; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+sse3 | FileCheck %s --check-prefix=SSE
; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx | FileCheck %s --check-prefix=AVX --check-prefix=AVX1
; RUN: llc < %s -mtriple=x86_64-unknown-linux-gnu -mattr=+avx512f | FileCheck %s --check-prefix=AVX --check-prefix=AVX512

; Test ADDSUB ISel patterns.

; Functions below are obtained from the following source:
;
; typedef double double2 __attribute__((ext_vector_type(2)));
; typedef double double4 __attribute__((ext_vector_type(4)));
; typedef float float4 __attribute__((ext_vector_type(4)));
; typedef float float8 __attribute__((ext_vector_type(8)));
;
; float4 test1(float4 A, float4 B) {
;   float4 X = A - B;
;   float4 Y = A + B;
;   return (float4){X[0], Y[1], X[2], Y[3]};
; }
;
; float8 test2(float8 A, float8 B) {
;   float8 X = A - B;
;   float8 Y = A + B;
;   return (float8){X[0], Y[1], X[2], Y[3], X[4], Y[5], X[6], Y[7]};
; }
;
; double4 test3(double4 A, double4 B) {
;   double4 X = A - B;
;   double4 Y = A + B;
;   return (double4){X[0], Y[1], X[2], Y[3]};
; }
;
; double2 test4(double2 A, double2 B) {
;   double2 X = A - B;
;   double2 Y = A + B;
;   return (double2){X[0], Y[1]};
; }

define <4 x float> @test1(<4 x float> %A, <4 x float> %B) {
; SSE-LABEL: test1:
; SSE:       # BB#0:
; SSE-NEXT:    addsubps %xmm1, %xmm0
; SSE-NEXT:    retq
;
; AVX-LABEL: test1:
; AVX:       # BB#0:
; AVX-NEXT:    vaddsubps %xmm1, %xmm0, %xmm0
; AVX-NEXT:    retq
  %sub = fsub <4 x float> %A, %B
  %add = fadd <4 x float> %A, %B
  %vecinit6 = shufflevector <4 x float> %sub, <4 x float> %add, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
  ret <4 x float> %vecinit6
}

define <8 x float> @test2(<8 x float> %A, <8 x float> %B) {
; SSE-LABEL: test2:
; SSE:       # BB#0:
; SSE-NEXT:    addsubps %xmm2, %xmm0
; SSE-NEXT:    addsubps %xmm3, %xmm1
; SSE-NEXT:    retq
;
; AVX-LABEL: test2:
; AVX:       # BB#0:
; AVX-NEXT:    vaddsubps %ymm1, %ymm0, %ymm0
; AVX-NEXT:    retq
  %sub = fsub <8 x float> %A, %B
  %add = fadd <8 x float> %A, %B
  %vecinit14 = shufflevector <8 x float> %sub, <8 x float> %add, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, i32 13, i32 6, i32 15>
  ret <8 x float> %vecinit14
}

define <4 x double> @test3(<4 x double> %A, <4 x double> %B) {
; SSE-LABEL: test3:
; SSE:       # BB#0:
; SSE-NEXT:    addsubpd %xmm2, %xmm0
; SSE-NEXT:    addsubpd %xmm3, %xmm1
; SSE-NEXT:    retq
;
; AVX-LABEL: test3:
; AVX:       # BB#0:
; AVX-NEXT:    vaddsubpd %ymm1, %ymm0, %ymm0
; AVX-NEXT:    retq
  %sub = fsub <4 x double> %A, %B
  %add = fadd <4 x double> %A, %B
  %vecinit6 = shufflevector <4 x double> %sub, <4 x double> %add, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
  ret <4 x double> %vecinit6
}

define <2 x double> @test4(<2 x double> %A, <2 x double> %B) #0 {
; SSE-LABEL: test4:
; SSE:       # BB#0:
; SSE-NEXT:    addsubpd %xmm1, %xmm0
; SSE-NEXT:    retq
;
; AVX-LABEL: test4:
; AVX:       # BB#0:
; AVX-NEXT:    vaddsubpd %xmm1, %xmm0, %xmm0
; AVX-NEXT:    retq
  %add = fadd <2 x double> %A, %B
  %sub = fsub <2 x double> %A, %B
  %vecinit2 = shufflevector <2 x double> %sub, <2 x double> %add, <2 x i32> <i32 0, i32 3>
  ret <2 x double> %vecinit2
}

