[CI][sglang] Add benchmark from sglang

.github/workflows/third-party-benchmarks.yml

@@ -15,10 +15,6 @@ on:
         description: JSON list of benchmarks to run. Leave empty to run all benchmarks.
         type: string
         default: ""
-      use_pyenv_python:
-        description: Use Python built with pyenv
-        type: boolean
-        default: false
   schedule:
     # About midnight PST (UTC-8)
     - cron: "5 10 * * *"
@@ -50,14 +46,7 @@ jobs:
       - name: Checkout repository
         uses: actions/checkout@v5
 
-      - name: Install Python
-        if: ${{ !(inputs.use_pyenv_python || false) }}
-        uses: actions/setup-python@v6
-        with:
-          python-version: ${{ env.PYTHON_VERSION }}
-
       - name: Install Python (from pyenv) ${{ inputs.python_version }}
-        if: ${{ inputs.use_pyenv_python }}
         uses: ./.github/actions/setup-pyenv-python
         with:
           python-version: ${{ env.PYTHON_VERSION }}
@@ -77,22 +66,40 @@ jobs:
       - name: Setup Triton
         uses: ./.github/actions/setup-triton
 
-      - name: Install benchmark dependencies
-        id: install
-        run: |
-          pip install transformers pandas pytest
-
       - name: Create reports dir
         run: |
           mkdir reports
           echo "REPORTS=$PWD/reports" >> $GITHUB_ENV
 
+      - name: Install benchmark dependencies
+        id: install
+        run: |
+          pip install transformers pandas pytest
+
       - name: Install benchmarks
         id: install-benchmarks
         run: |
           cd benchmarks
           pip install .
 
+      - name: Run sglang benchmark int8
+        if: ${{ steps.install-benchmarks.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'sglang')) }}
+        run: |
+          source ./scripts/capture-hw-details.sh
+
+          cd benchmarks/third_party/sglang
+          python scaled_mm_benchmark.py --reports $REPORTS
+          python ../vllm/transform_results.py $REPORTS/scaled_mm_benchmark.csv $REPORTS/scaled-mm-int8-report.csv --tag $TAG --benchmark scaled-mm-int8 --param_cols="M,N,K"
+
+      - name: Run sglang benchmark with fp8
+        if: ${{ steps.install-benchmarks.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'sglang')) }}
+        run: |
+          source ./scripts/capture-hw-details.sh
+
+          cd benchmarks/third_party/sglang
+          FP8="1" python scaled_mm_benchmark.py --reports $REPORTS
+          python ../vllm/transform_results.py $REPORTS/scaled_mm_benchmark.csv $REPORTS/scaled-mm-fp8-report.csv --tag $TAG --benchmark scaled-mm-fp8 --param_cols="M,N,K"
+
       - name: Run vllm benchmarks bf16
         if: ${{ steps.install-benchmarks.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'vllm')) }}
         run: |
@@ -103,7 +110,8 @@ jobs:
 
           cd benchmarks/third_party/vllm
           python batched_moe_benchmark.py --reports $REPORTS
-          python transform_results.py $REPORTS/moe-gemm-performance.csv $REPORTS/moe-gemm-report.csv --tag $TAG --benchmark moe-bf16-benchmark
+          python transform_results.py $REPORTS/moe-gemm-performance.csv $REPORTS/moe-gemm-report.csv --tag $TAG --benchmark moe-bf16-benchmark --param_cols="num_experts,max_tokens_per_expert,K,N"
+
 
       - name: Run vllm benchmarks fp8
         if: ${{ steps.install-benchmarks.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'vllm')) }}
@@ -112,7 +120,8 @@ jobs:
 
           cd benchmarks/third_party/vllm
           FP8="1" python batched_moe_benchmark.py --reports $REPORTS
-          python transform_results.py $REPORTS/moe-gemm-performance.csv $REPORTS/moe-gemm-fp8-report.csv --tag $TAG --benchmark moe-fp8-benchmark
+          python transform_results.py $REPORTS/moe-gemm-performance.csv $REPORTS/moe-gemm-fp8-report.csv --tag $TAG --benchmark moe-fp8-benchmark --param_cols="num_experts,max_tokens_per_expert,K,N"
+
 
       - name: Run Liger-Kernel benchmarks
         if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'liger-kernel')) }}