; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx512dq --show-mc-encoding | FileCheck %s --check-prefixes=CHECK,X86,AVX512DQ,X86-AVX512DQ
; RUN: llc < %s -mtriple=i686-unknown-unknown -mattr=+avx512dq,+avx512vl --show-mc-encoding | FileCheck %s --check-prefixes=CHECK,X86,AVX512DQVL,X86-AVX512DQVL
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512dq --show-mc-encoding | FileCheck %s --check-prefixes=CHECK,X64,AVX512DQ,X64-AVX512DQ
; RUN: llc < %s -mtriple=x86_64-unknown-unknown -mattr=+avx512dq,+avx512vl --show-mc-encoding | FileCheck %s --check-prefixes=CHECK,X64,AVX512DQVL,X64-AVX512DQVL

define i32 @test_int_x86_avx512_kadd_w(<16 x i32> %A, <16 x i32> %B) nounwind {
; CHECK-LABEL: test_int_x86_avx512_kadd_w:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vptestmd %zmm0, %zmm0, %k0 # encoding: [0x62,0xf2,0x7d,0x48,0x27,0xc0]
; CHECK-NEXT:    vptestmd %zmm1, %zmm1, %k1 # encoding: [0x62,0xf2,0x75,0x48,0x27,0xc9]
; CHECK-NEXT:    kaddw %k1, %k0, %k0 # encoding: [0xc5,0xfc,0x4a,0xc1]
; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; CHECK-NEXT:    kortestw %k0, %k0 # encoding: [0xc5,0xf8,0x98,0xc0]
; CHECK-NEXT:    sete %al # encoding: [0x0f,0x94,0xc0]
; CHECK-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
entry:
  %0 = icmp ne <16 x i32> %A, zeroinitializer
  %1 = icmp ne <16 x i32> %B, zeroinitializer
  %2 = call <16 x i1> @llvm.x86.avx512.kadd.w(<16 x i1> %0, <16 x i1> %1)
  %3 = bitcast <16 x i1> %2 to i16
  %4 = icmp eq i16 %3, 0
  %5 = zext i1 %4 to i32
  ret i32 %5
}
declare <16 x i1> @llvm.x86.avx512.kadd.w(<16 x i1>, <16 x i1>)

define i32 @test_int_x86_avx512_kadd_b(<8 x i64> %A, <8 x i64> %B) nounwind {
; CHECK-LABEL: test_int_x86_avx512_kadd_b:
; CHECK:       # %bb.0: # %entry
; CHECK-NEXT:    vptestmq %zmm0, %zmm0, %k0 # encoding: [0x62,0xf2,0xfd,0x48,0x27,0xc0]
; CHECK-NEXT:    vptestmq %zmm1, %zmm1, %k1 # encoding: [0x62,0xf2,0xf5,0x48,0x27,0xc9]
; CHECK-NEXT:    kaddb %k1, %k0, %k0 # encoding: [0xc5,0xfd,0x4a,0xc1]
; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; CHECK-NEXT:    kortestb %k0, %k0 # encoding: [0xc5,0xf9,0x98,0xc0]
; CHECK-NEXT:    sete %al # encoding: [0x0f,0x94,0xc0]
; CHECK-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
entry:
  %0 = icmp ne <8 x i64> %A, zeroinitializer
  %1 = icmp ne <8 x i64> %B, zeroinitializer
  %2 = call <8 x i1> @llvm.x86.avx512.kadd.b(<8 x i1> %0, <8 x i1> %1)
  %3 = bitcast <8 x i1> %2 to i8
  %4 = icmp eq i8 %3, 0
  %5 = zext i1 %4 to i32
  ret i32 %5
}
declare <8 x i1> @llvm.x86.avx512.kadd.b(<8 x i1>, <8 x i1>)

define i32 @test_x86_avx512_ktestc_w(<16 x i32> %A, <16 x i32> %B) {
; CHECK-LABEL: test_x86_avx512_ktestc_w:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vptestmd %zmm0, %zmm0, %k0 # encoding: [0x62,0xf2,0x7d,0x48,0x27,0xc0]
; CHECK-NEXT:    vptestmd %zmm1, %zmm1, %k1 # encoding: [0x62,0xf2,0x75,0x48,0x27,0xc9]
; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; CHECK-NEXT:    ktestw %k1, %k0 # encoding: [0xc5,0xf8,0x99,0xc1]
; CHECK-NEXT:    setb %al # encoding: [0x0f,0x92,0xc0]
; CHECK-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %1 = icmp ne <16 x i32> %A, zeroinitializer
  %2 = icmp ne <16 x i32> %B, zeroinitializer
  %res = call i32 @llvm.x86.avx512.ktestc.w(<16 x i1> %1, <16 x i1> %2) ; <i32> [#uses=1]
  ret i32 %res
}
declare i32 @llvm.x86.avx512.ktestc.w(<16 x i1>, <16 x i1>) nounwind readnone

