[llvm] [AMDGPU] Replace zext pattern from reg_sequence to cvt_u32_u16 (PR #208045)

[Guo Chen via llvm-commits]

================
@@ -797,10 +797,8 @@ define i32 @zext_fma_f16(half %x, half %y, half %z) {
 ; GFX11-TRUE16-LABEL: zext_fma_f16:
 ; GFX11-TRUE16:       ; %bb.0:
 ; GFX11-TRUE16-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
-; GFX11-TRUE16-NEXT:    v_mov_b16_e32 v0.h, v0.l
-; GFX11-TRUE16-NEXT:    v_mov_b16_e32 v0.l, v2.l
-; GFX11-TRUE16-NEXT:    v_fmac_f16_e32 v0.l, v0.h, v1.l
-; GFX11-TRUE16-NEXT:    v_mov_b16_e32 v0.h, 0
+; GFX11-TRUE16-NEXT:    v_fmac_f16_e32 v2.l, v0.l, v1.l
----------------
broxigarchen wrote:

another example to shows code quality improvement

https://github.com/llvm/llvm-project/pull/208045