mode=['short'], sample_func=cross_product ) @torch.jit.script def torch_matmul(a, b, iterations): # type: (Tensor, Tensor, int) result = torch.jit.annotate(torch.Tensor, None) for _ in range(iterations): result = torch.matmul(a, b) return result @benchmark_core.register_test def test_matmul(): generate_pt_test( [long_config, short_config], map_pt_config_matmul, [('matmul', torch_matmul)] ) generate_c2_test( [long_config, short_config], map_c2_config_matmul, [('matmul', 'MatMul')], ) if __name__ == "__main__": benchmark_runner.main()
mode=['short'], sample_func=cross_product ) @torch.jit.script def torch_add(a, b, iterations): # type: (Tensor, Tensor, int) result = torch.jit.annotate(torch.Tensor, None) for _ in range(iterations): result = torch.add(a, b) return result @benchmark_core.register_test def test_add(): generate_pt_test( [long_config, short_config], map_pt_config_add, [('add', torch_add)] ) generate_c2_test( [long_config, short_config], map_c2_config_add, [('add', 'Add')], ) if __name__ == "__main__": benchmark_runner.main()