define i32 @test_x86_avx512_ktestz_w(<16 x i32> %A, <16 x i32> %B) {
; CHECK-LABEL: test_x86_avx512_ktestz_w:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vptestmd %zmm0, %zmm0, %k0 # encoding: [0x62,0xf2,0x7d,0x48,0x27,0xc0]
; CHECK-NEXT:    vptestmd %zmm1, %zmm1, %k1 # encoding: [0x62,0xf2,0x75,0x48,0x27,0xc9]
; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; CHECK-NEXT:    ktestw %k1, %k0 # encoding: [0xc5,0xf8,0x99,0xc1]
; CHECK-NEXT:    sete %al # encoding: [0x0f,0x94,0xc0]
; CHECK-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %1 = icmp ne <16 x i32> %A, zeroinitializer
  %2 = icmp ne <16 x i32> %B, zeroinitializer
  %res = call i32 @llvm.x86.avx512.ktestz.w(<16 x i1> %1, <16 x i1> %2) ; <i32> [#uses=1]
  ret i32 %res
}
declare i32 @llvm.x86.avx512.ktestz.w(<16 x i1>, <16 x i1>) nounwind readnone

define i32 @test_x86_avx512_ktestc_b(<8 x i64> %A, <8 x i64> %B) {
; CHECK-LABEL: test_x86_avx512_ktestc_b:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vptestmq %zmm0, %zmm0, %k0 # encoding: [0x62,0xf2,0xfd,0x48,0x27,0xc0]
; CHECK-NEXT:    vptestmq %zmm1, %zmm1, %k1 # encoding: [0x62,0xf2,0xf5,0x48,0x27,0xc9]
; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; CHECK-NEXT:    ktestb %k1, %k0 # encoding: [0xc5,0xf9,0x99,0xc1]
; CHECK-NEXT:    setb %al # encoding: [0x0f,0x92,0xc0]
; CHECK-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %1 = icmp ne <8 x i64> %A, zeroinitializer
  %2 = icmp ne <8 x i64> %B, zeroinitializer
  %res = call i32 @llvm.x86.avx512.ktestc.b(<8 x i1> %1, <8 x i1> %2) ; <i32> [#uses=1]
  ret i32 %res
}
declare i32 @llvm.x86.avx512.ktestc.b(<8 x i1>, <8 x i1>) nounwind readnone

define i32 @test_x86_avx512_ktestz_b(<8 x i64> %A, <8 x i64> %B) {
; CHECK-LABEL: test_x86_avx512_ktestz_b:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vptestmq %zmm0, %zmm0, %k0 # encoding: [0x62,0xf2,0xfd,0x48,0x27,0xc0]
; CHECK-NEXT:    vptestmq %zmm1, %zmm1, %k1 # encoding: [0x62,0xf2,0xf5,0x48,0x27,0xc9]
; CHECK-NEXT:    xorl %eax, %eax # encoding: [0x31,0xc0]
; CHECK-NEXT:    ktestb %k1, %k0 # encoding: [0xc5,0xf9,0x99,0xc1]
; CHECK-NEXT:    sete %al # encoding: [0x0f,0x94,0xc0]
; CHECK-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %1 = icmp ne <8 x i64> %A, zeroinitializer
  %2 = icmp ne <8 x i64> %B, zeroinitializer
  %res = call i32 @llvm.x86.avx512.ktestz.b(<8 x i1> %1, <8 x i1> %2) ; <i32> [#uses=1]
  ret i32 %res
}
declare i32 @llvm.x86.avx512.ktestz.b(<8 x i1>, <8 x i1>) nounwind readnone

declare <8 x i64> @llvm.x86.avx512.mask.cvtpd2qq.512(<8 x double>, <8 x i64>, i8, i32)

