Symintify fused_scaled_matmul_reduce_scatter (pytorch#167122)

Symintify fused_scaled_matmul_reduce_scatter (pytorch#165086)

Fixes #ISSUE_NUMBER

Pull Request resolved: pytorch#165086
Approved by: https://github.com/zou3519, https://github.com/Skylion007

(cherry picked from commit 4a0df39)

Co-authored-by: angelayi <[email protected]>

Author: pytorchbot

torch/distributed/_symmetric_memory/__init__.py

@@ -450,7 +450,7 @@ def get_p2p_buf(rank: int, idx: int) -> torch.Tensor:
 lib.define(
     "fused_scaled_matmul_reduce_scatter("
     "Tensor A, Tensor B, Tensor A_scale, Tensor B_scale, "
-    "str reduce_op, int orig_scatter_dim, int scatter_dim_after_maybe_reshape, str group_name, int[]? output_shape, "
+    "str reduce_op, int orig_scatter_dim, int scatter_dim_after_maybe_reshape, str group_name, SymInt[]? output_shape, "
     "Tensor? bias = None, "
     "Tensor? result_scale = None, "
     "ScalarType? out_dtype = None, "