@@ -244,8 +244,8 @@ jobs:
244
244
python flash_attention_benchmark.py --reports $REPORTS --n_runs $N_RUNS
245
245
246
246
source ../../scripts/capture-hw-details.sh
247
- python 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
248
- python 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
247
+ python build_report.py $REPORTS/attn-performance.csv $REPORTS/attn-triton-report.csv --benchmark flash- attn --compiler triton --param_cols "Z,H,N_CTX,D_HEAD,CAUSAL" --tflops_col Triton-TFlops --hbm_col "Triton-GB/s" --tag $TAG
248
+ python build_report.py $REPORTS/attn-performance.csv $REPORTS/attn-xetla-report.csv --benchmark flash- attn --compiler xetla --param_cols "Z,H,N_CTX,D_HEAD,CAUSAL" --tflops_col XeTLA-TFlops --hbm_col "XeTLA-GB/s" --tag $TAG
249
249
250
250
- name : Run Triton FA bwd kernel benchmark
251
251
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') }}
@@ -256,8 +256,8 @@ jobs:
256
256
mv $REPORTS/attn-performance.csv $REPORTS/attn-bwd-performance.csv
257
257
258
258
source ../../scripts/capture-hw-details.sh
259
- python build_report.py $REPORTS/attn-bwd-performance.csv $REPORTS/attn-bwd-triton-report.csv --benchmark attn-bwd --compiler triton --param_cols "Z,H,N_CTX,D_HEAD,CAUSAL" --tflops_col Triton-TFlops --hbm_col "Triton-GB/s" --tag $TAG
260
- python build_report.py $REPORTS/attn-bwd-performance.csv $REPORTS/attn-bwd-xetla-report.csv --benchmark attn-bwd --compiler xetla --param_cols "Z,H,N_CTX,D_HEAD,CAUSAL" --tflops_col XeTLA-TFlops --hbm_col "XeTLA-GB/s" --tag $TAG
259
+ python build_report.py $REPORTS/attn-bwd-performance.csv $REPORTS/attn-bwd-triton-report.csv --benchmark flash- attn-bwd --compiler triton --param_cols "Z,H,N_CTX,D_HEAD,CAUSAL" --tflops_col Triton-TFlops --hbm_col "Triton-GB/s" --tag $TAG
260
+ python build_report.py $REPORTS/attn-bwd-performance.csv $REPORTS/attn-bwd-xetla-report.csv --benchmark flash- attn-bwd --compiler xetla --param_cols "Z,H,N_CTX,D_HEAD,CAUSAL" --tflops_col XeTLA-TFlops --hbm_col "XeTLA-GB/s" --tag $TAG
261
261
262
262
- name : Run Triton FA fwd kernel benchmark - with tensor descriptors
263
263
if : ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'flash_attention_tensor_desc_benchmark.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'flash_attention_tensor_desc_benchmark.py') }}
@@ -267,22 +267,8 @@ jobs:
267
267
mv $REPORTS/attn-performance.csv $REPORTS/attn-tensor-desc-performance.csv
268
268
269
269
source ../../scripts/capture-hw-details.sh
270
- python build_report.py $REPORTS/attn-tensor-desc-performance.csv $REPORTS/attn-tensor-desc-triton-report.csv --benchmark attn-tensor-desc --compiler triton --param_cols "Z,H,N_CTX,D_HEAD,CAUSAL" --tflops_col Triton-TFlops --hbm_col "Triton-GB/s" --tag $TAG
271
- python build_report.py $REPORTS/attn-tensor-desc-performance.csv $REPORTS/attn-tensor-desc-xetla-report.csv --benchmark attn-tensor-desc --compiler xetla --param_cols "Z,H,N_CTX,D_HEAD,CAUSAL" --tflops_col XeTLA-TFlops --hbm_col "XeTLA-GB/s" --tag $TAG
272
-
273
- - name : Run Prefix Sums kernel benchmark
274
- if : ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'prefix_sums.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'prefix_sums.py') }}
275
- run : |
276
- cd benchmarks/triton_kernels_benchmark
277
- python prefix_sums.py --reports $REPORTS --n_runs $N_RUNS
278
- source ../../scripts/capture-hw-details.sh
279
- python build_report.py $REPORTS/prefix-sums.csv $REPORTS/prefix_sums-triton-report.csv --benchmark prefix_sums --compiler triton --param_cols "N" --tflops_col Triton-TFlops --hbm_col "Triton-GB/s" --tag $TAG
280
-
281
- - name : Run micro benchmark
282
- if : ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'micro_benchmarks.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'micro_benchmarks') }}
283
- run : |
284
- cd benchmarks/micro_benchmarks
285
- python run_benchmarks.py --reports $REPORTS
270
+ python build_report.py $REPORTS/attn-tensor-desc-performance.csv $REPORTS/attn-tensor-desc-triton-report.csv --benchmark flash-attn-tensor-desc --compiler triton --param_cols "Z,H,N_CTX,D_HEAD,CAUSAL" --tflops_col Triton-TFlops --hbm_col "Triton-GB/s" --tag $TAG
