llvm/test/CodeGen/X86/avx512vl-arith.ll

; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=knl -mattr=+avx512vl --show-mc-encoding| FileCheck %s
; RUN: llc < %s -mtriple=x86_64-apple-darwin -mcpu=knl -mattr=+avx512vl,-evex512 --show-mc-encoding| FileCheck %s

; 256-bit

define <4 x i64> @vpaddq256_test(<4 x i64> %i, <4 x i64> %j) nounwind readnone {
; CHECK-LABEL: vpaddq256_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vpaddq %ymm1, %ymm0, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xfd,0xd4,0xc1]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %x = add <4 x i64> %i, %j
  ret <4 x i64> %x
}

define <4 x i64> @vpaddq256_fold_test(<4 x i64> %i, ptr %j) nounwind {
; CHECK-LABEL: vpaddq256_fold_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vpaddq (%rdi), %ymm0, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xfd,0xd4,0x07]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %tmp = load <4 x i64>, ptr %j, align 4
  %x = add <4 x i64> %i, %tmp
  ret <4 x i64> %x
}

define <4 x i64> @vpaddq256_broadcast_test(<4 x i64> %i) nounwind {
; CHECK-LABEL: vpaddq256_broadcast_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vpaddq {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %ymm0, %ymm0 ## encoding: [0x62,0xf1,0xfd,0x38,0xd4,0x05,A,A,A,A]
; CHECK-NEXT:    ## fixup A - offset: 6, value: {{\.?LCPI[0-9]+_[0-9]+}}-4, kind: reloc_riprel_4byte
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %x = add <4 x i64> %i, <i64 2, i64 2, i64 2, i64 2>
  ret <4 x i64> %x
}

define <4 x i64> @vpaddq256_broadcast2_test(<4 x i64> %i, ptr %j.ptr) nounwind {
; CHECK-LABEL: vpaddq256_broadcast2_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vpaddq (%rdi){1to4}, %ymm0, %ymm0 ## encoding: [0x62,0xf1,0xfd,0x38,0xd4,0x07]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %j = load i64, ptr %j.ptr
  %j.0 = insertelement <4 x i64> undef, i64 %j, i32 0
  %j.v = shufflevector <4 x i64> %j.0, <4 x i64> undef, <4 x i32> zeroinitializer
  %x = add <4 x i64> %i, %j.v
  ret <4 x i64> %x
}

define <8 x i32> @vpaddd256_test(<8 x i32> %i, <8 x i32> %j) nounwind readnone {
; CHECK-LABEL: vpaddd256_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vpaddd %ymm1, %ymm0, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xfd,0xfe,0xc1]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %x = add <8 x i32> %i, %j
  ret <8 x i32> %x
}

define <8 x i32> @vpaddd256_fold_test(<8 x i32> %i, ptr %j) nounwind {
; CHECK-LABEL: vpaddd256_fold_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vpaddd (%rdi), %ymm0, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xfd,0xfe,0x07]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %tmp = load <8 x i32>, ptr %j, align 4
  %x = add <8 x i32> %i, %tmp
  ret <8 x i32> %x
}

define <8 x i32> @vpaddd256_broadcast_test(<8 x i32> %i) nounwind {
; CHECK-LABEL: vpaddd256_broadcast_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vpaddd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to8}, %ymm0, %ymm0 ## encoding: [0x62,0xf1,0x7d,0x38,0xfe,0x05,A,A,A,A]
; CHECK-NEXT:    ## fixup A - offset: 6, value: {{\.?LCPI[0-9]+_[0-9]+}}-4, kind: reloc_riprel_4byte
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %x = add <8 x i32> %i, <i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3, i32 3>
  ret <8 x i32> %x
}

define <8 x i32> @vpaddd256_mask_test(<8 x i32> %i, <8 x i32> %j, <8 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: vpaddd256_mask_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %ymm2, %ymm2, %k1 ## encoding: [0x62,0xf2,0x6d,0x28,0x27,0xca]
; CHECK-NEXT:    vpaddd %ymm1, %ymm0, %ymm0 {%k1} ## encoding: [0x62,0xf1,0x7d,0x29,0xfe,0xc1]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <8 x i32> %mask1, zeroinitializer
  %x = add <8 x i32> %i, %j
  %r = select <8 x i1> %mask, <8 x i32> %x, <8 x i32> %i
  ret <8 x i32> %r
}

define <8 x i32> @vpaddd256_maskz_test(<8 x i32> %i, <8 x i32> %j, <8 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: vpaddd256_maskz_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %ymm2, %ymm2, %k1 ## encoding: [0x62,0xf2,0x6d,0x28,0x27,0xca]
; CHECK-NEXT:    vpaddd %ymm1, %ymm0, %ymm0 {%k1} {z} ## encoding: [0x62,0xf1,0x7d,0xa9,0xfe,0xc1]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <8 x i32> %mask1, zeroinitializer
  %x = add <8 x i32> %i, %j
  %r = select <8 x i1> %mask, <8 x i32> %x, <8 x i32> zeroinitializer
  ret <8 x i32> %r
}

define <8 x i32> @vpaddd256_mask_fold_test(<8 x i32> %i, ptr %j.ptr, <8 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: vpaddd256_mask_fold_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %ymm1, %ymm1, %k1 ## encoding: [0x62,0xf2,0x75,0x28,0x27,0xc9]
; CHECK-NEXT:    vpaddd (%rdi), %ymm0, %ymm0 {%k1} ## encoding: [0x62,0xf1,0x7d,0x29,0xfe,0x07]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <8 x i32> %mask1, zeroinitializer
  %j = load <8 x i32>, ptr %j.ptr
  %x = add <8 x i32> %i, %j
  %r = select <8 x i1> %mask, <8 x i32> %x, <8 x i32> %i
  ret <8 x i32> %r
}

