Bug fix for building triton kernel

Summary:
This diff includes a few bug fixes:

1. a bug in build_triton_func

2. https://www.internalfb.com/diff/D59195933 changed the format of grid information for triton kernel. et_replay also need to change the triton kernel execution function to pass in grid information correctly

3. updated resnet test files to have process_group:init node.

4. updated GPT2 PT2 trace file to have the new format for grid information

Reviewed By: briancoutinho

Differential Revision: D59195828

et_replay/lib/et_replay_utils.py

@@ -457,15 +457,15 @@ def build_triton_func(n, resources_dir, async_compile, device):
     with open(os.path.join(resources_dir, n.kernel_file), "r") as f:
         code = f.read()
 
+    func = None
     # TORCHINDUCTOR_UNIQUE_KERNEL_NAMES controls whether each triton
     # kernel is given a unique name or not, if it is not, then the
     # kernel name will be "triton_" for all triton kernels.
     try:
         func = async_compile.triton(n.name, code, device_str=device)
     except Exception:
         func = async_compile.triton("triton_", code, device_str=device)
-    finally:
-        func = None
+
     return func, 0
 
 

et_replay/tools/et_replay.py

@@ -1128,13 +1128,8 @@ def run_op(self, node, iter):
             outputs = []
             if output_count == 0:
                 if node.kernel_backend == "triton":
-                    # remove the last comma
-                    grid_info = inputs[-2]
-                    index = grid_info.rfind(",")
-                    if index >= 0:
-                        grid_info = grid_info[:index] + grid_info[index + 1 :]
                     exec(
-                        f"func.run(*inputs[:-2], grid={grid_info}, stream={inputs[-1]})"
+                        f"func.run(*inputs[:-2], grid={inputs[-2]}, stream={inputs[-1]})"
                     )
                 else:
                     func(*inputs)