Refactor gradient tests for addmm and addbmm

Updated gradient tests to only use long configurations for addmm and addbmm benchmarks.
Fix duplicate benchmarking for addmm
2025-11-01 21:47:15 +08:00 · 2025-10-30 10:04:21 -07:00 · 2025-10-29 19:19:33 -07:00
1 changed files with 3 additions and 7 deletions
--- a/benchmarks/operator_benchmark/pt/addmm_test.py
+++ b/benchmarks/operator_benchmark/pt/addmm_test.py
@ -53,10 +53,8 @@ class AddmmBenchmark(op_bench.TorchBenchmarkBase):
        return torch.addmm(input_one, mat1, mat2)


-op_bench.generate_pt_test(addmm_long_configs + addmm_long_configs, AddmmBenchmark)
-op_bench.generate_pt_gradient_test(
-    addmm_long_configs + addmm_long_configs, AddmmBenchmark
-)
+op_bench.generate_pt_test(addmm_short_configs + addmm_long_configs, AddmmBenchmark)
+op_bench.generate_pt_gradient_test(addmm_long_configs, AddmmBenchmark)

 """Mircobenchmark for addbmm operator."""

@ -107,9 +105,7 @@ addbmm_short_configs = op_bench.cross_product_configs(
 )

 op_bench.generate_pt_test(addbmm_long_configs + addbmm_short_configs, AddbmmBenchmark)
-op_bench.generate_pt_gradient_test(
-    addbmm_long_configs + addbmm_short_configs, AddbmmBenchmark
-)
+op_bench.generate_pt_gradient_test(addbmm_long_configs, AddbmmBenchmark)

 if __name__ == "__main__":
    op_bench.benchmark_runner.main()
Author	SHA1	Message	Date
Apurva Jain	85ba89cd8f	Refactor gradient tests for addmm and addbmm Updated gradient tests to only use long configurations for addmm and addbmm benchmarks.	2025-10-30 10:04:21 -07:00
jainapurva	eb35fb401a	Fix duplicate benchmarking for addmm	2025-10-29 19:19:33 -07:00