llvm/llvm/test/CodeGen/AArch64/sve-streaming-mode-fixed-length-optimize-ptrue.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc -mattr=+sve -force-streaming-compatible < %s | FileCheck %s
; RUN: llc -mattr=+sme -force-streaming < %s | FileCheck %s
; RUN: llc -force-streaming-compatible < %s | FileCheck %s --check-prefix=NONEON-NOSVE


target triple = "aarch64-unknown-linux-gnu"

define void @add_v4i8(ptr %a, ptr %b) {
; CHECK-LABEL: add_v4i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl4
; CHECK-NEXT:    ld1b { z0.h }, p0/z, [x0]
; CHECK-NEXT:    ld1b { z1.h }, p0/z, [x1]
; CHECK-NEXT:    add z0.h, z0.h, z1.h
; CHECK-NEXT:    st1b { z0.h }, p0, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: add_v4i8:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    ldrb w8, [x0, #3]
; NONEON-NOSVE-NEXT:    ldrb w9, [x1, #3]
; NONEON-NOSVE-NEXT:    ldrb w10, [x0, #2]
; NONEON-NOSVE-NEXT:    ldrb w11, [x0, #1]
; NONEON-NOSVE-NEXT:    ldrb w12, [x1, #2]
; NONEON-NOSVE-NEXT:    ldrb w13, [x0]
; NONEON-NOSVE-NEXT:    add w8, w8, w9
; NONEON-NOSVE-NEXT:    ldrb w14, [x1, #1]
; NONEON-NOSVE-NEXT:    ldrb w9, [x1]
; NONEON-NOSVE-NEXT:    add w10, w10, w12
; NONEON-NOSVE-NEXT:    strb w8, [x0, #3]
; NONEON-NOSVE-NEXT:    add w8, w11, w14
; NONEON-NOSVE-NEXT:    add w9, w13, w9
; NONEON-NOSVE-NEXT:    strb w10, [x0, #2]
; NONEON-NOSVE-NEXT:    strb w8, [x0, #1]
; NONEON-NOSVE-NEXT:    strb w9, [x0]
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <4 x i8>, ptr %a
  %op2 = load <4 x i8>, ptr %b
  %res = add <4 x i8> %op1, %op2
  store <4 x i8> %res, ptr %a
  ret void
}

define void @add_v8i8(ptr %a, ptr %b) {
; CHECK-LABEL: add_v8i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr d0, [x0]
; CHECK-NEXT:    ldr d1, [x1]
; CHECK-NEXT:    add z0.b, z0.b, z1.b
; CHECK-NEXT:    str d0, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: add_v8i8:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    sub sp, sp, #32
; NONEON-NOSVE-NEXT:    .cfi_def_cfa_offset 32
; NONEON-NOSVE-NEXT:    ldr d0, [x1]
; NONEON-NOSVE-NEXT:    ldr d1, [x0]
; NONEON-NOSVE-NEXT:    stp d1, d0, [sp, #8]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #23]
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #15]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #14]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #31]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #22]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #13]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #30]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #21]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #12]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #29]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #20]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #11]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #28]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #19]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #10]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #27]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #18]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #9]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #26]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #17]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #8]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #25]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #16]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    strb w8, [sp, #24]
; NONEON-NOSVE-NEXT:    ldr d0, [sp, #24]
; NONEON-NOSVE-NEXT:    str d0, [x0]
; NONEON-NOSVE-NEXT:    add sp, sp, #32
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <8 x i8>, ptr %a
  %op2 = load <8 x i8>, ptr %b
  %res = add <8 x i8> %op1, %op2
  store <8 x i8> %res, ptr %a
  ret void
}

define void @add_v16i8(ptr %a, ptr %b) {
; CHECK-LABEL: add_v16i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr q0, [x0]
; CHECK-NEXT:    ldr q1, [x1]
; CHECK-NEXT:    add z0.b, z0.b, z1.b
; CHECK-NEXT:    str q0, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: add_v16i8:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    ldr q0, [x1]
; NONEON-NOSVE-NEXT:    ldr q1, [x0]
; NONEON-NOSVE-NEXT:    stp q1, q0, [sp, #-48]!
; NONEON-NOSVE-NEXT:    .cfi_def_cfa_offset 48
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #31]
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #15]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #14]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #47]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #30]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #13]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #46]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #29]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #12]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #45]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #28]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #11]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #44]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #27]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #10]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #43]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #26]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #9]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #42]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #25]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #8]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #41]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #24]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #7]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #40]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #23]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #6]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #39]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #22]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #5]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #38]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #21]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #4]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #37]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #20]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #3]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #36]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #19]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #2]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #35]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #18]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #1]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #34]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #17]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #33]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #16]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    strb w8, [sp, #32]
; NONEON-NOSVE-NEXT:    ldr q0, [sp, #32]
; NONEON-NOSVE-NEXT:    str q0, [x0]
; NONEON-NOSVE-NEXT:    add sp, sp, #48
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <16 x i8>, ptr %a
  %op2 = load <16 x i8>, ptr %b
  %res = add <16 x i8> %op1, %op2
  store <16 x i8> %res, ptr %a
  ret void
}

