on May 19, 2026

• Remove //tensorflow/core/kernels/batching_util:shared_batch_scheduler_test as it timesout on CI

  hsharsha

  committed
  a090d58

  Copy full SHA for a090d58

Commits on May 18, 2026

• Remove .claude directory

  hsharsha

  committed
  0d8d798

  Copy full SHA for 0d8d798

Commits on May 15, 2026

• Guard explicit rocm deps under if_rocm in BUILD files for cpu builds

  hsharsha

  committed
  360f6b0

  Copy full SHA for 360f6b0

Commits on May 14, 2026

• Use ROCm 7.2.1 images

  

  Harsha Havanur Shamsundara

  committed
  67b382c

  Copy full SHA for 67b382c

Commits on May 13, 2026

• Remove cuda-only asyncmemalloc test and enable conv test

  hsharsha

  committed
  de0de69

  Copy full SHA for de0de69

• Add passing xla tests

  

  Harsha Havanur Shamsundara

  committed
  abb5949

  Copy full SHA for abb5949

• Fix test build break in xla

  

  Harsha Havanur Shamsundara

  committed
  ccdfac4

  Copy full SHA for ccdfac4

Commits on May 12, 2026

• Build: Bump ubuntu 22.04 base image

  Show description for a0db567
  

  Harsha Havanur Shamsundara

  committed
  a0db567

  Copy full SHA for a0db567

• Build: Drop support for Python 3.9.

  Show description for b30c0aa
  

  Harsha Havanur Shamsundara

  committed
  b30c0aa

  Copy full SHA for b30c0aa

• Update bazelisk, buildifier, buildozer and bats-core to the latest version in ml_build Dockerfile.rocm

  

  Harsha Havanur Shamsundara

  committed
  7829472

  Copy full SHA for 7829472

Commits on May 11, 2026

• Use explicit deps

  alekstheod

  committed
  b09217d

  Copy full SHA for b09217d

• alwayslink=True for kernel libs

  

  Harsha Havanur Shamsundara

  committed
  1eeaa42

  Copy full SHA for 1eeaa42

Commits on May 5, 2026

• Fix build errors

  

  Harsha Havanur Shamsundara

  committed
  2165ed8

  Copy full SHA for 2165ed8

• Resolve merge conflicts for sync_20260505

  

  Harsha Havanur Shamsundara

  committed
  39e4aa9

  Copy full SHA for 39e4aa9

• Merge remote-tracking branch 'upstream/master' into develop-upstream-sync-20260505

  

  Harsha Havanur Shamsundara

  committed
  c2045f6

  Copy full SHA for c2045f6

• [XLA:GPU] Fix accumulator type in GemmRewriter dot rewrite.

  Show description for 7598d09
  

  mooskagh

  authored and

  tensorflower-gardener

  committed
  7598d09

  Copy full SHA for 7598d09

• PR #41960: [ROCm] Use hermetic rpath for xla tests under rocm

  Show description for d102faa
  

  alekstheod

  authored and

  tensorflower-gardener

  committed
  d102faa

  Copy full SHA for d102faa

• [XLA:GPU] Emit copy in input parameter directly used by multi-host mosaic custom call.

  Show description for 4918381
  

  PatriosTheGreat

  authored and

  tensorflower-gardener

  committed
  4918381

  Copy full SHA for 4918381

• [XLA:GPU] Add missing dependency in OSS.

  Show description for 0b7e7d4
  

  PatriosTheGreat

  authored and

  tensorflower-gardener

  committed
  0b7e7d4

  Copy full SHA for 0b7e7d4

• Explicitly list v100 and h100 in CUB scratch data generation

  Show description for 3e0453e
  

  EusebioDM

  authored and

  tensorflower-gardener

  committed
  3e0453e

  Copy full SHA for 3e0453e

• Rollforward DotHandler sharding conversions with fixes.

  Show description for 8d324e8
  

  KanishAnand

  authored and

  tensorflower-gardener

  committed
  8d324e8

  Copy full SHA for 8d324e8

• [XLA:GPU] Reland: Color multi-host mosaic custom calls with collective memory space.

  Show description for c1c08b7
  

  PatriosTheGreat

  authored and

  tensorflower-gardener

  committed
  c1c08b7

  Copy full SHA for c1c08b7

• [Convolutions] Improve CuDNN Fusion Compiler logs to track down failing CuDNN fusions.

  Show description for c8641aa
  

  derdrdirk

  authored and

  tensorflower-gardener

  committed
  c8641aa

  Copy full SHA for c8641aa

• Remove 8-bit value types in CUB scratch space data generation

  Show description for c7b453d
  

  EusebioDM

  authored and

  tensorflower-gardener

  committed
  c7b453d

  Copy full SHA for c7b453d

• [XLA:GPU] Emit scaled dot in two stages

  Show description for 2f09e88
  

  vickyliu-go4it

  authored and

  tensorflower-gardener

  committed
  2f09e88

  Copy full SHA for 2f09e88

• Automated Code Change

  Show description for 72c64e2
  tensorflower-gardener

  committed
  72c64e2

  Copy full SHA for 72c64e2

• Automated Code Change

  Show description for 46869d1
  tensorflower-gardener

  committed
  46869d1

  Copy full SHA for 46869d1

• Integrate Triton up to 886365ca82b03edef597b0d2c8c32be62ac2546b

  Show description for 5c88c44
  tensorflower-gardener

  committed
  5c88c44

  Copy full SHA for 5c88c44

• Introduce DevicelessEstimateCubSortScratchSize HLO Pass

  Show description for c301062
  

  EusebioDM

  authored and

  tensorflower-gardener

  committed
  c301062

  Copy full SHA for c301062

• Add output_sym_offset to RaggedAllToAll WithSymmetricMemory kernel.

  Show description for 779726a
  

  apivovarov

  authored and

  tensorflower-gardener

  committed
  779726a

  Copy full SHA for 779726a

• Automated Code Change

  Show description for 379706d
  tensorflower-gardener

  committed
  379706d

  Copy full SHA for 379706d

• Automated Code Change

  Show description for 6f7fc44
  tensorflower-gardener

  committed
  6f7fc44

  Copy full SHA for 6f7fc44

• Automated Code Change

  Show description for 40eb4e4
  tensorflower-gardener

  committed
  40eb4e4

  Copy full SHA for 40eb4e4

• Automated Code Change

  Show description for d23e4be
  tensorflower-gardener

  committed
  d23e4be

  Copy full SHA for d23e4be