define <8 x i64>@test_int_x86_avx512_mask_cvt_pd2qq_512(<8 x double> %x0, <8 x i64> %x1, i8 %x2) {
; X86-LABEL: test_int_x86_avx512_mask_cvt_pd2qq_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vcvtpd2qq {ru-sae}, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0xfd,0x59,0x7b,0xc8]
; X86-NEXT:    vcvtpd2qq {rn-sae}, %zmm0, %zmm0 # encoding: [0x62,0xf1,0xfd,0x18,0x7b,0xc0]
; X86-NEXT:    vpaddq %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0xd4,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_cvt_pd2qq_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vcvtpd2qq {ru-sae}, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0xfd,0x59,0x7b,0xc8]
; X64-NEXT:    vcvtpd2qq {rn-sae}, %zmm0, %zmm0 # encoding: [0x62,0xf1,0xfd,0x18,0x7b,0xc0]
; X64-NEXT:    vpaddq %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0xd4,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x i64> @llvm.x86.avx512.mask.cvtpd2qq.512(<8 x double> %x0, <8 x i64> %x1, i8 %x2, i32 2)
  %res1 = call <8 x i64> @llvm.x86.avx512.mask.cvtpd2qq.512(<8 x double> %x0, <8 x i64> %x1, i8 -1, i32 0)
  %res2 = add <8 x i64> %res, %res1
  ret <8 x i64> %res2
}

declare <8 x i64> @llvm.x86.avx512.mask.cvtpd2uqq.512(<8 x double>, <8 x i64>, i8, i32)

define <8 x i64>@test_int_x86_avx512_mask_cvt_pd2uqq_512(<8 x double> %x0, <8 x i64> %x1, i8 %x2) {
; X86-LABEL: test_int_x86_avx512_mask_cvt_pd2uqq_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vcvtpd2uqq {ru-sae}, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0xfd,0x59,0x79,0xc8]
; X86-NEXT:    vcvtpd2uqq {rn-sae}, %zmm0, %zmm0 # encoding: [0x62,0xf1,0xfd,0x18,0x79,0xc0]
; X86-NEXT:    vpaddq %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0xd4,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_cvt_pd2uqq_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vcvtpd2uqq {ru-sae}, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0xfd,0x59,0x79,0xc8]
; X64-NEXT:    vcvtpd2uqq {rn-sae}, %zmm0, %zmm0 # encoding: [0x62,0xf1,0xfd,0x18,0x79,0xc0]
; X64-NEXT:    vpaddq %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0xd4,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x i64> @llvm.x86.avx512.mask.cvtpd2uqq.512(<8 x double> %x0, <8 x i64> %x1, i8 %x2, i32 2)
  %res1 = call <8 x i64> @llvm.x86.avx512.mask.cvtpd2uqq.512(<8 x double> %x0, <8 x i64> %x1, i8 -1, i32 0)
  %res2 = add <8 x i64> %res, %res1
  ret <8 x i64> %res2
}

declare <8 x i64> @llvm.x86.avx512.mask.cvtps2qq.512(<8 x float>, <8 x i64>, i8, i32)

define <8 x i64>@test_int_x86_avx512_mask_cvt_ps2qq_512(<8 x float> %x0, <8 x i64> %x1, i8 %x2) {
; X86-LABEL: test_int_x86_avx512_mask_cvt_ps2qq_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vcvtps2qq {ru-sae}, %ymm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7d,0x59,0x7b,0xc8]
; X86-NEXT:    vcvtps2qq {rn-sae}, %ymm0, %zmm0 # encoding: [0x62,0xf1,0x7d,0x18,0x7b,0xc0]
; X86-NEXT:    vpaddq %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0xd4,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_cvt_ps2qq_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vcvtps2qq {ru-sae}, %ymm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7d,0x59,0x7b,0xc8]
; X64-NEXT:    vcvtps2qq {rn-sae}, %ymm0, %zmm0 # encoding: [0x62,0xf1,0x7d,0x18,0x7b,0xc0]
; X64-NEXT:    vpaddq %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0xd4,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x i64> @llvm.x86.avx512.mask.cvtps2qq.512(<8 x float> %x0, <8 x i64> %x1, i8 %x2, i32 2)
  %res1 = call <8 x i64> @llvm.x86.avx512.mask.cvtps2qq.512(<8 x float> %x0, <8 x i64> %x1, i8 -1, i32 0)
  %res2 = add <8 x i64> %res, %res1
  ret <8 x i64> %res2
}

declare <8 x i64> @llvm.x86.avx512.mask.cvtps2uqq.512(<8 x float>, <8 x i64>, i8, i32)

