[release/2.9] Cherry-picks from release/2.8 10/06/2025#2701

Merged

jithunnair-amd merged 24 commits intorelease/2.9from

release_2.9_IFU_20251006

Oct 10, 2025

Commits on Oct 6, 2025

[release/2.8] Enable wheels

jithunnair-amd
authored and
pragupta
committed
[release/2.8] Upgrade numpy versions; Use different package versions for py3.9;

authored and
pragupta
committed

Commits on Oct 8, 2025

Use ROCm/triton and update triton.txt

jithunnair-amd
authored and
pragupta
committed
Add related_commits file (#2396)
pragupta
committed
Add QA automation scripts for running PyTorch unit tests

jithunnair-amd
authored and
pragupta
committed
[AUTOGENERATED] [release/2.5] [ROCm][layer_norm] Use __builtin_amdgcn_rcpf(x) instead of 1.f/x (#1800)

authored and
pragupta
committed
[release/2.7] Update test_binary_ufuncs.py after numpy upgrade (#2289)

ethanwee1
authored and
pragupta
committed
Clean up CUDA state between tests (#2335)

rraminen
authored and
pragupta
committed
[AUTOGENERATED] [release/2.7] [release/2.6] Fix dtype before comparing torch and numpy tensors (#2362)

authored and
pragupta
committed
[release/2.8] enable py3.13 (#2366)

ethanwee1
authored and
pragupta
committed
[SWDEV-539076] Initial naive foreach autotune support (#2377)

jataylo
authored and
pragupta
committed
[release/2.7] [SWDEV-543214] Reland #2416 Fix warps runtime (#2421)

authored and
pragupta
committed
[AUTOGENERATED] [release/2.8] [release/2.7] [SWDEV-543214] Reland #2416 Fix warps runtime part 2 (#2455)

authored and
pragupta
committed
[AUTOGENERATED] [release/2.8] [SWDEV-539215] - Autotune support for persistent reduction and no_x_dim removal (#2454)

authored and
pragupta
committed
[SWDEV-539119] [release/2.8] Add fast_tanh support (#2484)

jataylo
authored and
pragupta
committed
[AUTOGENERATED] [release/2.8] Change triton package name depending on rocm version (#2529)

authored and
pragupta
committed
[release/2.8] Define uint32 t when ROCM_VERSION >= 70000 (#2513)

rraminen
authored and
pragupta
committed
[AUTOGENERATED] [release/2.8] [ROCm] OffsetCalc Unroll Optimization (#2598)

authored and
pragupta
committed
Bug fix and optimisation for persistent reduction kernel tuning (#2596)

jataylo
authored and
pragupta
committed
[ROCm] Fix indexing_backward_kernel perf (#2667)

jerrymannil
authored and
pragupta
committed
[ROCm] Improve perf for elementwise broadcast with mixed dtype (#2672)

jerrymannil
authored and
pragupta
committed
[ROCm] Implement float32 copy kernel (#2683)

jerrymannil
authored and
pragupta
committed

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[release/2.9] Cherry-picks from release/2.8 10/06/2025#2701

[release/2.9] Cherry-picks from release/2.8 10/06/2025#2701
jithunnair-amd merged 24 commits intorelease/2.9from
release_2.9_IFU_20251006

Commits on Oct 6, 2025

[release/2.8] Enable wheels

[release/2.8] Upgrade numpy versions; Use different package versions for py3.9;

Commits on Oct 8, 2025

Use ROCm/triton and update triton.txt

Add related_commits file (#2396)

Add QA automation scripts for running PyTorch unit tests

[AUTOGENERATED] [release/2.5] [ROCm][layer_norm] Use __builtin_amdgcn_rcpf(x) instead of 1.f/x (#1800)

[release/2.7] Update test_binary_ufuncs.py after numpy upgrade (#2289)

Clean up CUDA state between tests (#2335)

[AUTOGENERATED] [release/2.7] [release/2.6] Fix dtype before comparing torch and numpy tensors (#2362)

[release/2.8] enable py3.13 (#2366)

[SWDEV-539076] Initial naive foreach autotune support (#2377)

[release/2.7] [SWDEV-543214] Reland #2416 Fix warps runtime (#2421)

[AUTOGENERATED] [release/2.8] [release/2.7] [SWDEV-543214] Reland #2416 Fix warps runtime part 2 (#2455)

[AUTOGENERATED] [release/2.8] [SWDEV-539215] - Autotune support for persistent reduction and no_x_dim removal (#2454)

[SWDEV-539119] [release/2.8] Add fast_tanh support (#2484)

[AUTOGENERATED] [release/2.8] Change triton package name depending on rocm version (#2529)

[release/2.8] Define uint32 t when ROCM_VERSION >= 70000 (#2513)

[AUTOGENERATED] [release/2.8] [ROCm] OffsetCalc Unroll Optimization (#2598)

Bug fix and optimisation for persistent reduction kernel tuning (#2596)

[ROCm] Fix indexing_backward_kernel perf (#2667)

[ROCm] Improve perf for elementwise broadcast with mixed dtype (#2672)

[ROCm] Implement float32 copy kernel (#2683)

Commits on Oct 9, 2025

Bump triton to 3.5.x

Update fbgemm submodule to avoid ck errors

[release/2.9] Cherry-picks from release/2.8 10/06/2025#2701jithunnair-amd merged 24 commits intorelease/2.9ROCm/pytorch:release/2.9from release_2.9_IFU_20251006ROCm/pytorch:release_2.9_IFU_20251006Copy head branch name to clipboard

Commits

Commits on Oct 6, 2025

Commits on Oct 8, 2025

Commits on Oct 9, 2025

[release/2.9] Cherry-picks from release/2.8 10/06/2025#2701
jithunnair-amd merged 24 commits intorelease/2.9from
release_2.9_IFU_20251006