define <8 x i32> @vpaddd256_mask_broadcast_test(<8 x i32> %i, <8 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: vpaddd256_mask_broadcast_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %ymm1, %ymm1, %k1 ## encoding: [0x62,0xf2,0x75,0x28,0x27,0xc9]
; CHECK-NEXT:    vpaddd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to8}, %ymm0, %ymm0 {%k1} ## encoding: [0x62,0xf1,0x7d,0x39,0xfe,0x05,A,A,A,A]
; CHECK-NEXT:    ## fixup A - offset: 6, value: {{\.?LCPI[0-9]+_[0-9]+}}-4, kind: reloc_riprel_4byte
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <8 x i32> %mask1, zeroinitializer
  %x = add <8 x i32> %i, <i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4, i32 4>
  %r = select <8 x i1> %mask, <8 x i32> %x, <8 x i32> %i
  ret <8 x i32> %r
}

define <8 x i32> @vpaddd256_maskz_fold_test(<8 x i32> %i, ptr %j.ptr, <8 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: vpaddd256_maskz_fold_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %ymm1, %ymm1, %k1 ## encoding: [0x62,0xf2,0x75,0x28,0x27,0xc9]
; CHECK-NEXT:    vpaddd (%rdi), %ymm0, %ymm0 {%k1} {z} ## encoding: [0x62,0xf1,0x7d,0xa9,0xfe,0x07]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <8 x i32> %mask1, zeroinitializer
  %j = load <8 x i32>, ptr %j.ptr
  %x = add <8 x i32> %i, %j
  %r = select <8 x i1> %mask, <8 x i32> %x, <8 x i32> zeroinitializer
  ret <8 x i32> %r
}

define <8 x i32> @vpaddd256_maskz_broadcast_test(<8 x i32> %i, <8 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: vpaddd256_maskz_broadcast_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %ymm1, %ymm1, %k1 ## encoding: [0x62,0xf2,0x75,0x28,0x27,0xc9]
; CHECK-NEXT:    vpaddd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to8}, %ymm0, %ymm0 {%k1} {z} ## encoding: [0x62,0xf1,0x7d,0xb9,0xfe,0x05,A,A,A,A]
; CHECK-NEXT:    ## fixup A - offset: 6, value: {{\.?LCPI[0-9]+_[0-9]+}}-4, kind: reloc_riprel_4byte
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <8 x i32> %mask1, zeroinitializer
  %x = add <8 x i32> %i, <i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5, i32 5>
  %r = select <8 x i1> %mask, <8 x i32> %x, <8 x i32> zeroinitializer
  ret <8 x i32> %r
}

define <4 x i64> @vpsubq256_test(<4 x i64> %i, <4 x i64> %j) nounwind readnone {
; CHECK-LABEL: vpsubq256_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vpsubq %ymm1, %ymm0, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xfd,0xfb,0xc1]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %x = sub <4 x i64> %i, %j
  ret <4 x i64> %x
}

define <8 x i32> @vpsubd256_test(<8 x i32> %i, <8 x i32> %j) nounwind readnone {
; CHECK-LABEL: vpsubd256_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vpsubd %ymm1, %ymm0, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xfd,0xfa,0xc1]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %x = sub <8 x i32> %i, %j
  ret <8 x i32> %x
}

define <8 x i32> @vpmulld256_test(<8 x i32> %i, <8 x i32> %j) {
; CHECK-LABEL: vpmulld256_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vpmulld %ymm1, %ymm0, %ymm0 ## EVEX TO VEX Compression encoding: [0xc4,0xe2,0x7d,0x40,0xc1]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %x = mul <8 x i32> %i, %j
  ret <8 x i32> %x
}

define <4 x double> @test_vaddpd_256(<4 x double> %y, <4 x double> %x) {
; CHECK-LABEL: test_vaddpd_256:
; CHECK:       ## %bb.0: ## %entry
; CHECK-NEXT:    vaddpd %ymm0, %ymm1, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf5,0x58,0xc0]
; CHECK-NEXT:    retq ## encoding: [0xc3]
entry:
  %add.i = fadd <4 x double> %x, %y
  ret <4 x double> %add.i
}

define <4 x double> @test_fold_vaddpd_256(<4 x double> %y) {
; CHECK-LABEL: test_fold_vaddpd_256:
; CHECK:       ## %bb.0: ## %entry
; CHECK-NEXT:    vaddpd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %ymm0, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xfd,0x58,0x05,A,A,A,A]
; CHECK-NEXT:    ## fixup A - offset: 4, value: {{\.?LCPI[0-9]+_[0-9]+}}-4, kind: reloc_riprel_4byte
; CHECK-NEXT:    retq ## encoding: [0xc3]
entry:
  %add.i = fadd <4 x double> %y, <double 4.500000e+00, double 3.400000e+00, double 4.500000e+00, double 5.600000e+00>
  ret <4 x double> %add.i
}

define <8 x float> @test_broadcast_vaddpd_256(<8 x float> %a) nounwind {
; CHECK-LABEL: test_broadcast_vaddpd_256:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vaddps {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to8}, %ymm0, %ymm0 ## encoding: [0x62,0xf1,0x7c,0x38,0x58,0x05,A,A,A,A]
; CHECK-NEXT:    ## fixup A - offset: 6, value: {{\.?LCPI[0-9]+_[0-9]+}}-4, kind: reloc_riprel_4byte
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %b = fadd <8 x float> %a, <float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000>
  ret <8 x float> %b
}

define <8 x float> @test_mask_vaddps_256(<8 x float> %dst, <8 x float> %i, <8 x float> %j, <8 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vaddps_256:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %ymm3, %ymm3, %k1 ## encoding: [0x62,0xf2,0x65,0x28,0x27,0xcb]
; CHECK-NEXT:    vaddps %ymm2, %ymm1, %ymm0 {%k1} ## encoding: [0x62,0xf1,0x74,0x29,0x58,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <8 x i32> %mask1, zeroinitializer
  %x = fadd <8 x float> %i, %j
  %r = select <8 x i1> %mask, <8 x float> %x, <8 x float> %dst
  ret <8 x float> %r
}

define <8 x float> @test_mask_vmulps_256(<8 x float> %dst, <8 x float> %i, <8 x float> %j, <8 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vmulps_256:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %ymm3, %ymm3, %k1 ## encoding: [0x62,0xf2,0x65,0x28,0x27,0xcb]
; CHECK-NEXT:    vmulps %ymm2, %ymm1, %ymm0 {%k1} ## encoding: [0x62,0xf1,0x74,0x29,0x59,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <8 x i32> %mask1, zeroinitializer
  %x = fmul <8 x float> %i, %j
  %r = select <8 x i1> %mask, <8 x float> %x, <8 x float> %dst
  ret <8 x float> %r
}