define <8 x i64>@test_int_x86_avx512_mask_cvt_ps2uqq_512(<8 x float> %x0, <8 x i64> %x1, i8 %x2) {
; X86-LABEL: test_int_x86_avx512_mask_cvt_ps2uqq_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vcvtps2uqq {ru-sae}, %ymm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7d,0x59,0x79,0xc8]
; X86-NEXT:    vcvtps2uqq {rn-sae}, %ymm0, %zmm0 # encoding: [0x62,0xf1,0x7d,0x18,0x79,0xc0]
; X86-NEXT:    vpaddq %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0xd4,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_cvt_ps2uqq_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vcvtps2uqq {ru-sae}, %ymm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7d,0x59,0x79,0xc8]
; X64-NEXT:    vcvtps2uqq {rn-sae}, %ymm0, %zmm0 # encoding: [0x62,0xf1,0x7d,0x18,0x79,0xc0]
; X64-NEXT:    vpaddq %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0xd4,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x i64> @llvm.x86.avx512.mask.cvtps2uqq.512(<8 x float> %x0, <8 x i64> %x1, i8 %x2, i32 2)
  %res1 = call <8 x i64> @llvm.x86.avx512.mask.cvtps2uqq.512(<8 x float> %x0, <8 x i64> %x1, i8 -1, i32 0)
  %res2 = add <8 x i64> %res, %res1
  ret <8 x i64> %res2
}

declare <8 x double> @llvm.x86.avx512.mask.cvtqq2pd.512(<8 x i64>, <8 x double>, i8, i32)

define <8 x double>@test_int_x86_avx512_mask_cvt_qq2pd_512(<8 x i64> %x0, <8 x double> %x1, i8 %x2) {
; X86-LABEL: test_int_x86_avx512_mask_cvt_qq2pd_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vcvtqq2pd %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0xfe,0x49,0xe6,0xc8]
; X86-NEXT:    vcvtqq2pd {rn-sae}, %zmm0, %zmm0 # encoding: [0x62,0xf1,0xfe,0x18,0xe6,0xc0]
; X86-NEXT:    vaddpd %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0x58,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_cvt_qq2pd_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vcvtqq2pd %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0xfe,0x49,0xe6,0xc8]
; X64-NEXT:    vcvtqq2pd {rn-sae}, %zmm0, %zmm0 # encoding: [0x62,0xf1,0xfe,0x18,0xe6,0xc0]
; X64-NEXT:    vaddpd %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0x58,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x double> @llvm.x86.avx512.mask.cvtqq2pd.512(<8 x i64> %x0, <8 x double> %x1, i8 %x2, i32 4)
  %res1 = call <8 x double> @llvm.x86.avx512.mask.cvtqq2pd.512(<8 x i64> %x0, <8 x double> %x1, i8 -1, i32 0)
  %res2 = fadd <8 x double> %res, %res1
  ret <8 x double> %res2
}

declare <8 x float> @llvm.x86.avx512.mask.cvtqq2ps.512(<8 x i64>, <8 x float>, i8, i32)