define void @add_v32i8(ptr %a, ptr %b) {
; CHECK-LABEL: add_v32i8:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldp q0, q3, [x1]
; CHECK-NEXT:    ldp q1, q2, [x0]
; CHECK-NEXT:    add z0.b, z1.b, z0.b
; CHECK-NEXT:    add z1.b, z2.b, z3.b
; CHECK-NEXT:    stp q0, q1, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: add_v32i8:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    sub sp, sp, #96
; NONEON-NOSVE-NEXT:    .cfi_def_cfa_offset 96
; NONEON-NOSVE-NEXT:    ldp q3, q0, [x1]
; NONEON-NOSVE-NEXT:    ldp q2, q1, [x0]
; NONEON-NOSVE-NEXT:    stp q2, q3, [sp]
; NONEON-NOSVE-NEXT:    stp q1, q0, [sp, #32]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #63]
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #47]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #46]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #95]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #62]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #45]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #94]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #61]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #44]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #93]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #60]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #43]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #92]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #59]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #42]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #91]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #58]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #41]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #90]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #57]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #40]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #89]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #56]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #39]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #88]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #55]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #38]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #87]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #54]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #37]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #86]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #53]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #36]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #85]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #52]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #35]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #84]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #51]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #34]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #83]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #50]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #33]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #82]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #49]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #32]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #81]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #48]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #15]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #80]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #31]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #14]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #79]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #30]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #13]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #78]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #29]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #12]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #77]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #28]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #11]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #76]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #27]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #10]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #75]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #26]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #9]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #74]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #25]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #8]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #73]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #24]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #7]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #72]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #23]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #6]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #71]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #22]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #5]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #70]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #21]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #4]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #69]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #20]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #3]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #68]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #19]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #2]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #67]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #18]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp, #1]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #66]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #17]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrb w9, [sp]
; NONEON-NOSVE-NEXT:    strb w8, [sp, #65]
; NONEON-NOSVE-NEXT:    ldrb w8, [sp, #16]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    strb w8, [sp, #64]
; NONEON-NOSVE-NEXT:    ldp q0, q1, [sp, #64]
; NONEON-NOSVE-NEXT:    stp q0, q1, [x0]
; NONEON-NOSVE-NEXT:    add sp, sp, #96
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <32 x i8>, ptr %a
  %op2 = load <32 x i8>, ptr %b
  %res = add <32 x i8> %op1, %op2
  store <32 x i8> %res, ptr %a
  ret void
}

define void @add_v2i16(ptr %a, ptr %b, ptr %c) {
; CHECK-LABEL: add_v2i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl2
; CHECK-NEXT:    ld1h { z0.s }, p0/z, [x0]
; CHECK-NEXT:    ld1h { z1.s }, p0/z, [x1]
; CHECK-NEXT:    add z0.s, z0.s, z1.s
; CHECK-NEXT:    st1h { z0.s }, p0, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: add_v2i16:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    ldrh w8, [x0]
; NONEON-NOSVE-NEXT:    ldrh w9, [x1]
; NONEON-NOSVE-NEXT:    ldrh w10, [x0, #2]
; NONEON-NOSVE-NEXT:    ldrh w11, [x1, #2]
; NONEON-NOSVE-NEXT:    add w8, w8, w9
; NONEON-NOSVE-NEXT:    add w9, w10, w11
; NONEON-NOSVE-NEXT:    strh w8, [x0]
; NONEON-NOSVE-NEXT:    strh w9, [x0, #2]
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <2 x i16>, ptr %a
  %op2 = load <2 x i16>, ptr %b
  %res = add <2 x i16> %op1, %op2
  store <2 x i16> %res, ptr %a
  ret void
}

define void @add_v4i16(ptr %a, ptr %b, ptr %c) {
; CHECK-LABEL: add_v4i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr d0, [x0]
; CHECK-NEXT:    ldr d1, [x1]
; CHECK-NEXT:    add z0.h, z0.h, z1.h
; CHECK-NEXT:    str d0, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: add_v4i16:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    sub sp, sp, #32
; NONEON-NOSVE-NEXT:    .cfi_def_cfa_offset 32
; NONEON-NOSVE-NEXT:    ldr d0, [x1]
; NONEON-NOSVE-NEXT:    ldr d1, [x0]
; NONEON-NOSVE-NEXT:    stp d1, d0, [sp, #8]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #22]
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #14]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #12]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #30]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #20]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #10]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #28]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #18]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #8]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #26]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #16]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    strh w8, [sp, #24]
; NONEON-NOSVE-NEXT:    ldr d0, [sp, #24]
; NONEON-NOSVE-NEXT:    str d0, [x0]
; NONEON-NOSVE-NEXT:    add sp, sp, #32
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <4 x i16>, ptr %a
  %op2 = load <4 x i16>, ptr %b
  %res = add <4 x i16> %op1, %op2
  store <4 x i16> %res, ptr %a
  ret void
}