define <8 x float> @test_mask_vminps_256(<8 x float> %dst, <8 x float> %i, <8 x float> %j, <8 x i32> %mask1)nounwind readnone {
; CHECK-LABEL: test_mask_vminps_256:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %ymm3, %ymm3, %k1 ## encoding: [0x62,0xf2,0x65,0x28,0x27,0xcb]
; CHECK-NEXT:    vminps %ymm2, %ymm1, %ymm0 {%k1} ## encoding: [0x62,0xf1,0x74,0x29,0x5d,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <8 x i32> %mask1, zeroinitializer
  %cmp_res = fcmp olt <8 x float> %i, %j
  %min = select <8 x i1> %cmp_res, <8 x float> %i, <8 x float> %j
  %r = select <8 x i1> %mask, <8 x float> %min, <8 x float> %dst
  ret <8 x float> %r
}

define <8 x float> @test_mask_vmaxps_256(<8 x float> %dst, <8 x float> %i, <8 x float> %j, <8 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vmaxps_256:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %ymm3, %ymm3, %k1 ## encoding: [0x62,0xf2,0x65,0x28,0x27,0xcb]
; CHECK-NEXT:    vmaxps %ymm2, %ymm1, %ymm0 {%k1} ## encoding: [0x62,0xf1,0x74,0x29,0x5f,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <8 x i32> %mask1, zeroinitializer
  %cmp_res = fcmp ogt <8 x float> %i, %j
  %max = select <8 x i1> %cmp_res, <8 x float> %i, <8 x float> %j
  %r = select <8 x i1> %mask, <8 x float> %max, <8 x float> %dst
  ret <8 x float> %r
}

define <8 x float> @test_mask_vsubps_256(<8 x float> %dst, <8 x float> %i, <8 x float> %j, <8 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vsubps_256:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %ymm3, %ymm3, %k1 ## encoding: [0x62,0xf2,0x65,0x28,0x27,0xcb]
; CHECK-NEXT:    vsubps %ymm2, %ymm1, %ymm0 {%k1} ## encoding: [0x62,0xf1,0x74,0x29,0x5c,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <8 x i32> %mask1, zeroinitializer
  %x = fsub <8 x float> %i, %j
  %r = select <8 x i1> %mask, <8 x float> %x, <8 x float> %dst
  ret <8 x float> %r
}

define <8 x float> @test_mask_vdivps_256(<8 x float> %dst, <8 x float> %i, <8 x float> %j, <8 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vdivps_256:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %ymm3, %ymm3, %k1 ## encoding: [0x62,0xf2,0x65,0x28,0x27,0xcb]
; CHECK-NEXT:    vdivps %ymm2, %ymm1, %ymm0 {%k1} ## encoding: [0x62,0xf1,0x74,0x29,0x5e,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <8 x i32> %mask1, zeroinitializer
  %x = fdiv <8 x float> %i, %j
  %r = select <8 x i1> %mask, <8 x float> %x, <8 x float> %dst
  ret <8 x float> %r
}

define <4 x double> @test_mask_vmulpd_256(<4 x double> %dst, <4 x double> %i, <4 x double> %j, <4 x i64> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vmulpd_256:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmq %ymm3, %ymm3, %k1 ## encoding: [0x62,0xf2,0xe5,0x28,0x27,0xcb]
; CHECK-NEXT:    vmulpd %ymm2, %ymm1, %ymm0 {%k1} ## encoding: [0x62,0xf1,0xf5,0x29,0x59,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i64> %mask1, zeroinitializer
  %x = fmul <4 x double> %i, %j
  %r = select <4 x i1> %mask, <4 x double> %x, <4 x double> %dst
  ret <4 x double> %r
}

define <4 x double> @test_mask_vminpd_256(<4 x double> %dst, <4 x double> %i, <4 x double> %j, <4 x i64> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vminpd_256:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmq %ymm3, %ymm3, %k1 ## encoding: [0x62,0xf2,0xe5,0x28,0x27,0xcb]
; CHECK-NEXT:    vminpd %ymm2, %ymm1, %ymm0 {%k1} ## encoding: [0x62,0xf1,0xf5,0x29,0x5d,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i64> %mask1, zeroinitializer
  %cmp_res = fcmp olt <4 x double> %i, %j
  %min = select <4 x i1> %cmp_res, <4 x double> %i, <4 x double> %j
  %r = select <4 x i1> %mask, <4 x double> %min, <4 x double> %dst
  ret <4 x double> %r
}

define <4 x double> @test_mask_vmaxpd_256(<4 x double> %dst, <4 x double> %i, <4 x double> %j, <4 x i64> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vmaxpd_256:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmq %ymm3, %ymm3, %k1 ## encoding: [0x62,0xf2,0xe5,0x28,0x27,0xcb]
; CHECK-NEXT:    vmaxpd %ymm2, %ymm1, %ymm0 {%k1} ## encoding: [0x62,0xf1,0xf5,0x29,0x5f,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i64> %mask1, zeroinitializer
  %cmp_res = fcmp ogt <4 x double> %i, %j
  %max = select <4 x i1> %cmp_res, <4 x double> %i, <4 x double> %j
  %r = select <4 x i1> %mask, <4 x double> %max, <4 x double> %dst
  ret <4 x double> %r
}

define <4 x double> @test_mask_vsubpd_256(<4 x double> %dst, <4 x double> %i, <4 x double> %j, <4 x i64> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vsubpd_256:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmq %ymm3, %ymm3, %k1 ## encoding: [0x62,0xf2,0xe5,0x28,0x27,0xcb]
; CHECK-NEXT:    vsubpd %ymm2, %ymm1, %ymm0 {%k1} ## encoding: [0x62,0xf1,0xf5,0x29,0x5c,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i64> %mask1, zeroinitializer
  %x = fsub <4 x double> %i, %j
  %r = select <4 x i1> %mask, <4 x double> %x, <4 x double> %dst
  ret <4 x double> %r
}

