6 people a7fda4c88b [Bugfix] Fix SM100 fp8_ds_mla cache scales (#49435) ...

Signed-off-by: Ariel Bereslavsky <arielb@getimpala.ai>
Signed-off-by: Ariel Bereslavsky <ariel@getimpala.ai>
Signed-off-by: Yongye Zhu <zyy1102000@gmail.com>
Co-authored-by: Ariel Bereslavsky <arielb@getimpala.ai>
Co-authored-by: Codex <noreply@openai.com>
Co-authored-by: Misha Goin <mgoin64@gmail.com>
Co-authored-by: Yongye Zhu <zyy1102000@gmail.com>
Co-authored-by: mergify[bot] <37929162+mergify[bot]@users.noreply.github.com>

2026-09-19 17:15:07 -07:00

..

attention

[9/n] Migrate attention and cache kernels to torch stable ABI (continued) (#43717)

2026-05-29 04:44:45 +00:00

core

[ROCm][Quark][7/N] Use MXFP4 linear kernel abstraction for emulation backend (#48949)

2026-07-31 10:07:15 -07:00

cpu

[CPU] Add CPU FP8 W8A8 linear/MoE support (#49942)

2026-09-18 17:59:42 +08:00

cutlass_extensions

New stable abi cleanup (#46656)

2026-07-03 14:02:26 +08:00

libtorch_stable

[Bugfix] Fix SM100 fp8_ds_mla cache scales (#49435)

2026-09-19 17:15:07 -07:00

moe

[CPU][BugFix] Multiple fixes to w4a8_int8 CPU MoE path (#46739)

2026-07-06 05:39:20 +00:00

quantization

[Refactor] Remove dead code quantization 2 (#52821)

2026-08-25 20:28:44 -07:00

quickreduce

Enable gfx1250 ROCm architecture (#46516)

2026-07-31 01:41:12 +00:00

rocm

[ROCm][Perf] W4A16: keep skinny GEMM zero-points packed 4-bit (#54965)

2026-09-14 17:40:25 +08:00

cache.h

[Bugfix] Sparse MLA: enable fp8_ds_mla dense prefill (#48642)

2026-07-16 22:44:04 +00:00

cuda_compat.h

[Bugfix][ROCm] Fix for warp_size uses on host (#21205)

2025-07-24 00:37:19 -07:00

cuda_utils.h

[Attention] MLA with chunked prefill (#12639)

2025-02-21 15:30:12 -08:00

cumem_allocator_compat.h

[Core][AMD] Migrate fully transparent sleep mode to ROCm platform (#12695)

2025-11-12 15:24:12 -08:00

cumem_allocator.cpp

[Test][V1] Add sleep/wake correctness regression test for hybrid GDN/… (#44972)

2026-08-02 22:02:42 +00:00

custom_all_gather_reduce_scatter.cuh

Fix MNNVL Lamport mailbox publication and cleanup (#53000)

2026-08-23 23:56:54 -07:00

custom_all_reduce.cuh

[Model] Add Kimi K3 support: model files and kernels [1/N] (#50089)

2026-07-29 14:10:58 +08:00

custom_collective_common.cuh

[Model] Add Kimi K3 support: model files and kernels [1/N] (#50089)

2026-07-29 14:10:58 +08:00

custom_quickreduce.cu

[ROCM] [Communication] Add INT3 quantization method for quickreduce (#45666)

2026-06-25 15:14:15 +00:00

dispatch_utils.h

[Performance] Fused blockwise quant RMS norm (#27883)

2025-12-07 16:38:04 +00:00

flashkda_registration.cpp

[Perf] Support internal prefill checkpoints for Mamba prefix caching, 9%~25% TTFT improvement (#52789)

2026-08-22 10:32:40 -07:00

fs_io.cpp

[KV Offload] Fix failed-load livelock by marking the lookup verdict as a miss (#49328)

2026-08-09 10:25:36 +03:00

ops.h

[Kernel] Implement CUDA kernel for ReLUSquaredActivation (relu^2) (#39058)

2026-07-12 19:18:03 -07:00

qutlass_registration.cpp

[12/n] final _C library kernel migration (#45415)

2026-06-19 06:57:26 -07:00

spinloop.cpp

[BugFix] Fix clang spinloop mwaitx include (#45532)

2026-07-28 19:00:06 +00:00

torch_bindings.cpp

[ROCm] [PyTorch] Move to stable abi since ROCm upgraded to torch 2.11 (#47128)

2026-07-02 18:34:07 +08:00

torch_utils.h

[6/n] Migrate activation kernels, gptq, gguf, non cutlass w8a8 to libtorch stable ABI (continued) (#42663)

2026-05-20 00:18:12 -07:00