define void @add_v8i16(ptr %a, ptr %b, ptr %c) {
; CHECK-LABEL: add_v8i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldr q0, [x0]
; CHECK-NEXT:    ldr q1, [x1]
; CHECK-NEXT:    add z0.h, z0.h, z1.h
; CHECK-NEXT:    str q0, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: add_v8i16:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    ldr q0, [x1]
; NONEON-NOSVE-NEXT:    ldr q1, [x0]
; NONEON-NOSVE-NEXT:    stp q1, q0, [sp, #-48]!
; NONEON-NOSVE-NEXT:    .cfi_def_cfa_offset 48
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #30]
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #14]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #12]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #46]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #28]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #10]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #44]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #26]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #8]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #42]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #24]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #6]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #40]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #22]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #4]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #38]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #20]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #2]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #36]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #18]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #34]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #16]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    strh w8, [sp, #32]
; NONEON-NOSVE-NEXT:    ldr q0, [sp, #32]
; NONEON-NOSVE-NEXT:    str q0, [x0]
; NONEON-NOSVE-NEXT:    add sp, sp, #48
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <8 x i16>, ptr %a
  %op2 = load <8 x i16>, ptr %b
  %res = add <8 x i16> %op1, %op2
  store <8 x i16> %res, ptr %a
  ret void
}

define void @add_v16i16(ptr %a, ptr %b, ptr %c) {
; CHECK-LABEL: add_v16i16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldp q0, q3, [x1]
; CHECK-NEXT:    ldp q1, q2, [x0]
; CHECK-NEXT:    add z0.h, z1.h, z0.h
; CHECK-NEXT:    add z1.h, z2.h, z3.h
; CHECK-NEXT:    stp q0, q1, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: add_v16i16:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    sub sp, sp, #96
; NONEON-NOSVE-NEXT:    .cfi_def_cfa_offset 96
; NONEON-NOSVE-NEXT:    ldp q3, q0, [x1]
; NONEON-NOSVE-NEXT:    ldp q2, q1, [x0]
; NONEON-NOSVE-NEXT:    stp q2, q3, [sp]
; NONEON-NOSVE-NEXT:    stp q1, q0, [sp, #32]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #62]
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #46]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #44]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #94]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #60]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #42]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #92]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #58]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #40]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #90]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #56]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #38]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #88]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #54]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #36]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #86]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #52]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #34]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #84]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #50]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #32]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #82]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #48]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #14]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #80]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #30]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #12]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #78]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #28]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #10]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #76]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #26]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #8]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #74]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #24]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #6]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #72]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #22]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #4]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #70]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #20]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp, #2]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #68]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #18]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    ldrh w9, [sp]
; NONEON-NOSVE-NEXT:    strh w8, [sp, #66]
; NONEON-NOSVE-NEXT:    ldrh w8, [sp, #16]
; NONEON-NOSVE-NEXT:    add w8, w9, w8
; NONEON-NOSVE-NEXT:    strh w8, [sp, #64]
; NONEON-NOSVE-NEXT:    ldp q0, q1, [sp, #64]
; NONEON-NOSVE-NEXT:    stp q0, q1, [x0]
; NONEON-NOSVE-NEXT:    add sp, sp, #96
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <16 x i16>, ptr %a
  %op2 = load <16 x i16>, ptr %b
  %res = add <16 x i16> %op1, %op2
  store <16 x i16> %res, ptr %a
  ret void
}

define void @abs_v2i32(ptr %a) {
; CHECK-LABEL: abs_v2i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl2
; CHECK-NEXT:    ldr d0, [x0]
; CHECK-NEXT:    abs z0.s, p0/m, z0.s
; CHECK-NEXT:    str d0, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: abs_v2i32:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    ldr d0, [x0]
; NONEON-NOSVE-NEXT:    str d0, [sp, #-16]!
; NONEON-NOSVE-NEXT:    .cfi_def_cfa_offset 16
; NONEON-NOSVE-NEXT:    ldr w8, [sp, #4]
; NONEON-NOSVE-NEXT:    cmp w8, #0
; NONEON-NOSVE-NEXT:    cneg w9, w8, mi
; NONEON-NOSVE-NEXT:    ldr w8, [sp]
; NONEON-NOSVE-NEXT:    cmp w8, #0
; NONEON-NOSVE-NEXT:    cneg w8, w8, mi
; NONEON-NOSVE-NEXT:    stp w8, w9, [sp, #8]
; NONEON-NOSVE-NEXT:    ldr d0, [sp, #8]
; NONEON-NOSVE-NEXT:    str d0, [x0]
; NONEON-NOSVE-NEXT:    add sp, sp, #16
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <2 x i32>, ptr %a
  %res = call <2 x i32> @llvm.abs.v2i32(<2 x i32> %op1, i1 false)
  store <2 x i32> %res, ptr %a
  ret void
}

