Commit Graph

116 Commits

Author SHA1 Message Date
Xiaoyu Zhang
55a7ec388f use warp shuffle style reduce and flashinfer vectorize (#3628) 2025-02-19 20:53:51 +08:00
Baizhou Zhang
67fc595bb8 [Feature] Apply Cublas Grouped Gemm kernel (#3629) 2025-02-18 15:18:31 +08:00
Xiaoyu Zhang
3efbdf68b9 fix sgl-kernel codestyle (#3563) 2025-02-14 18:05:52 +08:00
Yineng Zhang
e082142519 chore: bump 0.0.3.post6 sgl-kernel (#3555) 2025-02-14 08:55:15 +08:00
Xiaoyu Zhang
f076328bb7 fix moe_align_kernel shm init not sync bug (#3534) 2025-02-13 16:47:00 +08:00
Yineng Zhang
4430c0a513 chore: bump 0.0.3.post5 sgl-kernel (#3530) 2025-02-13 01:51:46 +08:00
yizhang2077
640363ad20 support blockwise fp8 matmul kernel (#3267) 2025-02-13 01:49:33 +08:00
Yineng Zhang
b96e92e6e6 chore: bump 0.0.3.post4 sgl-kernel (#3523) 2025-02-12 17:28:36 +08:00
Xiaoyu Zhang
bb418ced80 optimize per token group quant fp8 (#3490) 2025-02-11 22:19:05 +08:00
Yineng Zhang
6239d0b2e7 chore: bump sgl-kernel v0.0.3.post3 (#3440) 2025-02-10 04:00:52 +08:00
Yineng Zhang
4cfd3add6d support version in sgl-kernel (#3439) 2025-02-10 03:49:52 +08:00
Yineng Zhang
29daf498cd fix cu118 link issue (#3421) 2025-02-09 18:16:44 +08:00
Yineng Zhang
f9905d59a8 support speculative decoding kernel in sgl-kernel (#3373)
Co-authored-by: Ying Sheng <sqy1415@gmail.com>
2025-02-07 20:29:51 +08:00
Yineng Zhang
45c87e083f fix undefined symbol cudaGetDriverEntryPointByVersion (#3372) 2025-02-07 19:32:45 +08:00
Xiaoyu Zhang
cdae77b03d optimize moe_align_kernel cuda (#3347) 2025-02-07 00:53:46 +08:00
Yineng Zhang
adeee15204 fix sgl-kernel build failure on AMD (#3352) 2025-02-07 00:35:59 +08:00
Xiaoyu Zhang
ad3499858e clean moe align block kernel code and add acc test (#3332) 2025-02-06 16:42:36 +08:00
HAI
2c1a695ff1 ROCm: sgl-kernel enablement starting with sgl_moe_align_block (#3287) 2025-02-04 21:44:44 +08:00
Yineng Zhang
00fa7d0417 add copyright for sgl-kernel (#3270) 2025-02-03 21:34:44 +08:00
Yineng Zhang
7876279ea7 update cutlass dependency (#3240) 2025-02-01 03:13:44 +08:00
Yineng Zhang
3ee62235c6 revert the MoE dependence (#3230) 2025-01-31 16:51:41 +08:00
Yineng Zhang
9602c2aac7 keep the parts needed for moe_kernels (#3218) 2025-01-31 00:39:47 +08:00
Yineng Zhang
e81d7f11de add tensorrt_llm moe_gemm as 3rdparty (#3217) 2025-01-30 23:49:14 +08:00
Yineng Zhang
222ce6f1da add tensorrt_llm common and cutlass_extensions as 3rdparty (#3216)
Co-authored-by: BBuf <35585791+BBuf@users.noreply.github.com>
2025-01-30 23:04:41 +08:00
Yineng Zhang
468d23cff9 update setup for sgl-kernel (#3214) 2025-01-30 19:47:50 +08:00
Yineng Zhang
c38b5fb4f4 update 3rdparty and rms norm for sgl-kernel (#3213) 2025-01-30 19:32:21 +08:00
Xiaoyu Zhang
81262c7b72 clean up useless file (#3192) 2025-01-28 14:29:30 +08:00
