[llvm] [AMDGPU] Enable GCNRewritePartialRegUses pass by default. (PR #72975)

[Valery Pykhtin via llvm-commits]

================
@@ -136,13 +136,13 @@ define amdgpu_kernel void @max_256_vgprs_spill_9x32(ptr addrspace(1) %p) #1 {
 ; GFX908-DAG: v_accvgpr_read_b32
 
 ; GFX900: NumVgprs: 256
-; GFX908: NumVgprs: 254
-; GFX900: ScratchSize: 1796
+; GFX908: NumVgprs: 252
+; GFX900: ScratchSize: 132
----------------
vpykhtin wrote:

> Looks like rename-independent-subregs should be shrinking/narrowing registers, but will need a target hook to be able to find if that is possible and which register class to use.
That is what I originally thought to do, but I didn't want to complicate things dealing with other targets. Maybe once this pass get matured it can be made target-independent (it's almost indepent at the moment). I also remember @arsenm had an intention to make it more like an utility class to be used when needed.



https://github.com/llvm/llvm-project/pull/72975