define void @abs_v4i32(ptr %a) {
; CHECK-LABEL: abs_v4i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl4
; CHECK-NEXT:    ldr q0, [x0]
; CHECK-NEXT:    abs z0.s, p0/m, z0.s
; CHECK-NEXT:    str q0, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: abs_v4i32:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    ldr q0, [x0]
; NONEON-NOSVE-NEXT:    str q0, [sp, #-32]!
; NONEON-NOSVE-NEXT:    .cfi_def_cfa_offset 32
; NONEON-NOSVE-NEXT:    ldr w8, [sp, #12]
; NONEON-NOSVE-NEXT:    cmp w8, #0
; NONEON-NOSVE-NEXT:    cneg w9, w8, mi
; NONEON-NOSVE-NEXT:    ldr w8, [sp, #8]
; NONEON-NOSVE-NEXT:    cmp w8, #0
; NONEON-NOSVE-NEXT:    cneg w8, w8, mi
; NONEON-NOSVE-NEXT:    stp w8, w9, [sp, #24]
; NONEON-NOSVE-NEXT:    ldr w8, [sp, #4]
; NONEON-NOSVE-NEXT:    cmp w8, #0
; NONEON-NOSVE-NEXT:    cneg w9, w8, mi
; NONEON-NOSVE-NEXT:    ldr w8, [sp]
; NONEON-NOSVE-NEXT:    cmp w8, #0
; NONEON-NOSVE-NEXT:    cneg w8, w8, mi
; NONEON-NOSVE-NEXT:    stp w8, w9, [sp, #16]
; NONEON-NOSVE-NEXT:    ldr q0, [sp, #16]
; NONEON-NOSVE-NEXT:    str q0, [x0]
; NONEON-NOSVE-NEXT:    add sp, sp, #32
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <4 x i32>, ptr %a
  %res = call <4 x i32> @llvm.abs.v4i32(<4 x i32> %op1, i1 false)
  store <4 x i32> %res, ptr %a
  ret void
}

define void @abs_v8i32(ptr %a) {
; CHECK-LABEL: abs_v8i32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldp q0, q1, [x0]
; CHECK-NEXT:    ptrue p0.s, vl4
; CHECK-NEXT:    abs z0.s, p0/m, z0.s
; CHECK-NEXT:    abs z1.s, p0/m, z1.s
; CHECK-NEXT:    stp q0, q1, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: abs_v8i32:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    ldp q1, q0, [x0]
; NONEON-NOSVE-NEXT:    stp q1, q0, [sp, #-64]!
; NONEON-NOSVE-NEXT:    .cfi_def_cfa_offset 64
; NONEON-NOSVE-NEXT:    ldr w8, [sp, #28]
; NONEON-NOSVE-NEXT:    cmp w8, #0
; NONEON-NOSVE-NEXT:    cneg w9, w8, mi
; NONEON-NOSVE-NEXT:    ldr w8, [sp, #24]
; NONEON-NOSVE-NEXT:    cmp w8, #0
; NONEON-NOSVE-NEXT:    cneg w8, w8, mi
; NONEON-NOSVE-NEXT:    stp w8, w9, [sp, #56]
; NONEON-NOSVE-NEXT:    ldr w8, [sp, #20]
; NONEON-NOSVE-NEXT:    cmp w8, #0
; NONEON-NOSVE-NEXT:    cneg w9, w8, mi
; NONEON-NOSVE-NEXT:    ldr w8, [sp, #16]
; NONEON-NOSVE-NEXT:    cmp w8, #0
; NONEON-NOSVE-NEXT:    cneg w8, w8, mi
; NONEON-NOSVE-NEXT:    stp w8, w9, [sp, #48]
; NONEON-NOSVE-NEXT:    ldr w8, [sp, #12]
; NONEON-NOSVE-NEXT:    cmp w8, #0
; NONEON-NOSVE-NEXT:    cneg w9, w8, mi
; NONEON-NOSVE-NEXT:    ldr w8, [sp, #8]
; NONEON-NOSVE-NEXT:    cmp w8, #0
; NONEON-NOSVE-NEXT:    cneg w8, w8, mi
; NONEON-NOSVE-NEXT:    stp w8, w9, [sp, #40]
; NONEON-NOSVE-NEXT:    ldr w8, [sp, #4]
; NONEON-NOSVE-NEXT:    cmp w8, #0
; NONEON-NOSVE-NEXT:    cneg w9, w8, mi
; NONEON-NOSVE-NEXT:    ldr w8, [sp]
; NONEON-NOSVE-NEXT:    cmp w8, #0
; NONEON-NOSVE-NEXT:    cneg w8, w8, mi
; NONEON-NOSVE-NEXT:    stp w8, w9, [sp, #32]
; NONEON-NOSVE-NEXT:    ldp q0, q1, [sp, #32]
; NONEON-NOSVE-NEXT:    stp q0, q1, [x0]
; NONEON-NOSVE-NEXT:    add sp, sp, #64
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <8 x i32>, ptr %a
  %res = call <8 x i32> @llvm.abs.v8i32(<8 x i32> %op1, i1 false)
  store <8 x i32> %res, ptr %a
  ret void
}