define <16 x float> @test5(<16 x float> %A, <16 x float> %B) {
; SSE-LABEL: test5:
; SSE:       # BB#0:
; SSE-NEXT:    addsubps %xmm4, %xmm0
; SSE-NEXT:    addsubps %xmm5, %xmm1
; SSE-NEXT:    addsubps %xmm6, %xmm2
; SSE-NEXT:    addsubps %xmm7, %xmm3
; SSE-NEXT:    retq
;
; AVX1-LABEL: test5:
; AVX1:       # BB#0:
; AVX1-NEXT:    vaddsubps %ymm2, %ymm0, %ymm0
; AVX1-NEXT:    vaddsubps %ymm3, %ymm1, %ymm1
; AVX1-NEXT:    retq
;
; AVX512-LABEL: test5:
; AVX512:       # BB#0:
; AVX512-NEXT:    vsubps %zmm1, %zmm0, %zmm2
; AVX512-NEXT:    movw $-21846, %ax # imm = 0xAAAA
; AVX512-NEXT:    kmovw %eax, %k1
; AVX512-NEXT:    vaddps %zmm1, %zmm0, %zmm2 {%k1}
; AVX512-NEXT:    vmovaps %zmm2, %zmm0
; AVX512-NEXT:    retq
  %add = fadd <16 x float> %A, %B
  %sub = fsub <16 x float> %A, %B
  %vecinit2 = shufflevector <16 x float> %sub, <16 x float> %add, <16 x i32> <i32 0, i32 17, i32 2, i32 19, i32 4, i32 21, i32 6, i32 23, i32 8, i32 25, i32 10, i32 27, i32 12, i32 29, i32 14, i32 31>
  ret <16 x float> %vecinit2
}

define <8 x double> @test6(<8 x double> %A, <8 x double> %B) {
; SSE-LABEL: test6:
; SSE:       # BB#0:
; SSE-NEXT:    addsubpd %xmm4, %xmm0
; SSE-NEXT:    addsubpd %xmm5, %xmm1
; SSE-NEXT:    addsubpd %xmm6, %xmm2
; SSE-NEXT:    addsubpd %xmm7, %xmm3
; SSE-NEXT:    retq
;
; AVX1-LABEL: test6:
; AVX1:       # BB#0:
; AVX1-NEXT:    vaddsubpd %ymm2, %ymm0, %ymm0
; AVX1-NEXT:    vaddsubpd %ymm3, %ymm1, %ymm1
; AVX1-NEXT:    retq
;
; AVX512-LABEL: test6:
; AVX512:       # BB#0:
; AVX512-NEXT:    vaddpd %zmm1, %zmm0, %zmm2
; AVX512-NEXT:    vsubpd %zmm1, %zmm0, %zmm0
; AVX512-NEXT:    vshufpd {{.*#+}} zmm0 = zmm0[0],zmm2[1],zmm0[2],zmm2[3],zmm0[4],zmm2[5],zmm0[6],zmm2[7]
; AVX512-NEXT:    retq
  %add = fadd <8 x double> %A, %B
  %sub = fsub <8 x double> %A, %B
  %vecinit2 = shufflevector <8 x double> %sub, <8 x double> %add, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, i32 13, i32 6, i32 15>
  ret <8 x double> %vecinit2
}

define <4 x float> @test1b(<4 x float> %A, <4 x float>* %B) {
; SSE-LABEL: test1b:
; SSE:       # BB#0:
; SSE-NEXT:    addsubps (%rdi), %xmm0
; SSE-NEXT:    retq
;
; AVX-LABEL: test1b:
; AVX:       # BB#0:
; AVX-NEXT:    vaddsubps (%rdi), %xmm0, %xmm0
; AVX-NEXT:    retq
  %1 = load <4 x float>, <4 x float>* %B
  %add = fadd <4 x float> %A, %1
  %sub = fsub <4 x float> %A, %1
  %vecinit6 = shufflevector <4 x float> %sub, <4 x float> %add, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
  ret <4 x float> %vecinit6
}

define <8 x float> @test2b(<8 x float> %A, <8 x float>* %B) {
; SSE-LABEL: test2b:
; SSE:       # BB#0:
; SSE-NEXT:    addsubps (%rdi), %xmm0
; SSE-NEXT:    addsubps 16(%rdi), %xmm1
; SSE-NEXT:    retq
;
; AVX-LABEL: test2b:
; AVX:       # BB#0:
; AVX-NEXT:    vaddsubps (%rdi), %ymm0, %ymm0
; AVX-NEXT:    retq
  %1 = load <8 x float>, <8 x float>* %B
  %add = fadd <8 x float> %A, %1
  %sub = fsub <8 x float> %A, %1
  %vecinit14 = shufflevector <8 x float> %sub, <8 x float> %add, <8 x i32> <i32 0, i32 9, i32 2, i32 11, i32 4, i32 13, i32 6, i32 15>
  ret <8 x float> %vecinit14
}