Yineng Zhang
8a96f74988 chore: bump 0.0.3 for sgl-kernel (#3178)
Co-authored-by: ispobock <ispobaoke@hotmail.com>
Co-authored-by: BBuf <35585791+BBuf@users.noreply.github.com>
Co-authored-by: HandH1998 <007aabbcc411@gmail.com>
Co-authored-by: yizhang2077 <1109276519@qq.com>
Co-authored-by: ByronHsu <byronhsu1230@gmail.com>
2025-01-27 20:29:28 +08:00
Yineng Zhang
827aa8730b cleanup sgl-kernel kernels (#3175) 2025-01-27 19:11:01 +08:00
Lianmin Zheng
53cef81587 Improve weight loading and code style (#3174) 2025-01-27 03:00:41 -08:00
Byron Hsu
514f37c32b [kernel] Fix position ids in rope (#3173) 2025-01-27 17:09:51 +08:00
Byron Hsu
741fccd7bf Bump sgl kernel to 0.0.2.post19 (#3167) 2025-01-27 15:36:07 +08:00
Byron Hsu
fb11a43981 [kernel] Integrate flashinfer's rope with higher precision and better perf (#3134) 2025-01-27 15:28:00 +08:00
Yineng Zhang
f265d15b96 use self-hosted to build sgl-kernel (#3154) 2025-01-26 23:02:57 +08:00
Yineng Zhang
02431b9ad2 fix link in README (#3153) 2025-01-26 21:30:00 +08:00
Yineng Zhang
318260c0fa chore: bump 0.0.2.post18 for sgl-kernel (#3149) 2025-01-26 19:00:34 +08:00
HandH1998
82392da830 support w8a8 fp8 kernel with CUTLASS (#3047)
Co-authored-by: yych0745 <1398089567@qq.com>
2025-01-26 15:46:51 +08:00
Yineng Zhang
95f789adb0 minor: cleanup sgl-kernel (#3143) 2025-01-26 14:29:58 +08:00
yinfan98
9286740eff feat: refactor sgl-kernel and use TORCH_LIBRARY instead of PYBIND11_MODULE for custom ops (#3130)
Co-authored-by: yinfan.1024 <yinfan.1024@bytedance.com>
Co-authored-by: yinfan98 <1106110035@qq.com>
Co-authored-by: Yineng Zhang <me@zhyncs.com>
2025-01-26 02:55:08 +08:00
Yineng Zhang
896c07441e update installation doc for sgl-kernel (#3129) 2025-01-26 00:00:13 +08:00
Yineng Zhang
14e754a868 chore: bump v0.0.2.post17 for sgl-kernel (#3125) 2025-01-25 20:43:02 +08:00
yizhang2077
98522149ff mirror fix for custom allreduce (#3124) 2025-01-25 18:26:41 +08:00
Xiaoyu Zhang
5d9d15e70f support fp32 in sampling_scaling_penalties kernel (#3121) 2025-01-25 16:52:17 +08:00
Ke Bao
a22f60a313 Add workflow for sgl-kernel cu118 release (#3109) 2025-01-24 22:30:30 +08:00
Yineng Zhang
04f0b4cbef minor: update sgl-kernel setup (#3107) 2025-01-24 20:10:35 +08:00
Trevor Morris
685a5738a7 Allow local cutlass directory to be used in sgl-kernel build (#3037) 2025-01-24 03:59:47 -08:00
Yineng Zhang
153b414e83 minor: sync flashinfer and add turbomind as 3rdparty (#3105) 2025-01-24 19:22:39 +08:00
Ke Bao
6619f48e18 Fix cu118 group gemm compile issue (#3097) 2025-01-24 15:19:09 +08:00
Ke Bao
7bad7e75bf Add shapes for int8 gemm benchmark (#3093) 2025-01-24 12:27:30 +08:00
Yineng Zhang
54bac8af0b chore: bump sgl-kernel 0.0.2.post16 (#3087) 2025-01-24 01:57:48 +08:00