define <8 x float>@test_int_x86_avx512_mask_cvt_qq2ps_512(<8 x i64> %x0, <8 x float> %x1, i8 %x2) {
; X86-AVX512DQ-LABEL: test_int_x86_avx512_mask_cvt_qq2ps_512:
; X86-AVX512DQ:       # %bb.0:
; X86-AVX512DQ-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-AVX512DQ-NEXT:    vcvtqq2ps %zmm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0xfc,0x49,0x5b,0xc8]
; X86-AVX512DQ-NEXT:    vcvtqq2ps {rn-sae}, %zmm0, %ymm0 # encoding: [0x62,0xf1,0xfc,0x18,0x5b,0xc0]
; X86-AVX512DQ-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # encoding: [0xc5,0xf4,0x58,0xc0]
; X86-AVX512DQ-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512DQVL-LABEL: test_int_x86_avx512_mask_cvt_qq2ps_512:
; X86-AVX512DQVL:       # %bb.0:
; X86-AVX512DQVL-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-AVX512DQVL-NEXT:    vcvtqq2ps %zmm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0xfc,0x49,0x5b,0xc8]
; X86-AVX512DQVL-NEXT:    vcvtqq2ps {rn-sae}, %zmm0, %ymm0 # encoding: [0x62,0xf1,0xfc,0x18,0x5b,0xc0]
; X86-AVX512DQVL-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xf4,0x58,0xc0]
; X86-AVX512DQVL-NEXT:    retl # encoding: [0xc3]
;
; X64-AVX512DQ-LABEL: test_int_x86_avx512_mask_cvt_qq2ps_512:
; X64-AVX512DQ:       # %bb.0:
; X64-AVX512DQ-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-AVX512DQ-NEXT:    vcvtqq2ps %zmm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0xfc,0x49,0x5b,0xc8]
; X64-AVX512DQ-NEXT:    vcvtqq2ps {rn-sae}, %zmm0, %ymm0 # encoding: [0x62,0xf1,0xfc,0x18,0x5b,0xc0]
; X64-AVX512DQ-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # encoding: [0xc5,0xf4,0x58,0xc0]
; X64-AVX512DQ-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512DQVL-LABEL: test_int_x86_avx512_mask_cvt_qq2ps_512:
; X64-AVX512DQVL:       # %bb.0:
; X64-AVX512DQVL-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-AVX512DQVL-NEXT:    vcvtqq2ps %zmm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0xfc,0x49,0x5b,0xc8]
; X64-AVX512DQVL-NEXT:    vcvtqq2ps {rn-sae}, %zmm0, %ymm0 # encoding: [0x62,0xf1,0xfc,0x18,0x5b,0xc0]
; X64-AVX512DQVL-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xf4,0x58,0xc0]
; X64-AVX512DQVL-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x float> @llvm.x86.avx512.mask.cvtqq2ps.512(<8 x i64> %x0, <8 x float> %x1, i8 %x2, i32 4)
  %res1 = call <8 x float> @llvm.x86.avx512.mask.cvtqq2ps.512(<8 x i64> %x0, <8 x float> %x1, i8 -1, i32 0)
  %res2 = fadd <8 x float> %res, %res1
  ret <8 x float> %res2
}

declare <8 x i64> @llvm.x86.avx512.mask.cvttpd2qq.512(<8 x double>, <8 x i64>, i8, i32)

define <8 x i64>@test_int_x86_avx512_mask_cvtt_pd2qq_512(<8 x double> %x0, <8 x i64> %x1, i8 %x2) {
; X86-LABEL: test_int_x86_avx512_mask_cvtt_pd2qq_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vcvttpd2qq %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0xfd,0x49,0x7a,0xc8]
; X86-NEXT:    vcvttpd2qq {sae}, %zmm0, %zmm0 # encoding: [0x62,0xf1,0xfd,0x18,0x7a,0xc0]
; X86-NEXT:    vpaddq %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0xd4,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_cvtt_pd2qq_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vcvttpd2qq %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0xfd,0x49,0x7a,0xc8]
; X64-NEXT:    vcvttpd2qq {sae}, %zmm0, %zmm0 # encoding: [0x62,0xf1,0xfd,0x18,0x7a,0xc0]
; X64-NEXT:    vpaddq %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0xd4,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x i64> @llvm.x86.avx512.mask.cvttpd2qq.512(<8 x double> %x0, <8 x i64> %x1, i8 %x2, i32 4)
  %res1 = call <8 x i64> @llvm.x86.avx512.mask.cvttpd2qq.512(<8 x double> %x0, <8 x i64> %x1, i8 -1, i32 8)
  %res2 = add <8 x i64> %res, %res1
  ret <8 x i64> %res2
}

declare <8 x i64> @llvm.x86.avx512.mask.cvttpd2uqq.512(<8 x double>, <8 x i64>, i8, i32)

define <8 x i64>@test_int_x86_avx512_mask_cvtt_pd2uqq_512(<8 x double> %x0, <8 x i64> %x1, i8 %x2) {
; X86-LABEL: test_int_x86_avx512_mask_cvtt_pd2uqq_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vcvttpd2uqq %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0xfd,0x49,0x78,0xc8]
; X86-NEXT:    vcvttpd2uqq {sae}, %zmm0, %zmm0 # encoding: [0x62,0xf1,0xfd,0x18,0x78,0xc0]
; X86-NEXT:    vpaddq %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0xd4,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_cvtt_pd2uqq_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vcvttpd2uqq %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0xfd,0x49,0x78,0xc8]
; X64-NEXT:    vcvttpd2uqq {sae}, %zmm0, %zmm0 # encoding: [0x62,0xf1,0xfd,0x18,0x78,0xc0]
; X64-NEXT:    vpaddq %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0xd4,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x i64> @llvm.x86.avx512.mask.cvttpd2uqq.512(<8 x double> %x0, <8 x i64> %x1, i8 %x2, i32 4)
  %res1 = call <8 x i64> @llvm.x86.avx512.mask.cvttpd2uqq.512(<8 x double> %x0, <8 x i64> %x1, i8 -1, i32 8)
  %res2 = add <8 x i64> %res, %res1
  ret <8 x i64> %res2
}