define <4 x double> @test3b(<4 x double> %A, <4 x double>* %B) {
; SSE-LABEL: test3b:
; SSE:       # BB#0:
; SSE-NEXT:    addsubpd (%rdi), %xmm0
; SSE-NEXT:    addsubpd 16(%rdi), %xmm1
; SSE-NEXT:    retq
;
; AVX-LABEL: test3b:
; AVX:       # BB#0:
; AVX-NEXT:    vaddsubpd (%rdi), %ymm0, %ymm0
; AVX-NEXT:    retq
  %1 = load <4 x double>, <4 x double>* %B
  %add = fadd <4 x double> %A, %1
  %sub = fsub <4 x double> %A, %1
  %vecinit6 = shufflevector <4 x double> %sub, <4 x double> %add, <4 x i32> <i32 0, i32 5, i32 2, i32 7>
  ret <4 x double> %vecinit6
}

define <2 x double> @test4b(<2 x double> %A, <2 x double>* %B) {
; SSE-LABEL: test4b:
; SSE:       # BB#0:
; SSE-NEXT:    addsubpd (%rdi), %xmm0
; SSE-NEXT:    retq
;
; AVX-LABEL: test4b:
; AVX:       # BB#0:
; AVX-NEXT:    vaddsubpd (%rdi), %xmm0, %xmm0
; AVX-NEXT:    retq
  %1 = load <2 x double>, <2 x double>* %B
  %sub = fsub <2 x double> %A, %1
  %add = fadd <2 x double> %A, %1
  %vecinit2 = shufflevector <2 x double> %sub, <2 x double> %add, <2 x i32> <i32 0, i32 3>
  ret <2 x double> %vecinit2
}

define <4 x float> @test1c(<4 x float> %A, <4 x float>* %B) {
; SSE-LABEL: test1c:
; SSE:       # BB#0:
; SSE-NEXT:    addsubps (%rdi), %xmm0
; SSE-NEXT:    retq
;
; AVX-LABEL: test1c:
; AVX:       # BB#0:
; AVX-NEXT:    vaddsubps (%rdi), %xmm0, %xmm0
; AVX-NEXT:    retq
  %1 = load <4 x float>, <4 x float>* %B
  %add = fadd <4 x float> %A, %1
  %sub = fsub <4 x float> %A, %1
  %vecinit6 = shufflevector <4 x float> %add, <4 x float> %sub, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
  ret <4 x float> %vecinit6
}

define <8 x float> @test2c(<8 x float> %A, <8 x float>* %B) {
; SSE-LABEL: test2c:
; SSE:       # BB#0:
; SSE-NEXT:    addsubps (%rdi), %xmm0
; SSE-NEXT:    addsubps 16(%rdi), %xmm1
; SSE-NEXT:    retq
;
; AVX-LABEL: test2c:
; AVX:       # BB#0:
; AVX-NEXT:    vaddsubps (%rdi), %ymm0, %ymm0
; AVX-NEXT:    retq
  %1 = load <8 x float>, <8 x float>* %B
  %add = fadd <8 x float> %A, %1
  %sub = fsub <8 x float> %A, %1
  %vecinit14 = shufflevector <8 x float> %add, <8 x float> %sub, <8 x i32> <i32 8, i32 1, i32 10, i32 3, i32 12, i32 5, i32 14, i32 7>
  ret <8 x float> %vecinit14
}

define <4 x double> @test3c(<4 x double> %A, <4 x double>* %B) {
; SSE-LABEL: test3c:
; SSE:       # BB#0:
; SSE-NEXT:    addsubpd (%rdi), %xmm0
; SSE-NEXT:    addsubpd 16(%rdi), %xmm1
; SSE-NEXT:    retq
;
; AVX-LABEL: test3c:
; AVX:       # BB#0:
; AVX-NEXT:    vaddsubpd (%rdi), %ymm0, %ymm0
; AVX-NEXT:    retq
  %1 = load <4 x double>, <4 x double>* %B
  %add = fadd <4 x double> %A, %1
  %sub = fsub <4 x double> %A, %1
  %vecinit6 = shufflevector <4 x double> %add, <4 x double> %sub, <4 x i32> <i32 4, i32 1, i32 6, i32 3>
  ret <4 x double> %vecinit6
}

define <2 x double> @test4c(<2 x double> %A, <2 x double>* %B) {
; SSE-LABEL: test4c:
; SSE:       # BB#0:
; SSE-NEXT:    addsubpd (%rdi), %xmm0
; SSE-NEXT:    retq
;
; AVX-LABEL: test4c:
; AVX:       # BB#0:
; AVX-NEXT:    vaddsubpd (%rdi), %xmm0, %xmm0
; AVX-NEXT:    retq
  %1 = load <2 x double>, <2 x double>* %B
  %sub = fsub <2 x double> %A, %1
  %add = fadd <2 x double> %A, %1
  %vecinit2 = shufflevector <2 x double> %add, <2 x double> %sub, <2 x i32> <i32 2, i32 1>
  ret <2 x double> %vecinit2
}