define <4 x double> @test_mask_vdivpd_256(<4 x double> %dst, <4 x double> %i, <4 x double> %j, <4 x i64> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vdivpd_256:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmq %ymm3, %ymm3, %k1 ## encoding: [0x62,0xf2,0xe5,0x28,0x27,0xcb]
; CHECK-NEXT:    vdivpd %ymm2, %ymm1, %ymm0 {%k1} ## encoding: [0x62,0xf1,0xf5,0x29,0x5e,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i64> %mask1, zeroinitializer
  %x = fdiv <4 x double> %i, %j
  %r = select <4 x i1> %mask, <4 x double> %x, <4 x double> %dst
  ret <4 x double> %r
}

define <4 x double> @test_mask_vaddpd_256(<4 x double> %dst, <4 x double> %i, <4 x double> %j, <4 x i64> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vaddpd_256:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmq %ymm3, %ymm3, %k1 ## encoding: [0x62,0xf2,0xe5,0x28,0x27,0xcb]
; CHECK-NEXT:    vaddpd %ymm2, %ymm1, %ymm0 {%k1} ## encoding: [0x62,0xf1,0xf5,0x29,0x58,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i64> %mask1, zeroinitializer
  %x = fadd <4 x double> %i, %j
  %r = select <4 x i1> %mask, <4 x double> %x, <4 x double> %dst
  ret <4 x double> %r
}

define <4 x double> @test_maskz_vaddpd_256(<4 x double> %i, <4 x double> %j, <4 x i64> %mask1) nounwind readnone {
; CHECK-LABEL: test_maskz_vaddpd_256:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmq %ymm2, %ymm2, %k1 ## encoding: [0x62,0xf2,0xed,0x28,0x27,0xca]
; CHECK-NEXT:    vaddpd %ymm1, %ymm0, %ymm0 {%k1} {z} ## encoding: [0x62,0xf1,0xfd,0xa9,0x58,0xc1]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i64> %mask1, zeroinitializer
  %x = fadd <4 x double> %i, %j
  %r = select <4 x i1> %mask, <4 x double> %x, <4 x double> zeroinitializer
  ret <4 x double> %r
}

define <4 x double> @test_mask_fold_vaddpd_256(<4 x double> %dst, <4 x double> %i, ptr %j,  <4 x i64> %mask1) nounwind {
; CHECK-LABEL: test_mask_fold_vaddpd_256:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmq %ymm2, %ymm2, %k1 ## encoding: [0x62,0xf2,0xed,0x28,0x27,0xca]
; CHECK-NEXT:    vaddpd (%rdi), %ymm1, %ymm0 {%k1} ## encoding: [0x62,0xf1,0xf5,0x29,0x58,0x07]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i64> %mask1, zeroinitializer
  %tmp = load <4 x double>, ptr %j
  %x = fadd <4 x double> %i, %tmp
  %r = select <4 x i1> %mask, <4 x double> %x, <4 x double> %dst
  ret <4 x double> %r
}

define <4 x double> @test_maskz_fold_vaddpd_256(<4 x double> %i, ptr %j, <4 x i64> %mask1) nounwind {
; CHECK-LABEL: test_maskz_fold_vaddpd_256:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmq %ymm1, %ymm1, %k1 ## encoding: [0x62,0xf2,0xf5,0x28,0x27,0xc9]
; CHECK-NEXT:    vaddpd (%rdi), %ymm0, %ymm0 {%k1} {z} ## encoding: [0x62,0xf1,0xfd,0xa9,0x58,0x07]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i64> %mask1, zeroinitializer
  %tmp = load <4 x double>, ptr %j
  %x = fadd <4 x double> %i, %tmp
  %r = select <4 x i1> %mask, <4 x double> %x, <4 x double> zeroinitializer
  ret <4 x double> %r
}

define <4 x double> @test_broadcast2_vaddpd_256(<4 x double> %i, ptr %j) nounwind {
; CHECK-LABEL: test_broadcast2_vaddpd_256:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vaddpd (%rdi){1to4}, %ymm0, %ymm0 ## encoding: [0x62,0xf1,0xfd,0x38,0x58,0x07]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %tmp = load double, ptr %j
  %b = insertelement <4 x double> undef, double %tmp, i32 0
  %c = shufflevector <4 x double> %b, <4 x double> undef, <4 x i32> zeroinitializer
  %x = fadd <4 x double> %c, %i
  ret <4 x double> %x
}

define <4 x double> @test_mask_broadcast_vaddpd_256(<4 x double> %dst, <4 x double> %i, ptr %j, <4 x i64> %mask1) nounwind {
; CHECK-LABEL: test_mask_broadcast_vaddpd_256:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vmovapd %ymm1, %ymm0 ## EVEX TO VEX Compression encoding: [0xc5,0xfd,0x28,0xc1]
; CHECK-NEXT:    vptestmq %ymm2, %ymm2, %k1 ## encoding: [0x62,0xf2,0xed,0x28,0x27,0xca]
; CHECK-NEXT:    vaddpd (%rdi){1to4}, %ymm1, %ymm0 {%k1} ## encoding: [0x62,0xf1,0xf5,0x39,0x58,0x07]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i64> %mask1, zeroinitializer
  %tmp = load double, ptr %j
  %b = insertelement <4 x double> undef, double %tmp, i32 0
  %c = shufflevector <4 x double> %b, <4 x double> undef, <4 x i32> zeroinitializer
  %x = fadd <4 x double> %c, %i
  %r = select <4 x i1> %mask, <4 x double> %x, <4 x double> %i
  ret <4 x double> %r
}

define <4 x double> @test_maskz_broadcast_vaddpd_256(<4 x double> %i, ptr %j, <4 x i64> %mask1) nounwind {
; CHECK-LABEL: test_maskz_broadcast_vaddpd_256:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmq %ymm1, %ymm1, %k1 ## encoding: [0x62,0xf2,0xf5,0x28,0x27,0xc9]
; CHECK-NEXT:    vaddpd (%rdi){1to4}, %ymm0, %ymm0 {%k1} {z} ## encoding: [0x62,0xf1,0xfd,0xb9,0x58,0x07]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i64> %mask1, zeroinitializer
  %tmp = load double, ptr %j
  %b = insertelement <4 x double> undef, double %tmp, i32 0
  %c = shufflevector <4 x double> %b, <4 x double> undef, <4 x i32> zeroinitializer
  %x = fadd <4 x double> %c, %i
  %r = select <4 x i1> %mask, <4 x double> %x, <4 x double> zeroinitializer
  ret <4 x double> %r
}