define void @abs_v2i64(ptr %a) {
; CHECK-LABEL: abs_v2i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl2
; CHECK-NEXT:    ldr q0, [x0]
; CHECK-NEXT:    abs z0.d, p0/m, z0.d
; CHECK-NEXT:    str q0, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: abs_v2i64:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    ldr q0, [x0]
; NONEON-NOSVE-NEXT:    str q0, [sp, #-32]!
; NONEON-NOSVE-NEXT:    .cfi_def_cfa_offset 32
; NONEON-NOSVE-NEXT:    ldr x8, [sp, #8]
; NONEON-NOSVE-NEXT:    cmp x8, #0
; NONEON-NOSVE-NEXT:    cneg x9, x8, mi
; NONEON-NOSVE-NEXT:    ldr x8, [sp]
; NONEON-NOSVE-NEXT:    cmp x8, #0
; NONEON-NOSVE-NEXT:    cneg x8, x8, mi
; NONEON-NOSVE-NEXT:    stp x8, x9, [sp, #16]
; NONEON-NOSVE-NEXT:    ldr q0, [sp, #16]
; NONEON-NOSVE-NEXT:    str q0, [x0]
; NONEON-NOSVE-NEXT:    add sp, sp, #32
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <2 x i64>, ptr %a
  %res = call <2 x i64> @llvm.abs.v2i64(<2 x i64> %op1, i1 false)
  store <2 x i64> %res, ptr %a
  ret void
}

define void @abs_v4i64(ptr %a) {
; CHECK-LABEL: abs_v4i64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldp q0, q1, [x0]
; CHECK-NEXT:    ptrue p0.d, vl2
; CHECK-NEXT:    abs z0.d, p0/m, z0.d
; CHECK-NEXT:    abs z1.d, p0/m, z1.d
; CHECK-NEXT:    stp q0, q1, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: abs_v4i64:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    ldp q1, q0, [x0]
; NONEON-NOSVE-NEXT:    stp q1, q0, [sp, #-64]!
; NONEON-NOSVE-NEXT:    .cfi_def_cfa_offset 64
; NONEON-NOSVE-NEXT:    ldr x8, [sp, #24]
; NONEON-NOSVE-NEXT:    cmp x8, #0
; NONEON-NOSVE-NEXT:    cneg x9, x8, mi
; NONEON-NOSVE-NEXT:    ldr x8, [sp, #16]
; NONEON-NOSVE-NEXT:    cmp x8, #0
; NONEON-NOSVE-NEXT:    cneg x8, x8, mi
; NONEON-NOSVE-NEXT:    stp x8, x9, [sp, #48]
; NONEON-NOSVE-NEXT:    ldr x8, [sp, #8]
; NONEON-NOSVE-NEXT:    cmp x8, #0
; NONEON-NOSVE-NEXT:    cneg x9, x8, mi
; NONEON-NOSVE-NEXT:    ldr x8, [sp]
; NONEON-NOSVE-NEXT:    cmp x8, #0
; NONEON-NOSVE-NEXT:    cneg x8, x8, mi
; NONEON-NOSVE-NEXT:    stp x8, x9, [sp, #32]
; NONEON-NOSVE-NEXT:    ldp q0, q1, [sp, #32]
; NONEON-NOSVE-NEXT:    stp q0, q1, [x0]
; NONEON-NOSVE-NEXT:    add sp, sp, #64
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <4 x i64>, ptr %a
  %res = call <4 x i64> @llvm.abs.v4i64(<4 x i64> %op1, i1 false)
  store <4 x i64> %res, ptr %a
  ret void
}

define void @fadd_v2f16(ptr %a, ptr %b) {
; CHECK-LABEL: fadd_v2f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl4
; CHECK-NEXT:    ldr s0, [x0]
; CHECK-NEXT:    ldr s1, [x1]
; CHECK-NEXT:    fadd z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    fmov w8, s0
; CHECK-NEXT:    str w8, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: fadd_v2f16:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    ldr w8, [x0]
; NONEON-NOSVE-NEXT:    str w8, [sp, #-48]!
; NONEON-NOSVE-NEXT:    .cfi_def_cfa_offset 48
; NONEON-NOSVE-NEXT:    ldr w8, [x1]
; NONEON-NOSVE-NEXT:    str w8, [sp, #8]
; NONEON-NOSVE-NEXT:    ldp d0, d1, [sp]
; NONEON-NOSVE-NEXT:    stp d0, d1, [sp, #16]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #26]
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #18]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #16]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #34]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #24]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #32]
; NONEON-NOSVE-NEXT:    ldr d0, [sp, #32]
; NONEON-NOSVE-NEXT:    str d0, [sp, #40]
; NONEON-NOSVE-NEXT:    ldr w8, [sp, #40]
; NONEON-NOSVE-NEXT:    str w8, [x0]
; NONEON-NOSVE-NEXT:    add sp, sp, #48
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <2 x half>, ptr %a
  %op2 = load <2 x half>, ptr %b
  %res = fadd <2 x half> %op1, %op2
  store <2 x half> %res, ptr %a
  ret void
}

