[llvm] [AMDGPU][DAG] Enable ganging up of memcpy loads/stores for AMDGPU (PR #96185)

[Fabian Ritter via llvm-commits]

================
@@ -67,6 +67,9 @@ AMDGPUTargetLowering::AMDGPUTargetLowering(const TargetMachine &TM,
   MaxStoresPerMemcpy = MaxStoresPerMemcpyOptSize = ~0U;
   MaxStoresPerMemmove = MaxStoresPerMemmoveOptSize = ~0U;
 
+  // Enable ganging up loads and stores in the memcpy DAG lowering.
----------------
ritter-x2a wrote:

I just added a generated test (with alignments 8 and 16 instead of 4, leading to 97k lines) to the PR, so that you can see what it looks like. Running it alone costs ca. 5 seconds on my workstation.

https://github.com/llvm/llvm-project/pull/96185