; 128-bit

define <2 x i64> @vpaddq128_test(<2 x i64> %i, <2 x i64> %j) nounwind readnone {
; CHECK-LABEL: vpaddq128_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vpaddq %xmm1, %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf9,0xd4,0xc1]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %x = add <2 x i64> %i, %j
  ret <2 x i64> %x
}

define <2 x i64> @vpaddq128_fold_test(<2 x i64> %i, ptr %j) nounwind {
; CHECK-LABEL: vpaddq128_fold_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vpaddq (%rdi), %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf9,0xd4,0x07]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %tmp = load <2 x i64>, ptr %j, align 4
  %x = add <2 x i64> %i, %tmp
  ret <2 x i64> %x
}

define <2 x i64> @vpaddq128_broadcast2_test(<2 x i64> %i, ptr %j) nounwind {
; CHECK-LABEL: vpaddq128_broadcast2_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vpaddq (%rdi){1to2}, %xmm0, %xmm0 ## encoding: [0x62,0xf1,0xfd,0x18,0xd4,0x07]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %tmp = load i64, ptr %j
  %j.0 = insertelement <2 x i64> undef, i64 %tmp, i32 0
  %j.1 = insertelement <2 x i64> %j.0, i64 %tmp, i32 1
  %x = add <2 x i64> %i, %j.1
  ret <2 x i64> %x
}

define <4 x i32> @vpaddd128_test(<4 x i32> %i, <4 x i32> %j) nounwind readnone {
; CHECK-LABEL: vpaddd128_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vpaddd %xmm1, %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf9,0xfe,0xc1]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %x = add <4 x i32> %i, %j
  ret <4 x i32> %x
}

define <4 x i32> @vpaddd128_fold_test(<4 x i32> %i, ptr %j) nounwind {
; CHECK-LABEL: vpaddd128_fold_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vpaddd (%rdi), %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf9,0xfe,0x07]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %tmp = load <4 x i32>, ptr %j, align 4
  %x = add <4 x i32> %i, %tmp
  ret <4 x i32> %x
}

define <4 x i32> @vpaddd128_broadcast_test(<4 x i32> %i) nounwind {
; CHECK-LABEL: vpaddd128_broadcast_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vpaddd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %xmm0, %xmm0 ## encoding: [0x62,0xf1,0x7d,0x18,0xfe,0x05,A,A,A,A]
; CHECK-NEXT:    ## fixup A - offset: 6, value: {{\.?LCPI[0-9]+_[0-9]+}}-4, kind: reloc_riprel_4byte
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %x = add <4 x i32> %i, <i32 6, i32 6, i32 6, i32 6>
  ret <4 x i32> %x
}

define <4 x i32> @vpaddd128_mask_test(<4 x i32> %i, <4 x i32> %j, <4 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: vpaddd128_mask_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %xmm2, %xmm2, %k1 ## encoding: [0x62,0xf2,0x6d,0x08,0x27,0xca]
; CHECK-NEXT:    vpaddd %xmm1, %xmm0, %xmm0 {%k1} ## encoding: [0x62,0xf1,0x7d,0x09,0xfe,0xc1]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i32> %mask1, zeroinitializer
  %x = add <4 x i32> %i, %j
  %r = select <4 x i1> %mask, <4 x i32> %x, <4 x i32> %i
  ret <4 x i32> %r
}

define <4 x i32> @vpaddd128_maskz_test(<4 x i32> %i, <4 x i32> %j, <4 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: vpaddd128_maskz_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %xmm2, %xmm2, %k1 ## encoding: [0x62,0xf2,0x6d,0x08,0x27,0xca]
; CHECK-NEXT:    vpaddd %xmm1, %xmm0, %xmm0 {%k1} {z} ## encoding: [0x62,0xf1,0x7d,0x89,0xfe,0xc1]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i32> %mask1, zeroinitializer
  %x = add <4 x i32> %i, %j
  %r = select <4 x i1> %mask, <4 x i32> %x, <4 x i32> zeroinitializer
  ret <4 x i32> %r
}

define <4 x i32> @vpaddd128_mask_fold_test(<4 x i32> %i, ptr %j.ptr, <4 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: vpaddd128_mask_fold_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %xmm1, %xmm1, %k1 ## encoding: [0x62,0xf2,0x75,0x08,0x27,0xc9]
; CHECK-NEXT:    vpaddd (%rdi), %xmm0, %xmm0 {%k1} ## encoding: [0x62,0xf1,0x7d,0x09,0xfe,0x07]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i32> %mask1, zeroinitializer
  %j = load <4 x i32>, ptr %j.ptr
  %x = add <4 x i32> %i, %j
  %r = select <4 x i1> %mask, <4 x i32> %x, <4 x i32> %i
  ret <4 x i32> %r
}

define <4 x i32> @vpaddd128_mask_broadcast_test(<4 x i32> %i, <4 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: vpaddd128_mask_broadcast_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %xmm1, %xmm1, %k1 ## encoding: [0x62,0xf2,0x75,0x08,0x27,0xc9]
; CHECK-NEXT:    vpaddd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %xmm0, %xmm0 {%k1} ## encoding: [0x62,0xf1,0x7d,0x19,0xfe,0x05,A,A,A,A]
; CHECK-NEXT:    ## fixup A - offset: 6, value: {{\.?LCPI[0-9]+_[0-9]+}}-4, kind: reloc_riprel_4byte
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i32> %mask1, zeroinitializer
  %x = add <4 x i32> %i, <i32 7, i32 7, i32 7, i32 7>
  %r = select <4 x i1> %mask, <4 x i32> %x, <4 x i32> %i
  ret <4 x i32> %r
}