271
+ python build_report.py $REPORTS/attn-tensor-desc-performance.csv $REPORTS/attn-tensor-desc-xetla-report.csv --benchmark flash-attn-tensor-desc --compiler xetla --param_cols "Z,H,N_CTX,D_HEAD,CAUSAL" --tflops_col XeTLA-TFlops --hbm_col "XeTLA-GB/s" --tag $TAG
286
272
287
273
- name : Run Triton FlexAttention Causal Mask fwd kernel benchmark
288
274
if : ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'flex_attention_benchmark_causal_mask.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'flex_attention_benchmark_causal_mask.py') }}
@@ -291,7 +277,7 @@ jobs:
291
277
python flex_attention_benchmark_causal_mask.py --reports $REPORTS --n_runs $N_RUNS
292
278
293
279
source ../../scripts/capture-hw-details.sh
294
- python build_report.py $REPORTS/flexAttnCausal-performance.csv $REPORTS/flexAttnCausal-triton-report.csv --benchmark flexAttnCausal --compiler triton --param_cols "Z,H_q,H_kv,N_CTX_q,N_CTX_kv,D_HEAD_qk,D_HEAD_v" --tflops_col Triton-TFlops --hbm_col "Triton-GB/s" --tag $TAG
280
+ python build_report.py $REPORTS/flexAttnCausal-performance.csv $REPORTS/flexAttnCausal-triton-report.csv --benchmark flex-attn-causal --compiler triton --param_cols "Z,H_q,H_kv,N_CTX_q,N_CTX_kv,D_HEAD_qk,D_HEAD_v" --tflops_col Triton-TFlops --hbm_col "Triton-GB/s" --tag $TAG
295
281
296
282
- name : Run Triton FlexAttention Custom Masks fwd kernel benchmark
297
283
if : ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'flex_attention_benchmark_custom_masks.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'flex_attention_benchmark_custom_masks.py') }}
@@ -300,9 +286,22 @@ jobs:
300
286
python flex_attention_benchmark_custom_masks.py --reports $REPORTS --n_runs $N_RUNS
301
287
302
288
source ../../scripts/capture-hw-details.sh
303
- python build_report.py $REPORTS/flexAttnMasks-performance.csv $REPORTS/flexAttnMasks-triton-report.csv --benchmark flexAttnMasks --compiler triton --param_cols "Z,H,N_CTX,D_HEAD,MASK" --tflops_col Triton-TFlops --hbm_col "Triton-GB/s" --tag $TAG --mask
304
- python build_report.py $REPORTS/flexAttnMasks-performance.csv $REPORTS/flexAttnMasks-onednn-report.csv --benchmark flexAttnMasks --compiler onednn --param_cols "Z,H,N_CTX,D_HEAD,MASK" --tflops_col OneDNN-TFlops --hbm_col "OneDNN-GB/s" --tag $TAG --mask
289
+ python build_report.py $REPORTS/flexAttnMasks-performance.csv $REPORTS/flexAttnMasks-triton-report.csv --benchmark flex-attn-masks --compiler triton --param_cols "Z,H,N_CTX,D_HEAD,MASK" --tflops_col Triton-TFlops --hbm_col "Triton-GB/s" --tag $TAG --mask
290
+ python build_report.py $REPORTS/flexAttnMasks-performance.csv $REPORTS/flexAttnMasks-onednn-report.csv --benchmark flex-attn-masks --compiler onednn --param_cols "Z,H,N_CTX,D_HEAD,MASK" --tflops_col OneDNN-TFlops --hbm_col "OneDNN-GB/s" --tag $TAG --mask
305
291
292
+ - name : Run Prefix Sums kernel benchmark
293
+ if : ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'prefix_sums.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'prefix_sums.py') }}
294
+ run : |
295
+ cd benchmarks/triton_kernels_benchmark
296
+ python prefix_sums.py --reports $REPORTS --n_runs $N_RUNS
297
+ source ../../scripts/capture-hw-details.sh
298
+ python build_report.py $REPORTS/prefix-sums.csv $REPORTS/prefix_sums-triton-report.csv --benchmark prefix_sums --compiler triton --param_cols "N" --tflops_col Triton-TFlops --hbm_col "Triton-GB/s" --tag $TAG
299
+
300
+ - name : Run micro benchmark
301
+ if : ${{ steps.install.outcome == 'success' && !cancelled() && (inputs.benchmarks == '' || contains(fromJson(inputs.benchmarks || '[]'), 'micro_benchmarks.py')) && !contains(fromJson(inputs.skip_benchmarks || '[]'), 'micro_benchmarks') }}
302
+ run : |
303
+ cd benchmarks/micro_benchmarks
304
+ python run_benchmarks.py --reports $REPORTS
306
305
307
306
- name : Upload benchmark reports
308
307
if : ${{ steps.install.outcome == 'success' && !cancelled() }}
0 commit comments