EngineX-Hygon/sglang
5
0
Fork 0
Code Issues Pull Requests Actions 7 Projects Releases Wiki Activity

[Refactor] Rename n_share_experts_fusion as num_fused_shared_experts (#6735)

Browse Source
This commit is contained in:
Cheng Wan
2025-06-03 17:48:24 -07:00
committed by GitHub
parent b6d0ce9f78
commit 8a5480528d
14 changed files with 82 additions and 93 deletions
Download Patch File Download Diff File Expand all files Collapse all files

11
benchmark/kernels/fused_moe_triton/README.md
View File

@@ -27,19 +27,17 @@ python benchmark/kernels/fused_moe_triton/tuning_fused_moe_triton.py \
--dtype fp8_w8a8 \
--tune
# Tune DeepSeek-V3 with FP8, TP=8 and n_share_experts_fusion=8
# Tune DeepSeek-V3 with FP8 and TP=8
python benchmark/kernels/fused_moe_triton/tuning_fused_moe_triton.py \
--model deepseek-ai/DeepSeek-V3-0324 \
--tp-size 8 \
--n-share-experts-fusion 8 \
--dtype fp8_w8a8 \
--tune
# Tune DeepSeek-R1 with channel-wise INT8, TP=16 and n_share_experts_fusion=16
# Tune DeepSeek-R1 with channel-wise INT8 and TP=16
python benchmark/kernels/fused_moe_triton/tuning_fused_moe_triton.py \
--model meituan/DeepSeek-R1-Channel-INT8 \
--tp-size 16 \
--n-share-experts-fusion 16 \
--dtype int8_w8a8 \
--tune
```
@@ -65,11 +63,10 @@ python benchmark/kernels/fused_moe_triton/benchmark_vllm_vs_sglang_fused_moe_tri
--model deepseek-ai/DeepSeek-V3-0324 \
--tp-size 8
# Compare with custom TP size and n_share_experts_fusion
# Compare with custom TP size
python benchmark/kernels/fused_moe_triton/benchmark_vllm_vs_sglang_fused_moe_triton.py \
--model deepseek-ai/DeepSeek-V3-0324 \
--tp-size 8 \
--n-share-experts-fusion 8
--tp-size 8
```
The benchmark results will be saved as plots and data files in the specified output directory (default: `./configs/benchmark_ops/vllm_sglang_fused_moe/`).