define <4 x i32> @vpaddd128_maskz_fold_test(<4 x i32> %i, ptr %j.ptr, <4 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: vpaddd128_maskz_fold_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %xmm1, %xmm1, %k1 ## encoding: [0x62,0xf2,0x75,0x08,0x27,0xc9]
; CHECK-NEXT:    vpaddd (%rdi), %xmm0, %xmm0 {%k1} {z} ## encoding: [0x62,0xf1,0x7d,0x89,0xfe,0x07]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i32> %mask1, zeroinitializer
  %j = load <4 x i32>, ptr %j.ptr
  %x = add <4 x i32> %i, %j
  %r = select <4 x i1> %mask, <4 x i32> %x, <4 x i32> zeroinitializer
  ret <4 x i32> %r
}

define <4 x i32> @vpaddd128_maskz_broadcast_test(<4 x i32> %i, <4 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: vpaddd128_maskz_broadcast_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %xmm1, %xmm1, %k1 ## encoding: [0x62,0xf2,0x75,0x08,0x27,0xc9]
; CHECK-NEXT:    vpaddd {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %xmm0, %xmm0 {%k1} {z} ## encoding: [0x62,0xf1,0x7d,0x99,0xfe,0x05,A,A,A,A]
; CHECK-NEXT:    ## fixup A - offset: 6, value: {{\.?LCPI[0-9]+_[0-9]+}}-4, kind: reloc_riprel_4byte
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i32> %mask1, zeroinitializer
  %x = add <4 x i32> %i, <i32 8, i32 8, i32 8, i32 8>
  %r = select <4 x i1> %mask, <4 x i32> %x, <4 x i32> zeroinitializer
  ret <4 x i32> %r
}

define <2 x i64> @vpsubq128_test(<2 x i64> %i, <2 x i64> %j) nounwind readnone {
; CHECK-LABEL: vpsubq128_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vpsubq %xmm1, %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf9,0xfb,0xc1]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %x = sub <2 x i64> %i, %j
  ret <2 x i64> %x
}

define <4 x i32> @vpsubd128_test(<4 x i32> %i, <4 x i32> %j) nounwind readnone {
; CHECK-LABEL: vpsubd128_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vpsubd %xmm1, %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf9,0xfa,0xc1]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %x = sub <4 x i32> %i, %j
  ret <4 x i32> %x
}

define <4 x i32> @vpmulld128_test(<4 x i32> %i, <4 x i32> %j) {
; CHECK-LABEL: vpmulld128_test:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vpmulld %xmm1, %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc4,0xe2,0x79,0x40,0xc1]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %x = mul <4 x i32> %i, %j
  ret <4 x i32> %x
}

define <2 x double> @test_vaddpd_128(<2 x double> %y, <2 x double> %x) {
; CHECK-LABEL: test_vaddpd_128:
; CHECK:       ## %bb.0: ## %entry
; CHECK-NEXT:    vaddpd %xmm0, %xmm1, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf1,0x58,0xc0]
; CHECK-NEXT:    retq ## encoding: [0xc3]
entry:
  %add.i = fadd <2 x double> %x, %y
  ret <2 x double> %add.i
}

define <2 x double> @test_fold_vaddpd_128(<2 x double> %y) {
; CHECK-LABEL: test_fold_vaddpd_128:
; CHECK:       ## %bb.0: ## %entry
; CHECK-NEXT:    vaddpd {{\.?LCPI[0-9]+_[0-9]+}}(%rip), %xmm0, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf9,0x58,0x05,A,A,A,A]
; CHECK-NEXT:    ## fixup A - offset: 4, value: {{\.?LCPI[0-9]+_[0-9]+}}-4, kind: reloc_riprel_4byte
; CHECK-NEXT:    retq ## encoding: [0xc3]
entry:
  %add.i = fadd <2 x double> %y, <double 4.500000e+00, double 3.400000e+00>
  ret <2 x double> %add.i
}

define <4 x float> @test_broadcast_vaddpd_128(<4 x float> %a) nounwind {
; CHECK-LABEL: test_broadcast_vaddpd_128:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vaddps {{\.?LCPI[0-9]+_[0-9]+}}(%rip){1to4}, %xmm0, %xmm0 ## encoding: [0x62,0xf1,0x7c,0x18,0x58,0x05,A,A,A,A]
; CHECK-NEXT:    ## fixup A - offset: 6, value: {{\.?LCPI[0-9]+_[0-9]+}}-4, kind: reloc_riprel_4byte
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %b = fadd <4 x float> %a, <float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000, float 0x3FB99999A0000000>
  ret <4 x float> %b
}

define <4 x float> @test_mask_vaddps_128(<4 x float> %dst, <4 x float> %i, <4 x float> %j, <4 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vaddps_128:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %xmm3, %xmm3, %k1 ## encoding: [0x62,0xf2,0x65,0x08,0x27,0xcb]
; CHECK-NEXT:    vaddps %xmm2, %xmm1, %xmm0 {%k1} ## encoding: [0x62,0xf1,0x74,0x09,0x58,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i32> %mask1, zeroinitializer
  %x = fadd <4 x float> %i, %j
  %r = select <4 x i1> %mask, <4 x float> %x, <4 x float> %dst
  ret <4 x float> %r
}

define <4 x float> @test_mask_vmulps_128(<4 x float> %dst, <4 x float> %i, <4 x float> %j, <4 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vmulps_128:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %xmm3, %xmm3, %k1 ## encoding: [0x62,0xf2,0x65,0x08,0x27,0xcb]
; CHECK-NEXT:    vmulps %xmm2, %xmm1, %xmm0 {%k1} ## encoding: [0x62,0xf1,0x74,0x09,0x59,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i32> %mask1, zeroinitializer
  %x = fmul <4 x float> %i, %j
  %r = select <4 x i1> %mask, <4 x float> %x, <4 x float> %dst
  ret <4 x float> %r
}

define <4 x float> @test_mask_vminps_128(<4 x float> %dst, <4 x float> %i, <4 x float> %j, <4 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vminps_128:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %xmm3, %xmm3, %k1 ## encoding: [0x62,0xf2,0x65,0x08,0x27,0xcb]
; CHECK-NEXT:    vminps %xmm2, %xmm1, %xmm0 {%k1} ## encoding: [0x62,0xf1,0x74,0x09,0x5d,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i32> %mask1, zeroinitializer
  %cmp_res = fcmp olt <4 x float> %i, %j
  %min = select <4 x i1> %cmp_res, <4 x float> %i, <4 x float> %j
  %r = select <4 x i1> %mask, <4 x float> %min, <4 x float> %dst
  ret <4 x float> %r
}