declare <8 x i64> @llvm.x86.avx512.mask.cvttps2qq.512(<8 x float>, <8 x i64>, i8, i32)

define <8 x i64>@test_int_x86_avx512_mask_cvtt_ps2qq_512(<8 x float> %x0, <8 x i64> %x1, i8 %x2) {
; X86-LABEL: test_int_x86_avx512_mask_cvtt_ps2qq_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vcvttps2qq %ymm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7d,0x49,0x7a,0xc8]
; X86-NEXT:    vcvttps2qq {sae}, %ymm0, %zmm0 # encoding: [0x62,0xf1,0x7d,0x18,0x7a,0xc0]
; X86-NEXT:    vpaddq %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0xd4,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_cvtt_ps2qq_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vcvttps2qq %ymm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7d,0x49,0x7a,0xc8]
; X64-NEXT:    vcvttps2qq {sae}, %ymm0, %zmm0 # encoding: [0x62,0xf1,0x7d,0x18,0x7a,0xc0]
; X64-NEXT:    vpaddq %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0xd4,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x i64> @llvm.x86.avx512.mask.cvttps2qq.512(<8 x float> %x0, <8 x i64> %x1, i8 %x2, i32 4)
  %res1 = call <8 x i64> @llvm.x86.avx512.mask.cvttps2qq.512(<8 x float> %x0, <8 x i64> %x1, i8 -1, i32 8)
  %res2 = add <8 x i64> %res, %res1
  ret <8 x i64> %res2
}

declare <8 x i64> @llvm.x86.avx512.mask.cvttps2uqq.512(<8 x float>, <8 x i64>, i8, i32)

define <8 x i64>@test_int_x86_avx512_mask_cvtt_ps2uqq_512(<8 x float> %x0, <8 x i64> %x1, i8 %x2) {
; X86-LABEL: test_int_x86_avx512_mask_cvtt_ps2uqq_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vcvttps2uqq %ymm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7d,0x49,0x78,0xc8]
; X86-NEXT:    vcvttps2uqq {sae}, %ymm0, %zmm0 # encoding: [0x62,0xf1,0x7d,0x18,0x78,0xc0]
; X86-NEXT:    vpaddq %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0xd4,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_cvtt_ps2uqq_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vcvttps2uqq %ymm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0x7d,0x49,0x78,0xc8]
; X64-NEXT:    vcvttps2uqq {sae}, %ymm0, %zmm0 # encoding: [0x62,0xf1,0x7d,0x18,0x78,0xc0]
; X64-NEXT:    vpaddq %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0xd4,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x i64> @llvm.x86.avx512.mask.cvttps2uqq.512(<8 x float> %x0, <8 x i64> %x1, i8 %x2, i32 4)
  %res1 = call <8 x i64> @llvm.x86.avx512.mask.cvttps2uqq.512(<8 x float> %x0, <8 x i64> %x1, i8 -1, i32 8)
  %res2 = add <8 x i64> %res, %res1
  ret <8 x i64> %res2
}

declare <8 x double> @llvm.x86.avx512.mask.cvtuqq2pd.512(<8 x i64>, <8 x double>, i8, i32)

define <8 x double>@test_int_x86_avx512_mask_cvt_uqq2pd_512(<8 x i64> %x0, <8 x double> %x1, i8 %x2) {
; X86-LABEL: test_int_x86_avx512_mask_cvt_uqq2pd_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vcvtuqq2pd %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0xfe,0x49,0x7a,0xc8]
; X86-NEXT:    vcvtuqq2pd {rn-sae}, %zmm0, %zmm0 # encoding: [0x62,0xf1,0xfe,0x18,0x7a,0xc0]
; X86-NEXT:    vaddpd %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0x58,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_cvt_uqq2pd_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vcvtuqq2pd %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf1,0xfe,0x49,0x7a,0xc8]
; X64-NEXT:    vcvtuqq2pd {rn-sae}, %zmm0, %zmm0 # encoding: [0x62,0xf1,0xfe,0x18,0x7a,0xc0]
; X64-NEXT:    vaddpd %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0x58,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x double> @llvm.x86.avx512.mask.cvtuqq2pd.512(<8 x i64> %x0, <8 x double> %x1, i8 %x2, i32 4)
  %res1 = call <8 x double> @llvm.x86.avx512.mask.cvtuqq2pd.512(<8 x i64> %x0, <8 x double> %x1, i8 -1, i32 0)
  %res2 = fadd <8 x double> %res, %res1
  ret <8 x double> %res2
}

