Skip test_mm_triton_kernel_benchmark (#1376)

* Running triton kernel on ROCM only has one GB/s metric reported

* Update test_kernel_benchmark.py

Author: pragupta

test/inductor/test_kernel_benchmark.py

@@ -12,7 +12,7 @@
 from torch._inductor.utils import fresh_inductor_cache
 from torch.testing import FileCheck
 from torch.testing._internal.inductor_utils import GPU_TYPE, HAS_GPU
-
+from torch.testing._internal.common_utils import skipIfRocm
 
 class TestKernelBenchmark(TestCase):
     @classmethod
@@ -97,6 +97,7 @@ def f(a, b):
 
     @config.patch(max_autotune=True, max_autotune_gemm_backends="TRITON")
     @fresh_inductor_cache()
+    @skipIfRocm #This seems to be disabled upstream https://github.com/pytorch/pytorch/issues/118346
     def test_mm_triton_kernel_benchmark(self):
         M = 2048
         N = 2432