minke.yu/sglang
1
0
Fork 0
Code Issues Pull Requests Actions 12 Packages Projects Releases Wiki Activity

[Kernel] Migrate scattered quantization kernels to sglang.kernels (RFC #29630, Phase 2.5, 1/7) (#30784)

Browse Source 
Co-authored-by: Claude Fable 5 <noreply@anthropic.com>
This commit is contained in:
Xiaoyu Zhang
2026-07-13 16:17:01 +08:00
committed by GitHub
co-authored by Claude Fable 5
parent 2225817424
commit 874fc07d9b
252 changed files with 294 additions and 214 deletions
Download Patch File Download Diff File Expand all files Collapse all files
benchmark/kernels/flashinfer_allreduce_fusion/benchmark_fused_collective.py
+2 -2
View File
@@ -28,6 +28,8 @@ from typing import Optional
import torch # type: ignore
import torch.distributed as dist # type: ignore
from sglang.kernels.ops.quantization.fp8_kernel import fp8_dtype as SGLANG_FP8_DTYPE
from sglang.kernels.ops.quantization.fp8_kernel import static_quant_fp8
from sglang.srt.distributed import get_tp_group, tensor_model_parallel_all_reduce
from sglang.srt.distributed.parallel_state import (
cleanup_dist_env_and_memory,
@@ -36,8 +38,6 @@ from sglang.srt.distributed.parallel_state import (
initialize_model_parallel,
)
from sglang.srt.layers.layernorm import RMSNorm # noqa
from sglang.srt.layers.quantization.fp8_kernel import fp8_dtype as SGLANG_FP8_DTYPE
from sglang.srt.layers.quantization.fp8_kernel import static_quant_fp8
try:
from sgl_kernel import fused_add_rmsnorm as SGL_FUSED_ADD_RMS_NORM