Commits

Commits on Apr 9, 2025

[ROCm] Introduce AMD specific inductor gemm tuning (#147315)

jataylo
authored and
pytorchmergebot
committed

Commits on Apr 1, 2025

[ROCm] Enable more inductor UTs (#149513)

jataylo
authored and
pytorchmergebot
committed

Commits on Mar 26, 2025

Reland "Introduce new template heuristic for triton autotune configs" (#147452)

jataylo
authored and
pytorchmergebot
committed

Commits on Mar 7, 2025

[ROCm] Incorporate ROCm triton specific tuning parameters (#148437)

jataylo
authored and
pytorchmergebot
committed

Commits on Feb 14, 2025

Introduce new template heuristic for triton autotune configs (#144985)

jataylo
authored and
pytorchmergebot
committed

Commits on Feb 11, 2025

Introduce new template heuristic for triton autotune configs (#144985)

jataylo
authored and
pytorchmergebot
committed

Commits on Feb 10, 2025

[ROCm] Unskip std:bad_alloc failures (#146407)

jataylo
authored and
pytorchmergebot
committed

Commits on Jan 29, 2025

[64-bit] Int64 casting for UpSampleNearest3D (#144865)

jataylo
authored and
pytorchmergebot
committed

Commits on Dec 24, 2024

[ROCm] Add miopen_batch_norm to meta_registrations to fix AOTI issue (#143569)

jataylo
authored and
pytorchmergebot
committed

Commits on Dec 19, 2024

[ROCm] Guard triton backend call around cuda.is_available (#143570)

jataylo
authored and
pytorchmergebot
committed

Commits on Nov 21, 2024

[ROCm] AMDSMI memory usage unification (#139900)

jataylo
authored and
pytorchmergebot
committed

Commits on Nov 15, 2024

[ROCm] Bug fix for flex attention configs avoiding ROCm path (#140270)

jataylo
authored and
pytorchmergebot
committed

Commits on Nov 11, 2024

[ROCm] Correct numerical issues in layer norm backwards kernel (#140259)

jataylo
authored and
pytorchmergebot
committed

Commits on Nov 8, 2024

[ROCm] Support new AMD triton stream pipeliner (#139881)

jataylo
authored and
pytorchmergebot
committed

Commits on Nov 7, 2024

[ROCm] Tune flex-attention and decode to num_stages=1 (#139883)

jataylo
authored and
pytorchmergebot
committed

Commits on Nov 6, 2024

[ROCm] re-enable flex attention UTs (#139632)

jataylo
authored and
pytorchmergebot
committed

Commits on Oct 15, 2024

[ROCm] Add AMDSMI support for UUID input (#129741)

jataylo
authored and
pytorchmergebot
committed

Commits on Sep 18, 2024

[ROCm] upgrade ROCm CI builds to py3.10 (#134108)

jataylo
authored and
pytorchmergebot
committed

Commits on Sep 14, 2024

[ROCm] Skip pointwise associative scan tests due to regression (#135995)

jataylo
authored and
pytorchmergebot
committed

Commits on Sep 13, 2024

[ROCm] Enable ROCm support for inductor's dynamic_rblock_scaling (#129663)

jataylo
authored and
pytorchmergebot
committed

Commits on Sep 12, 2024

[ROCm] Use ieee precision for fp32 in flex attention (#135702)

jataylo
authored and
pytorchmergebot
committed

Commits on Sep 11, 2024

[ROCm] [BUGFIX] Re-enable rocm-specific tuning parameters v2 (#133852)

jataylo
authored and
pytorchmergebot
committed

Commits on Sep 5, 2024

[ROCm] remove triton-rocm commit pin and merge pins with triton.txt (#133438)

authored and
pytorchmergebot
committed

Commits on Aug 30, 2024

[ROCm] remove triton-rocm commit pin and merge pins with triton.txt (#133438)

authored and
pytorchmergebot
committed

Commits on Aug 24, 2024

[ROCm] remove triton-rocm commit pin and merge pins with triton.txt (#133438)

jataylo
authored and
pytorchmergebot
committed

Commits on Aug 19, 2024

[ROCm] Set _HAS_PYNVML to false if amdsmi not installed (#132990)

jataylo
authored and
pytorchmergebot
committed

Commits on Aug 16, 2024

[ROCm] Fix text_export cudnn_attention UT (#133234)

jataylo
authored and
pytorchmergebot
committed

Commits on Aug 5, 2024

Scale XBLOCK in triton reduction configs to avoid hitting max grid (#128826)

authored and
pytorchmergebot
committed

Commits on Jul 19, 2024

[ROCm] Enable ROCm support for inductor's dynamic_rblock_scaling (#129663)

jataylo
authored and
pytorchmergebot
committed

Commits on Jul 17, 2024

[ROCm] Return correct AMDSMI socket_power metric (#130331)

jataylo
authored and
pytorchmergebot
committed

Commits on Jul 12, 2024

[ROCm] [BUGFIX] Re-enable rocm-specific tuning parameters (#130617)

jataylo
authored and
pytorchmergebot
committed

Commits on Jul 2, 2024

[ROCm] Update nightly triton-rocm pin to release branch (#129361)

jataylo
authored and
pytorchmergebot
committed

Commits on Jul 1, 2024

[ROCm] CUDA_VISIBLE_DEVICES fallback option for device_count (#129650)

jataylo
authored and
pytorchmergebot
committed

Commits on Jun 26, 2024

[ROCm] Use additional shard for inductor workflow to resolve timeouts (#129480)

jataylo
authored and
pytorchmergebot
committed

Commits on Jun 4, 2024

[ROCm] Fix error in torch.cuda initialisation if amdsmi is not available (#127528)

jataylo
authored and
pytorchmergebot
committed