llvm/test/CodeGen/AMDGPU/rotr.i64.ll

; RUN: llc -mtriple=amdgpu6.00 < %s | FileCheck -check-prefix=SI -check-prefix=BOTH %s
; RUN: llc -mtriple=amdgpu8.02 < %s | FileCheck -check-prefix=VI -check-prefix=BOTH %s

; BOTH-LABEL: {{^}}s_rotr_i64:
; BOTH-DAG: s_sub_i32
; BOTH-DAG: s_lshr_b64
; BOTH-DAG: s_lshl_b64
; BOTH: s_or_b64
define amdgpu_kernel void @s_rotr_i64(ptr addrspace(1) %in, i64 %x, i64 %y) {
entry:
  %tmp0 = sub i64 64, %y
  %tmp1 = shl i64 %x, %tmp0
  %tmp2 = lshr i64 %x, %y
  %tmp3 = or i64 %tmp1, %tmp2
  store i64 %tmp3, ptr addrspace(1) %in
  ret void
}

; BOTH-LABEL: {{^}}v_rotr_i64:
; BOTH-DAG: v_sub_{{[iu]}}32
; SI-DAG: v_lshr_b64
; SI-DAG: v_lshl_b64
; VI-DAG: v_lshrrev_b64
; VI-DAG: v_lshlrev_b64
; BOTH: v_or_b32
; BOTH: v_or_b32
define i64 @v_rotr_i64(i64 %x, i64 %y) {
  %tmp0 = sub i64 64, %y
  %tmp1 = shl i64 %x, %tmp0
  %tmp2 = lshr i64 %x, %y
  %tmp3 = or i64 %tmp1, %tmp2
  ret i64 %tmp3
}

; BOTH-LABEL: {{^}}s_rotr_v2i64:
define amdgpu_kernel void @s_rotr_v2i64(ptr addrspace(1) %in, <2 x i64> %x, <2 x i64> %y) {
entry:
  %tmp0 = sub <2 x i64> <i64 64, i64 64>, %y
  %tmp1 = shl <2 x i64> %x, %tmp0
  %tmp2 = lshr <2 x i64> %x, %y
  %tmp3 = or <2 x i64> %tmp1, %tmp2
  store <2 x i64> %tmp3, ptr addrspace(1) %in
  ret void
}

; BOTH-LABEL: {{^}}v_rotr_v2i64:
define <2 x i64> @v_rotr_v2i64(<2 x i64> %x, <2 x i64> %y) {
  %tmp0 = sub <2 x i64> <i64 64, i64 64>, %y
  %tmp1 = shl <2 x i64> %x, %tmp0
  %tmp2 = lshr <2 x i64> %x, %y
  %tmp3 = or <2 x i64> %tmp1, %tmp2
  ret <2 x i64> %tmp3
}