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

Refactor FP4 quantization and remove deprecated JIT kernels (#30448)

Browse Source 
Co-authored-by: root <root@sgl-b300-inference.datacrunch.io>
This commit is contained in:
Brayden Zhong
2026-07-14 09:22:07 +08:00
committed by GitHub
co-authored by root
parent 0c01971eeb
commit 9756f768a6
38 changed files with 120 additions and 5705 deletions
Download Patch File Download Diff File Expand all files Collapse all files
docs_new/docs/advanced_features/quantization.mdx
+1 -6
View File
@@ -284,11 +284,6 @@ Backend selection applies to **blockwise FP8**, **MXFP8** (dense linear), and **
<td>SM80+</td>
<td>Auto-selects: <code>flashinfer_cutedsl</code> on SM100; <code>marlin</code> on SM80-SM90; <code>flashinfer_cutlass</code> otherwise (including SM120)</td>
</tr>
<tr>
<td><code>cutlass</code></td>
<td>SM100/120</td>
<td>SGLang CUTLASS kernel</td>
</tr>
<tr>
<td><code>flashinfer_cutlass</code></td>
<td>SM100/120</td>
@@ -317,7 +312,7 @@ Backend selection applies to **blockwise FP8**, **MXFP8** (dense linear), and **
</tbody>
</table>
On Blackwell, when FlashInfer is unavailable for NVFP4, the SGLang CUTLASS kernel is used as an automatic fallback. On SM80-SM90, `auto` selects Marlin for NVFP4.
On SM80-SM90, `auto` selects Marlin for NVFP4. NVFP4 GEMM requires FlashInfer to be installed.
## Offline Quantization
docs_new/docs/advanced_features/server_arguments.mdx
+2 -2
View File
@@ -1436,9 +1436,9 @@ Please consult the documentation below and [server_args.py](https://github.com/s
</tr>
<tr>
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>`--fp4-gemm-backend`</td>
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>Choose the runner backend for NVFP4 GEMM operations. Options: 'auto' (default; selects <code>flashinfer_cutedsl</code> on SM100, <code>marlin</code> on SM80-SM90, <code>flashinfer_cutlass</code> otherwise (including SM120)), 'cutlass' (SGLang CUTLASS kernel), 'flashinfer_cutlass' (FlashInfer CUTLASS backend), 'flashinfer_cudnn' (FlashInfer cuDNN backend, optimal on CUDA 13+ with cuDNN 9.15+), 'flashinfer_cutedsl' (FlashInfer CuTe DSL backend), 'flashinfer_trtllm' (FlashInfer TensorRT-LLM backend, requires different weight preparation with shuffling), 'marlin' (weight-only W4A16 fallback for SM80-SM90). All FlashInfer backends fall back to sgl-kernel CUTLASS when FlashInfer is unavailable.</td>
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}>Choose the runner backend for NVFP4 GEMM operations. Options: 'auto' (default; selects <code>flashinfer_cutedsl</code> on SM100, <code>marlin</code> on SM80-SM90, <code>flashinfer_cutlass</code> otherwise (including SM120)), 'flashinfer_cutlass' (FlashInfer CUTLASS backend), 'flashinfer_cudnn' (FlashInfer cuDNN backend, optimal on CUDA 13+ with cuDNN 9.15+), 'flashinfer_cutedsl' (FlashInfer CuTe DSL backend), 'flashinfer_trtllm' (FlashInfer TensorRT-LLM backend, requires different weight preparation with shuffling), 'marlin' (weight-only W4A16 fallback for SM80-SM90). Requires FlashInfer to be installed.</td>
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.02)"}}>`auto`</td>
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>auto</code>, <code>cutlass</code>, <code>flashinfer_cudnn</code>, <code>flashinfer_cutedsl</code>, <code>flashinfer_cutlass</code>, <code>flashinfer_trtllm</code>, <code>marlin</code></td>
<td style={{padding: "9px 12px", backgroundColor: "rgba(255,255,255,0.05)"}}><code>auto</code>, <code>flashinfer_cudnn</code>, <code>flashinfer_cutedsl</code>, <code>flashinfer_cutlass</code>, <code>flashinfer_trtllm</code>, <code>marlin</code></td>
</tr>
<tr>
<td style={{padding: "9px 12px", fontWeight: 500, backgroundColor: "rgba(255,255,255,0.02)"}}>`--disable-flashinfer-autotune`</td>