define <4 x float> @test_mask_vmaxps_128(<4 x float> %dst, <4 x float> %i, <4 x float> %j, <4 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vmaxps_128:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %xmm3, %xmm3, %k1 ## encoding: [0x62,0xf2,0x65,0x08,0x27,0xcb]
; CHECK-NEXT:    vmaxps %xmm2, %xmm1, %xmm0 {%k1} ## encoding: [0x62,0xf1,0x74,0x09,0x5f,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i32> %mask1, zeroinitializer
  %cmp_res = fcmp ogt <4 x float> %i, %j
  %max = select <4 x i1> %cmp_res, <4 x float> %i, <4 x float> %j
  %r = select <4 x i1> %mask, <4 x float> %max, <4 x float> %dst
  ret <4 x float> %r
}

define <4 x float> @test_mask_vsubps_128(<4 x float> %dst, <4 x float> %i, <4 x float> %j, <4 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vsubps_128:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %xmm3, %xmm3, %k1 ## encoding: [0x62,0xf2,0x65,0x08,0x27,0xcb]
; CHECK-NEXT:    vsubps %xmm2, %xmm1, %xmm0 {%k1} ## encoding: [0x62,0xf1,0x74,0x09,0x5c,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i32> %mask1, zeroinitializer
  %x = fsub <4 x float> %i, %j
  %r = select <4 x i1> %mask, <4 x float> %x, <4 x float> %dst
  ret <4 x float> %r
}


define <4 x float> @test_mask_vdivps_128(<4 x float> %dst, <4 x float> %i, <4 x float> %j, <4 x i32> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vdivps_128:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmd %xmm3, %xmm3, %k1 ## encoding: [0x62,0xf2,0x65,0x08,0x27,0xcb]
; CHECK-NEXT:    vdivps %xmm2, %xmm1, %xmm0 {%k1} ## encoding: [0x62,0xf1,0x74,0x09,0x5e,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <4 x i32> %mask1, zeroinitializer
  %x = fdiv <4 x float> %i, %j
  %r = select <4 x i1> %mask, <4 x float> %x, <4 x float> %dst
  ret <4 x float> %r
}

define <2 x double> @test_mask_vmulpd_128(<2 x double> %dst, <2 x double> %i, <2 x double> %j, <2 x i64> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vmulpd_128:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmq %xmm3, %xmm3, %k1 ## encoding: [0x62,0xf2,0xe5,0x08,0x27,0xcb]
; CHECK-NEXT:    vmulpd %xmm2, %xmm1, %xmm0 {%k1} ## encoding: [0x62,0xf1,0xf5,0x09,0x59,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <2 x i64> %mask1, zeroinitializer
  %x = fmul <2 x double> %i, %j
  %r = select <2 x i1> %mask, <2 x double> %x, <2 x double> %dst
  ret <2 x double> %r
}

define <2 x double> @test_mask_vminpd_128(<2 x double> %dst, <2 x double> %i, <2 x double> %j, <2 x i64> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vminpd_128:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmq %xmm3, %xmm3, %k1 ## encoding: [0x62,0xf2,0xe5,0x08,0x27,0xcb]
; CHECK-NEXT:    vminpd %xmm2, %xmm1, %xmm0 {%k1} ## encoding: [0x62,0xf1,0xf5,0x09,0x5d,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <2 x i64> %mask1, zeroinitializer
  %cmp_res = fcmp olt <2 x double> %i, %j
  %min = select <2 x i1> %cmp_res, <2 x double> %i, <2 x double> %j
  %r = select <2 x i1> %mask, <2 x double> %min, <2 x double> %dst
  ret <2 x double> %r
}

define <2 x double> @test_mask_vmaxpd_128(<2 x double> %dst, <2 x double> %i, <2 x double> %j, <2 x i64> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vmaxpd_128:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmq %xmm3, %xmm3, %k1 ## encoding: [0x62,0xf2,0xe5,0x08,0x27,0xcb]
; CHECK-NEXT:    vmaxpd %xmm2, %xmm1, %xmm0 {%k1} ## encoding: [0x62,0xf1,0xf5,0x09,0x5f,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <2 x i64> %mask1, zeroinitializer
  %cmp_res = fcmp ogt <2 x double> %i, %j
  %max = select <2 x i1> %cmp_res, <2 x double> %i, <2 x double> %j
  %r = select <2 x i1> %mask, <2 x double> %max, <2 x double> %dst
  ret <2 x double> %r
}

define <2 x double> @test_mask_vsubpd_128(<2 x double> %dst, <2 x double> %i, <2 x double> %j, <2 x i64> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vsubpd_128:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmq %xmm3, %xmm3, %k1 ## encoding: [0x62,0xf2,0xe5,0x08,0x27,0xcb]
; CHECK-NEXT:    vsubpd %xmm2, %xmm1, %xmm0 {%k1} ## encoding: [0x62,0xf1,0xf5,0x09,0x5c,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <2 x i64> %mask1, zeroinitializer
  %x = fsub <2 x double> %i, %j
  %r = select <2 x i1> %mask, <2 x double> %x, <2 x double> %dst
  ret <2 x double> %r
}

define <2 x double> @test_mask_vdivpd_128(<2 x double> %dst, <2 x double> %i, <2 x double> %j, <2 x i64> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vdivpd_128:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmq %xmm3, %xmm3, %k1 ## encoding: [0x62,0xf2,0xe5,0x08,0x27,0xcb]
; CHECK-NEXT:    vdivpd %xmm2, %xmm1, %xmm0 {%k1} ## encoding: [0x62,0xf1,0xf5,0x09,0x5e,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <2 x i64> %mask1, zeroinitializer
  %x = fdiv <2 x double> %i, %j
  %r = select <2 x i1> %mask, <2 x double> %x, <2 x double> %dst
  ret <2 x double> %r
}

