remove sglang benchmark from triton-benchmarks

Author: leonling-ll

.github/workflows/triton-benchmarks.yml

@@ -258,6 +258,18 @@ jobs:
           python ../../scripts/build_report.py $REPORTS/attn-performance.csv $REPORTS/attn-triton-report.csv --benchmark attn --compiler triton --param_cols "Z,H,N_CTX,D_HEAD,CAUSAL" --tflops_col Triton-TFlops --hbm_col "Triton-GB/s" --tag $TAG
           python ../../scripts/build_report.py $REPORTS/attn-performance.csv $REPORTS/attn-xetla-report.csv --benchmark attn --compiler xetla --param_cols "Z,H,N_CTX,D_HEAD,CAUSAL" --tflops_col XeTLA-TFlops --hbm_col "XeTLA-GB/s" --tag $TAG
 
+      - name: Run Triton FA fwd kernel benchmark - advanced path
+        if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'flash_attention_benchmark.py_advanced')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'flash_attention_benchmark.py_advanced') }}
+        run: |
+          cd benchmarks/triton_kernels_benchmark
+          TRITON_INTEL_ADVANCED_PATH=1 \
+          IGC_VISAOptions=" -enableBCR" \
+          python flash_attention_benchmark.py --reports $REPORTS --n_runs $N_RUNS
+
+          TAG="${TAG}-adv"
+          source ../../scripts/capture-hw-details.sh
+          python ../../scripts/build_report.py $REPORTS/attn-performance.csv $REPORTS/attn-triton-advanced-report.csv --benchmark attn --compiler triton --param_cols "Z,H,N_CTX,D_HEAD,CAUSAL" --tflops_col Triton-TFlops --hbm_col "Triton-GB/s" --tag $TAG
+
       - name: Run Triton FA bwd kernel benchmark
         if: ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'flash_attention_bwd_benchmark.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'flash_attention_bwd_benchmark.py') }}
         run: |

python/src/ir.cc

@@ -1782,7 +1782,7 @@ void init_triton_ir(py::module &&m) {
                context->disableMultithreading();
                auto printingFlags = OpPrintingFlags();
                printingFlags.elideLargeElementsAttrs(16);
-               //  printingFlags.enableDebugInfo();
+               printingFlags.enableDebugInfo();
                auto printAlways = [funcToDump](Pass *, Operation *op) -> bool {
                  if (funcToDump.empty())
                    return true;