declare <8 x float> @llvm.x86.avx512.mask.cvtuqq2ps.512(<8 x i64>, <8 x float>, i8, i32)

define <8 x float>@test_int_x86_avx512_mask_cvt_uqq2ps_512(<8 x i64> %x0, <8 x float> %x1, i8 %x2) {
; X86-AVX512DQ-LABEL: test_int_x86_avx512_mask_cvt_uqq2ps_512:
; X86-AVX512DQ:       # %bb.0:
; X86-AVX512DQ-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-AVX512DQ-NEXT:    vcvtuqq2ps %zmm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0xff,0x49,0x7a,0xc8]
; X86-AVX512DQ-NEXT:    vcvtuqq2ps {rn-sae}, %zmm0, %ymm0 # encoding: [0x62,0xf1,0xff,0x18,0x7a,0xc0]
; X86-AVX512DQ-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # encoding: [0xc5,0xf4,0x58,0xc0]
; X86-AVX512DQ-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512DQVL-LABEL: test_int_x86_avx512_mask_cvt_uqq2ps_512:
; X86-AVX512DQVL:       # %bb.0:
; X86-AVX512DQVL-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-AVX512DQVL-NEXT:    vcvtuqq2ps %zmm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0xff,0x49,0x7a,0xc8]
; X86-AVX512DQVL-NEXT:    vcvtuqq2ps {rn-sae}, %zmm0, %ymm0 # encoding: [0x62,0xf1,0xff,0x18,0x7a,0xc0]
; X86-AVX512DQVL-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xf4,0x58,0xc0]
; X86-AVX512DQVL-NEXT:    retl # encoding: [0xc3]
;
; X64-AVX512DQ-LABEL: test_int_x86_avx512_mask_cvt_uqq2ps_512:
; X64-AVX512DQ:       # %bb.0:
; X64-AVX512DQ-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-AVX512DQ-NEXT:    vcvtuqq2ps %zmm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0xff,0x49,0x7a,0xc8]
; X64-AVX512DQ-NEXT:    vcvtuqq2ps {rn-sae}, %zmm0, %ymm0 # encoding: [0x62,0xf1,0xff,0x18,0x7a,0xc0]
; X64-AVX512DQ-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # encoding: [0xc5,0xf4,0x58,0xc0]
; X64-AVX512DQ-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512DQVL-LABEL: test_int_x86_avx512_mask_cvt_uqq2ps_512:
; X64-AVX512DQVL:       # %bb.0:
; X64-AVX512DQVL-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-AVX512DQVL-NEXT:    vcvtuqq2ps %zmm0, %ymm1 {%k1} # encoding: [0x62,0xf1,0xff,0x49,0x7a,0xc8]
; X64-AVX512DQVL-NEXT:    vcvtuqq2ps {rn-sae}, %zmm0, %ymm0 # encoding: [0x62,0xf1,0xff,0x18,0x7a,0xc0]
; X64-AVX512DQVL-NEXT:    vaddps %ymm0, %ymm1, %ymm0 # EVEX TO VEX Compression encoding: [0xc5,0xf4,0x58,0xc0]
; X64-AVX512DQVL-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x float> @llvm.x86.avx512.mask.cvtuqq2ps.512(<8 x i64> %x0, <8 x float> %x1, i8 %x2, i32 4)
  %res1 = call <8 x float> @llvm.x86.avx512.mask.cvtuqq2ps.512(<8 x i64> %x0, <8 x float> %x1, i8 -1, i32 0)
  %res2 = fadd <8 x float> %res, %res1
  ret <8 x float> %res2
}

declare <8 x double> @llvm.x86.avx512.mask.reduce.pd.512(<8 x double>, i32, <8 x double>, i8, i32)

