[llvm-branch-commits] [llvm] [AMDGPU] Extract byte lanes of a split vector from the 32-bit source (PR #228424)

[Diana Picus via llvm-branch-commits]

================
@@ -453,13 +453,14 @@ define amdgpu_kernel void @lshr_v_imm_v2i16(ptr addrspace(1) %out, ptr addrspace
 ; VI-NEXT:    v_add_u32_e32 v0, vcc, s2, v2
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    flat_load_dword v3, v[0:1]
-; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v2
 ; VI-NEXT:    v_mov_b32_e32 v1, s1
+; VI-NEXT:    v_add_u32_e32 v0, vcc, s0, v2
+; VI-NEXT:    s_mov_b32 s0, 0x5040c00
 ; VI-NEXT:    v_addc_u32_e32 v1, vcc, 0, v1, vcc
 ; VI-NEXT:    s_waitcnt vmcnt(0)
-; VI-NEXT:    v_lshrrev_b32_e32 v2, 24, v3
-; VI-NEXT:    v_lshlrev_b32_e32 v2, 16, v2
-; VI-NEXT:    v_or_b32_sdwa v2, v3, v2 dst_sel:DWORD dst_unused:UNUSED_PAD src0_sel:BYTE_1 src1_sel:DWORD
+; VI-NEXT:    v_lshrrev_b32_e32 v2, 8, v3
+; VI-NEXT:    v_lshrrev_b32_e32 v3, 24, v3
+; VI-NEXT:    v_perm_b32 v2, v3, v2, s0
----------------
rovka wrote:

Would be nice if we could get rid of the shifts here.

https://github.com/llvm/llvm-project/pull/228424