define void @fadd_v4f16(ptr %a, ptr %b) {
; CHECK-LABEL: fadd_v4f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl4
; CHECK-NEXT:    ldr d0, [x0]
; CHECK-NEXT:    ldr d1, [x1]
; CHECK-NEXT:    fadd z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    str d0, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: fadd_v4f16:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    sub sp, sp, #32
; NONEON-NOSVE-NEXT:    .cfi_def_cfa_offset 32
; NONEON-NOSVE-NEXT:    ldr d0, [x1]
; NONEON-NOSVE-NEXT:    ldr d1, [x0]
; NONEON-NOSVE-NEXT:    stp d1, d0, [sp, #8]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #22]
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #14]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #12]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #30]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #20]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #10]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #28]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #18]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #8]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #26]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #16]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #24]
; NONEON-NOSVE-NEXT:    ldr d0, [sp, #24]
; NONEON-NOSVE-NEXT:    str d0, [x0]
; NONEON-NOSVE-NEXT:    add sp, sp, #32
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <4 x half>, ptr %a
  %op2 = load <4 x half>, ptr %b
  %res = fadd <4 x half> %op1, %op2
  store <4 x half> %res, ptr %a
  ret void
}

define void @fadd_v8f16(ptr %a, ptr %b) {
; CHECK-LABEL: fadd_v8f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.h, vl8
; CHECK-NEXT:    ldr q0, [x0]
; CHECK-NEXT:    ldr q1, [x1]
; CHECK-NEXT:    fadd z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    str q0, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: fadd_v8f16:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    ldr q0, [x1]
; NONEON-NOSVE-NEXT:    ldr q1, [x0]
; NONEON-NOSVE-NEXT:    stp q1, q0, [sp, #-48]!
; NONEON-NOSVE-NEXT:    .cfi_def_cfa_offset 48
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #30]
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #14]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #12]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #46]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #28]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #10]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #44]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #26]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #8]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #42]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #24]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #6]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #40]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #22]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #4]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #38]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #20]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #2]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #36]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #18]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #34]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #16]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #32]
; NONEON-NOSVE-NEXT:    ldr q0, [sp, #32]
; NONEON-NOSVE-NEXT:    str q0, [x0]
; NONEON-NOSVE-NEXT:    add sp, sp, #48
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <8 x half>, ptr %a
  %op2 = load <8 x half>, ptr %b
  %res = fadd <8 x half> %op1, %op2
  store <8 x half> %res, ptr %a
  ret void
}

define void @fadd_v16f16(ptr %a, ptr %b) {
; CHECK-LABEL: fadd_v16f16:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldp q0, q3, [x1]
; CHECK-NEXT:    ptrue p0.h, vl8
; CHECK-NEXT:    ldp q1, q2, [x0]
; CHECK-NEXT:    fadd z0.h, p0/m, z0.h, z1.h
; CHECK-NEXT:    movprfx z1, z2
; CHECK-NEXT:    fadd z1.h, p0/m, z1.h, z3.h
; CHECK-NEXT:    stp q0, q1, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: fadd_v16f16:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    sub sp, sp, #96
; NONEON-NOSVE-NEXT:    .cfi_def_cfa_offset 96
; NONEON-NOSVE-NEXT:    ldp q3, q0, [x1]
; NONEON-NOSVE-NEXT:    ldp q2, q1, [x0]
; NONEON-NOSVE-NEXT:    stp q2, q3, [sp]
; NONEON-NOSVE-NEXT:    stp q1, q0, [sp, #32]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #62]
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #46]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #44]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #94]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #60]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #42]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #92]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #58]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #40]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #90]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #56]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #38]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #88]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #54]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #36]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #86]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #52]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #34]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #84]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #50]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #32]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #82]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #48]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #14]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #80]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #30]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #12]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #78]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #28]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #10]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #76]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #26]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #8]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #74]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #24]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #6]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #72]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #22]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #4]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #70]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #20]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp, #2]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #68]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #18]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldr h1, [sp]
; NONEON-NOSVE-NEXT:    fcvt s1, h1
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #66]
; NONEON-NOSVE-NEXT:    ldr h0, [sp, #16]
; NONEON-NOSVE-NEXT:    fcvt s0, h0
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    fcvt h0, s0
; NONEON-NOSVE-NEXT:    str h0, [sp, #64]
; NONEON-NOSVE-NEXT:    ldp q0, q1, [sp, #64]
; NONEON-NOSVE-NEXT:    stp q0, q1, [x0]
; NONEON-NOSVE-NEXT:    add sp, sp, #96
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <16 x half>, ptr %a
  %op2 = load <16 x half>, ptr %b
  %res = fadd <16 x half> %op1, %op2
  store <16 x half> %res, ptr %a
  ret void
}

