[llvm] [AMDGPU][True16][CodeGen] use vgpr16 for zext patterns (reopen #153894) (PR #154211)

[Brox Chen via llvm-commits]

================
@@ -0,0 +1,197 @@
+; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 5
+; RUN:  llc -amdgpu-scalarize-global-loads=false  -mtriple=amdgcn -mcpu=gfx1100 -mattr=-flat-for-global,+real-true16 < %s | FileCheck -check-prefixes=GFX11 %s
+; RUN:  llc -amdgpu-scalarize-global-loads=false  -mtriple=amdgcn -mcpu=gfx1100 -mattr=-flat-for-global,-real-true16 < %s | FileCheck -check-prefixes=GFX11 %s
+
+define amdgpu_kernel void @zextload_global_i18_to_i16(ptr addrspace(1) %out, ptr addrspace(1) %in) nounwind {
+; GFX11-LABEL: zextload_global_i18_to_i16:
+; GFX11:       ; %bb.0:
+; GFX11-NEXT:    s_load_b128 s[0:3], s[4:5], 0x24
+; GFX11-NEXT:    s_mov_b32 s6, -1
+; GFX11-NEXT:    s_mov_b32 s7, 0x31016000
+; GFX11-NEXT:    s_mov_b32 s10, s6
+; GFX11-NEXT:    s_mov_b32 s11, s7
+; GFX11-NEXT:    s_waitcnt lgkmcnt(0)
+; GFX11-NEXT:    s_mov_b32 s8, s2
+; GFX11-NEXT:    s_mov_b32 s9, s3
+; GFX11-NEXT:    s_mov_b32 s4, s0
+; GFX11-NEXT:    buffer_load_u8 v0, off, s[8:11], 0
----------------
broxigarchen wrote:

right there are a few issues with this test, just updated

https://github.com/llvm/llvm-project/pull/154211