define <2 x double> @test_mask_vaddpd_128(<2 x double> %dst, <2 x double> %i, <2 x double> %j, <2 x i64> %mask1) nounwind readnone {
; CHECK-LABEL: test_mask_vaddpd_128:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmq %xmm3, %xmm3, %k1 ## encoding: [0x62,0xf2,0xe5,0x08,0x27,0xcb]
; CHECK-NEXT:    vaddpd %xmm2, %xmm1, %xmm0 {%k1} ## encoding: [0x62,0xf1,0xf5,0x09,0x58,0xc2]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <2 x i64> %mask1, zeroinitializer
  %x = fadd <2 x double> %i, %j
  %r = select <2 x i1> %mask, <2 x double> %x, <2 x double> %dst
  ret <2 x double> %r
}

define <2 x double> @test_maskz_vaddpd_128(<2 x double> %i, <2 x double> %j,
; CHECK-LABEL: test_maskz_vaddpd_128:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmq %xmm2, %xmm2, %k1 ## encoding: [0x62,0xf2,0xed,0x08,0x27,0xca]
; CHECK-NEXT:    vaddpd %xmm1, %xmm0, %xmm0 {%k1} {z} ## encoding: [0x62,0xf1,0xfd,0x89,0x58,0xc1]
; CHECK-NEXT:    retq ## encoding: [0xc3]
                                          <2 x i64> %mask1) nounwind readnone {
  %mask = icmp ne <2 x i64> %mask1, zeroinitializer
  %x = fadd <2 x double> %i, %j
  %r = select <2 x i1> %mask, <2 x double> %x, <2 x double> zeroinitializer
  ret <2 x double> %r
}

define <2 x double> @test_mask_fold_vaddpd_128(<2 x double> %dst, <2 x double> %i, ptr %j,  <2 x i64> %mask1) nounwind {
; CHECK-LABEL: test_mask_fold_vaddpd_128:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmq %xmm2, %xmm2, %k1 ## encoding: [0x62,0xf2,0xed,0x08,0x27,0xca]
; CHECK-NEXT:    vaddpd (%rdi), %xmm1, %xmm0 {%k1} ## encoding: [0x62,0xf1,0xf5,0x09,0x58,0x07]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <2 x i64> %mask1, zeroinitializer
  %tmp = load <2 x double>, ptr %j
  %x = fadd <2 x double> %i, %tmp
  %r = select <2 x i1> %mask, <2 x double> %x, <2 x double> %dst
  ret <2 x double> %r
}

define <2 x double> @test_maskz_fold_vaddpd_128(<2 x double> %i, ptr %j, <2 x i64> %mask1) nounwind {
; CHECK-LABEL: test_maskz_fold_vaddpd_128:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmq %xmm1, %xmm1, %k1 ## encoding: [0x62,0xf2,0xf5,0x08,0x27,0xc9]
; CHECK-NEXT:    vaddpd (%rdi), %xmm0, %xmm0 {%k1} {z} ## encoding: [0x62,0xf1,0xfd,0x89,0x58,0x07]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <2 x i64> %mask1, zeroinitializer
  %tmp = load <2 x double>, ptr %j
  %x = fadd <2 x double> %i, %tmp
  %r = select <2 x i1> %mask, <2 x double> %x, <2 x double> zeroinitializer
  ret <2 x double> %r
}

define <2 x double> @test_broadcast2_vaddpd_128(<2 x double> %i, ptr %j) nounwind {
; CHECK-LABEL: test_broadcast2_vaddpd_128:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vaddpd (%rdi){1to2}, %xmm0, %xmm0 ## encoding: [0x62,0xf1,0xfd,0x18,0x58,0x07]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %tmp = load double, ptr %j
  %j.0 = insertelement <2 x double> undef, double %tmp, i64 0
  %j.1 = insertelement <2 x double> %j.0, double %tmp, i64 1
  %x = fadd <2 x double> %j.1, %i
  ret <2 x double> %x
}

define <2 x double> @test_mask_broadcast_vaddpd_128(<2 x double> %dst, <2 x double> %i, ptr %j, <2 x i64> %mask1) nounwind {
; CHECK-LABEL: test_mask_broadcast_vaddpd_128:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vmovapd %xmm1, %xmm0 ## EVEX TO VEX Compression encoding: [0xc5,0xf9,0x28,0xc1]
; CHECK-NEXT:    vptestmq %xmm2, %xmm2, %k1 ## encoding: [0x62,0xf2,0xed,0x08,0x27,0xca]
; CHECK-NEXT:    vaddpd (%rdi){1to2}, %xmm1, %xmm0 {%k1} ## encoding: [0x62,0xf1,0xf5,0x19,0x58,0x07]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <2 x i64> %mask1, zeroinitializer
  %tmp = load double, ptr %j
  %j.0 = insertelement <2 x double> undef, double %tmp, i64 0
  %j.1 = insertelement <2 x double> %j.0, double %tmp, i64 1
  %x = fadd <2 x double> %j.1, %i
  %r = select <2 x i1> %mask, <2 x double> %x, <2 x double> %i
  ret <2 x double> %r
}

define <2 x double> @test_maskz_broadcast_vaddpd_128(<2 x double> %i, ptr %j, <2 x i64> %mask1) nounwind {
; CHECK-LABEL: test_maskz_broadcast_vaddpd_128:
; CHECK:       ## %bb.0:
; CHECK-NEXT:    vptestmq %xmm1, %xmm1, %k1 ## encoding: [0x62,0xf2,0xf5,0x08,0x27,0xc9]
; CHECK-NEXT:    vaddpd (%rdi){1to2}, %xmm0, %xmm0 {%k1} {z} ## encoding: [0x62,0xf1,0xfd,0x99,0x58,0x07]
; CHECK-NEXT:    retq ## encoding: [0xc3]
  %mask = icmp ne <2 x i64> %mask1, zeroinitializer
  %tmp = load double, ptr %j
  %j.0 = insertelement <2 x double> undef, double %tmp, i64 0
  %j.1 = insertelement <2 x double> %j.0, double %tmp, i64 1
  %x = fadd <2 x double> %j.1, %i
  %r = select <2 x i1> %mask, <2 x double> %x, <2 x double> zeroinitializer
  ret <2 x double> %r
}