define void @fadd_v2f32(ptr %a, ptr %b) {
; CHECK-LABEL: fadd_v2f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl2
; CHECK-NEXT:    ldr d0, [x0]
; CHECK-NEXT:    ldr d1, [x1]
; CHECK-NEXT:    fadd z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    str d0, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: fadd_v2f32:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    sub sp, sp, #32
; NONEON-NOSVE-NEXT:    .cfi_def_cfa_offset 32
; NONEON-NOSVE-NEXT:    ldr d0, [x1]
; NONEON-NOSVE-NEXT:    ldr d1, [x0]
; NONEON-NOSVE-NEXT:    stp d1, d0, [sp, #8]
; NONEON-NOSVE-NEXT:    ldp s1, s2, [sp, #8]
; NONEON-NOSVE-NEXT:    ldr s0, [sp, #20]
; NONEON-NOSVE-NEXT:    fadd s3, s2, s0
; NONEON-NOSVE-NEXT:    ldr s0, [sp, #16]
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    stp s0, s3, [sp, #24]
; NONEON-NOSVE-NEXT:    ldr d0, [sp, #24]
; NONEON-NOSVE-NEXT:    str d0, [x0]
; NONEON-NOSVE-NEXT:    add sp, sp, #32
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <2 x float>, ptr %a
  %op2 = load <2 x float>, ptr %b
  %res = fadd <2 x float> %op1, %op2
  store <2 x float> %res, ptr %a
  ret void
}

define void @fadd_v4f32(ptr %a, ptr %b) {
; CHECK-LABEL: fadd_v4f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.s, vl4
; CHECK-NEXT:    ldr q0, [x0]
; CHECK-NEXT:    ldr q1, [x1]
; CHECK-NEXT:    fadd z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    str q0, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: fadd_v4f32:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    ldr q0, [x1]
; NONEON-NOSVE-NEXT:    ldr q1, [x0]
; NONEON-NOSVE-NEXT:    stp q1, q0, [sp, #-48]!
; NONEON-NOSVE-NEXT:    .cfi_def_cfa_offset 48
; NONEON-NOSVE-NEXT:    ldp s1, s2, [sp, #8]
; NONEON-NOSVE-NEXT:    ldr s0, [sp, #28]
; NONEON-NOSVE-NEXT:    fadd s3, s2, s0
; NONEON-NOSVE-NEXT:    ldr s0, [sp, #24]
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldp s1, s2, [sp]
; NONEON-NOSVE-NEXT:    stp s0, s3, [sp, #40]
; NONEON-NOSVE-NEXT:    ldr s0, [sp, #20]
; NONEON-NOSVE-NEXT:    fadd s3, s2, s0
; NONEON-NOSVE-NEXT:    ldr s0, [sp, #16]
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    stp s0, s3, [sp, #32]
; NONEON-NOSVE-NEXT:    ldr q0, [sp, #32]
; NONEON-NOSVE-NEXT:    str q0, [x0]
; NONEON-NOSVE-NEXT:    add sp, sp, #48
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <4 x float>, ptr %a
  %op2 = load <4 x float>, ptr %b
  %res = fadd <4 x float> %op1, %op2
  store <4 x float> %res, ptr %a
  ret void
}

define void @fadd_v8f32(ptr %a, ptr %b) {
; CHECK-LABEL: fadd_v8f32:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldp q0, q3, [x1]
; CHECK-NEXT:    ptrue p0.s, vl4
; CHECK-NEXT:    ldp q1, q2, [x0]
; CHECK-NEXT:    fadd z0.s, p0/m, z0.s, z1.s
; CHECK-NEXT:    movprfx z1, z2
; CHECK-NEXT:    fadd z1.s, p0/m, z1.s, z3.s
; CHECK-NEXT:    stp q0, q1, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: fadd_v8f32:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    sub sp, sp, #96
; NONEON-NOSVE-NEXT:    .cfi_def_cfa_offset 96
; NONEON-NOSVE-NEXT:    ldp q3, q0, [x1]
; NONEON-NOSVE-NEXT:    ldp q2, q1, [x0]
; NONEON-NOSVE-NEXT:    stp q2, q3, [sp]
; NONEON-NOSVE-NEXT:    stp q1, q0, [sp, #32]
; NONEON-NOSVE-NEXT:    ldp s1, s2, [sp, #40]
; NONEON-NOSVE-NEXT:    ldr s0, [sp, #60]
; NONEON-NOSVE-NEXT:    fadd s3, s2, s0
; NONEON-NOSVE-NEXT:    ldr s0, [sp, #56]
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldp s1, s2, [sp, #32]
; NONEON-NOSVE-NEXT:    stp s0, s3, [sp, #88]
; NONEON-NOSVE-NEXT:    ldr s0, [sp, #52]
; NONEON-NOSVE-NEXT:    fadd s3, s2, s0
; NONEON-NOSVE-NEXT:    ldr s0, [sp, #48]
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldp s1, s2, [sp, #8]
; NONEON-NOSVE-NEXT:    stp s0, s3, [sp, #80]
; NONEON-NOSVE-NEXT:    ldr s0, [sp, #28]
; NONEON-NOSVE-NEXT:    fadd s3, s2, s0
; NONEON-NOSVE-NEXT:    ldr s0, [sp, #24]
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    ldp s1, s2, [sp]
; NONEON-NOSVE-NEXT:    stp s0, s3, [sp, #72]
; NONEON-NOSVE-NEXT:    ldr s0, [sp, #20]
; NONEON-NOSVE-NEXT:    fadd s3, s2, s0
; NONEON-NOSVE-NEXT:    ldr s0, [sp, #16]
; NONEON-NOSVE-NEXT:    fadd s0, s1, s0
; NONEON-NOSVE-NEXT:    stp s0, s3, [sp, #64]
; NONEON-NOSVE-NEXT:    ldp q0, q1, [sp, #64]
; NONEON-NOSVE-NEXT:    stp q0, q1, [x0]
; NONEON-NOSVE-NEXT:    add sp, sp, #96
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <8 x float>, ptr %a
  %op2 = load <8 x float>, ptr %b
  %res = fadd <8 x float> %op1, %op2
  store <8 x float> %res, ptr %a
  ret void
}

