Set use_cuda_graphs in fp8_gemm_rowwise

Summary: The default value for use_cuda_graphs was changed to False in D64471087 and this caused slowdowns in triton/ck kernels for fp8_gemm_rowwise.

Reviewed By: danzimm

Differential Revision: D65140285

fbshipit-source-id: 4ab77537afeb9108dab7cdef6cac34aaa39d7d73

tritonbench/operators/fp8_gemm_rowwise/operator.py

@@ -111,6 +111,7 @@ def __init__(
         self, tb_args: argparse.Namespace, extra_args: Optional[List[str]] = None
     ):
         super().__init__(tb_args, extra_args)
+        self.use_cuda_graphs = True
         addmm_args = parse_args(self.extra_args)
         if addmm_args.m and addmm_args.n and addmm_args.k:
             self.shapes = [(addmm_args.m, addmm_args.n, addmm_args.k)]