[PyTorch] Check is_same instead of data_ptr in addmm_out_cuda_impl (p…

…ytorch#55111)

Summary:
Pull Request resolved: pytorch#55111

I don't see how we could have ended up with !is_same but also identical data_ptr, and is_same is cheaper.
ghstack-source-id: 125438822

Test Plan: Existing CI?

Reviewed By: ngimel

Differential Revision: D27484914

fbshipit-source-id: 22125b29e6e09d312a2b92e893d08c69059e4435

aten/src/ATen/native/cuda/LinearAlgebra.cu

@@ -168,7 +168,7 @@ Tensor& addmm_out_cuda_impl(Tensor& result, const Tensor& self, const Tensor& ma
       result_ptr, result_ld
     );
   });
-  if (result.data_ptr() != result_->data_ptr()) {
+  if (!result.is_same(*result_)) {
     result.copy_(*result_);
   }
   return result;