define void @fadd_v2f64(ptr %a, ptr %b) {
; CHECK-LABEL: fadd_v2f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ptrue p0.d, vl2
; CHECK-NEXT:    ldr q0, [x0]
; CHECK-NEXT:    ldr q1, [x1]
; CHECK-NEXT:    fadd z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    str q0, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: fadd_v2f64:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    ldr q0, [x1]
; NONEON-NOSVE-NEXT:    ldr q1, [x0]
; NONEON-NOSVE-NEXT:    stp q1, q0, [sp, #-48]!
; NONEON-NOSVE-NEXT:    .cfi_def_cfa_offset 48
; NONEON-NOSVE-NEXT:    ldp d1, d2, [sp]
; NONEON-NOSVE-NEXT:    ldr d0, [sp, #24]
; NONEON-NOSVE-NEXT:    fadd d3, d2, d0
; NONEON-NOSVE-NEXT:    ldr d0, [sp, #16]
; NONEON-NOSVE-NEXT:    fadd d0, d1, d0
; NONEON-NOSVE-NEXT:    stp d0, d3, [sp, #32]
; NONEON-NOSVE-NEXT:    ldr q0, [sp, #32]
; NONEON-NOSVE-NEXT:    str q0, [x0]
; NONEON-NOSVE-NEXT:    add sp, sp, #48
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <2 x double>, ptr %a
  %op2 = load <2 x double>, ptr %b
  %res = fadd <2 x double> %op1, %op2
  store <2 x double> %res, ptr %a
  ret void
}

define void @fadd_v4f64(ptr %a, ptr %b) {
; CHECK-LABEL: fadd_v4f64:
; CHECK:       // %bb.0:
; CHECK-NEXT:    ldp q0, q3, [x1]
; CHECK-NEXT:    ptrue p0.d, vl2
; CHECK-NEXT:    ldp q1, q2, [x0]
; CHECK-NEXT:    fadd z0.d, p0/m, z0.d, z1.d
; CHECK-NEXT:    movprfx z1, z2
; CHECK-NEXT:    fadd z1.d, p0/m, z1.d, z3.d
; CHECK-NEXT:    stp q0, q1, [x0]
; CHECK-NEXT:    ret
;
; NONEON-NOSVE-LABEL: fadd_v4f64:
; NONEON-NOSVE:       // %bb.0:
; NONEON-NOSVE-NEXT:    sub sp, sp, #96
; NONEON-NOSVE-NEXT:    .cfi_def_cfa_offset 96
; NONEON-NOSVE-NEXT:    ldp q3, q0, [x1]
; NONEON-NOSVE-NEXT:    ldp q2, q1, [x0]
; NONEON-NOSVE-NEXT:    stp q2, q3, [sp]
; NONEON-NOSVE-NEXT:    stp q1, q0, [sp, #32]
; NONEON-NOSVE-NEXT:    ldp d1, d2, [sp, #32]
; NONEON-NOSVE-NEXT:    ldr d0, [sp, #56]
; NONEON-NOSVE-NEXT:    fadd d3, d2, d0
; NONEON-NOSVE-NEXT:    ldr d0, [sp, #48]
; NONEON-NOSVE-NEXT:    fadd d0, d1, d0
; NONEON-NOSVE-NEXT:    ldp d1, d2, [sp]
; NONEON-NOSVE-NEXT:    stp d0, d3, [sp, #80]
; NONEON-NOSVE-NEXT:    ldr d0, [sp, #24]
; NONEON-NOSVE-NEXT:    fadd d3, d2, d0
; NONEON-NOSVE-NEXT:    ldr d0, [sp, #16]
; NONEON-NOSVE-NEXT:    fadd d0, d1, d0
; NONEON-NOSVE-NEXT:    stp d0, d3, [sp, #64]
; NONEON-NOSVE-NEXT:    ldp q0, q1, [sp, #64]
; NONEON-NOSVE-NEXT:    stp q0, q1, [x0]
; NONEON-NOSVE-NEXT:    add sp, sp, #96
; NONEON-NOSVE-NEXT:    ret
  %op1 = load <4 x double>, ptr %a
  %op2 = load <4 x double>, ptr %b
  %res = fadd <4 x double> %op1, %op2
  store <4 x double> %res, ptr %a
  ret void
}

declare <2 x i32> @llvm.abs.v2i32(<2 x i32>, i1)
declare <4 x i32> @llvm.abs.v4i32(<4 x i32>, i1)
declare <8 x i32> @llvm.abs.v8i32(<8 x i32>, i1)
declare <2 x i64> @llvm.abs.v2i64(<2 x i64>, i1)
declare <4 x i64> @llvm.abs.v4i64(<4 x i64>, i1)