Jia Guo 587fd15bd2 perf: eliminate attention DtoD copy by passing pre-allocated output to FA (#21985) ...

Co-authored-by: Claude Opus 4.6 <noreply@anthropic.com>

2026-04-24 12:05:16 -07:00

..

quantization

[7/n] decouple quantization impl from vllm dependency - gguf kernel (#11019)

2025-10-11 14:04:57 -07:00

testing

[kernel slimming] Clean many useless sgl-kernel deprecated kernels (#20277)

2026-03-14 16:45:54 +08:00

__init__.py

Remove obsolete sgl-kernel legacy paths (#21528)

2026-04-01 09:00:20 +08:00

allreduce.py

[amd] Add deterministic all-reduce kernel for AMD (ROCm) (#15340)

2025-12-18 23:36:03 -08:00

attention.py

Remove obsolete sgl-kernel legacy paths (#21528)

2026-04-01 09:00:20 +08:00

cutlass_moe.py

[1/n]: add cutlass W4A8 moe kernel for hopper architecture (#7772)

2025-07-04 20:50:12 -07:00

debug_utils.py

Add SGLang CUDA crash API logging inspired by FlashInfer (#20910)

2026-03-22 16:39:40 +08:00

elementwise.py

Clean up bench_one_batch warning and simplify norm dispatch (#23110)

2026-04-17 17:42:20 -07:00

expert_specialization.py

[sgl-kernel][Feat][B200][1/N] Support MXFP8 Grouped GEMM in Blackwell (#13731)

2025-12-04 10:09:09 +08:00

flash_attn.py

perf: eliminate attention DtoD copy by passing pre-allocated output to FA (#21985)

2026-04-24 12:05:16 -07:00

flash_mla.py

Revert "Rollback flashmla to older version [1/2]" (#21922)

2026-04-02 00:27:02 -07:00

gemm.py

[Kernel] Set sgl_per_token_group_quant_8bit_v2 as default choice (#22467)

2026-04-11 01:59:57 -07:00

grammar.py

fix sgl-kernel unit tests (#5666)

2025-04-23 01:18:30 -07:00

kvcacheio.py

[HiCache]Page head layout IO kernel (#11615)

2025-10-26 15:53:50 +08:00

load_utils.py

Release sglang kernel 0.4.0 (#20440)

2026-03-16 20:34:58 +08:00

mamba.py

[CPU] Optimize Qwen3-next model on CPU (#12525)

2026-01-29 22:03:58 -08:00

memory.py

[kernel slimming] Clean many useless sgl-kernel deprecated kernels (#20277)

2026-03-14 16:45:54 +08:00

moe.py

[JIT Kernel] Reland NVFP4 kernels to JIT (#20012)

2026-03-07 10:31:08 +08:00

sampling.py

Remove obsolete sgl-kernel legacy paths (#21528)

2026-04-01 09:00:20 +08:00

scalar_type.py

refactor: Move scalar_types.py to sgl-kernel to avoid circular import (#8720)

2025-08-07 19:22:16 -07:00

sparse_flash_attn.py

[Feat] QWen-1M context support[1/2]: Update block sparse attention backend utils kernel (#5847)

2025-04-28 11:03:17 -07:00

spatial.py

Optional extension for green context (#9231)

2025-08-15 21:33:52 +08:00

speculative.py

move all get_stream in sgl_kernel to c++ to reduce the launch overhead (#12521)

2025-11-02 13:15:05 -08:00

test_utils.py

Reland [1/2] Optimizations and refactors about quant kernel (#10312)

2025-10-11 15:59:03 +08:00

top_k.py

[sgl-kernel][Deepseek V3.2] Add row_starts to topk kernel (#12582)

2025-11-07 18:33:27 -08:00

utils.py

[Misc] Use cache_once for is_arch_support_pdl in sgl-kernel (#22725)

2026-04-14 15:22:10 -07:00

version.py

chore: bump sgl-kernel version to 0.4.1 (#21447)

2026-04-02 22:31:59 -07:00