llvm/test/CodeGen/AMDGPU/insert-vector-elt-build-vector-poison.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py UTC_ARGS: --version 6
; RUN: llc -mtriple=amdgpu9.00 < %s | FileCheck %s

; Two vectors are derived from the same poison-padded base constant. CSE shares
; the base BUILD_VECTOR; the combine rebuilds each insert as its own
; BUILD_VECTOR so the shared base does not force copies of its unused lanes.
define void @shared_poison_base(i32 %a, i32 %b, ptr addrspace(1) %out0, ptr addrspace(1) %out1) {
; CHECK-LABEL: shared_poison_base:
; CHECK:       ; %bb.0:
; CHECK-NEXT:    s_waitcnt vmcnt(0) expcnt(0) lgkmcnt(0)
; CHECK-NEXT:    v_mov_b32_e32 v7, v0
; CHECK-NEXT:    v_mov_b32_e32 v6, 1
; CHECK-NEXT:    global_store_dwordx4 v[2:3], v[6:9], off
; CHECK-NEXT:    global_store_dwordx4 v[4:5], v[1:4], off
; CHECK-NEXT:    s_waitcnt vmcnt(0)
; CHECK-NEXT:    s_setpc_b64 s[30:31]
  %v0 = insertelement <4 x i32> <i32 1, i32 poison, i32 poison, i32 poison>, i32 %a, i64 1
  %v1 = insertelement <4 x i32> <i32 1, i32 poison, i32 poison, i32 poison>, i32 %b, i64 0
  store <4 x i32> %v0, ptr addrspace(1) %out0
  store <4 x i32> %v1, ptr addrspace(1) %out1
  ret void
}