define <8 x double>@test_int_x86_avx512_mask_reduce_pd_512(<8 x double> %x0, <8 x double> %x2, i8 %x3) {
; X86-LABEL: test_int_x86_avx512_mask_reduce_pd_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vreducepd $8, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0x56,0xc8,0x08]
; X86-NEXT:    vreducepd $4, {sae}, %zmm0, %zmm0 # encoding: [0x62,0xf3,0xfd,0x18,0x56,0xc0,0x04]
; X86-NEXT:    vaddpd %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0x58,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_reduce_pd_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vreducepd $8, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0x56,0xc8,0x08]
; X64-NEXT:    vreducepd $4, {sae}, %zmm0, %zmm0 # encoding: [0x62,0xf3,0xfd,0x18,0x56,0xc0,0x04]
; X64-NEXT:    vaddpd %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0xf5,0x48,0x58,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x double> @llvm.x86.avx512.mask.reduce.pd.512(<8 x double> %x0, i32 8, <8 x double> %x2, i8 %x3, i32 4)
  %res1 = call <8 x double> @llvm.x86.avx512.mask.reduce.pd.512(<8 x double> %x0, i32 4, <8 x double> %x2, i8 -1, i32 8)
  %res2 = fadd <8 x double> %res, %res1
  ret <8 x double> %res2
}

declare <16 x float> @llvm.x86.avx512.mask.reduce.ps.512(<16 x float>, i32, <16 x float>, i16, i32)

define <16 x float>@test_int_x86_avx512_mask_reduce_ps_512(<16 x float> %x0, <16 x float> %x2, i16 %x3) {
; X86-LABEL: test_int_x86_avx512_mask_reduce_ps_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vreduceps $44, {sae}, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf3,0x7d,0x19,0x56,0xc8,0x2c]
; X86-NEXT:    vreduceps $11, %zmm0, %zmm0 # encoding: [0x62,0xf3,0x7d,0x48,0x56,0xc0,0x0b]
; X86-NEXT:    vaddps %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0x74,0x48,0x58,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_reduce_ps_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vreduceps $44, {sae}, %zmm0, %zmm1 {%k1} # encoding: [0x62,0xf3,0x7d,0x19,0x56,0xc8,0x2c]
; X64-NEXT:    vreduceps $11, %zmm0, %zmm0 # encoding: [0x62,0xf3,0x7d,0x48,0x56,0xc0,0x0b]
; X64-NEXT:    vaddps %zmm0, %zmm1, %zmm0 # encoding: [0x62,0xf1,0x74,0x48,0x58,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <16 x float> @llvm.x86.avx512.mask.reduce.ps.512(<16 x float> %x0, i32 44, <16 x float> %x2, i16 %x3, i32 8)
  %res1 = call <16 x float> @llvm.x86.avx512.mask.reduce.ps.512(<16 x float> %x0, i32 11, <16 x float> %x2, i16 -1, i32 4)
  %res2 = fadd <16 x float> %res, %res1
  ret <16 x float> %res2
}

declare <8 x double> @llvm.x86.avx512.mask.range.pd.512(<8 x double>, <8 x double>, i32, <8 x double>, i8, i32)

define <8 x double>@test_int_x86_avx512_mask_range_pd_512(<8 x double> %x0, <8 x double> %x1, <8 x double> %x3, i8 %x4) {
; X86-LABEL: test_int_x86_avx512_mask_range_pd_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vrangepd $8, %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0x50,0xd1,0x08]
; X86-NEXT:    vrangepd $4, {sae}, %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf3,0xfd,0x18,0x50,0xc1,0x04]
; X86-NEXT:    vaddpd %zmm0, %zmm2, %zmm0 # encoding: [0x62,0xf1,0xed,0x48,0x58,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_range_pd_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vrangepd $8, %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0x50,0xd1,0x08]
; X64-NEXT:    vrangepd $4, {sae}, %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf3,0xfd,0x18,0x50,0xc1,0x04]
; X64-NEXT:    vaddpd %zmm0, %zmm2, %zmm0 # encoding: [0x62,0xf1,0xed,0x48,0x58,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <8 x double> @llvm.x86.avx512.mask.range.pd.512(<8 x double> %x0, <8 x double> %x1, i32 8, <8 x double> %x3, i8 %x4, i32 4)
  %res1 = call <8 x double> @llvm.x86.avx512.mask.range.pd.512(<8 x double> %x0, <8 x double> %x1, i32 4, <8 x double> %x3, i8 -1, i32 8)
  %res2 = fadd <8 x double> %res, %res1
  ret <8 x double> %res2
}

declare <16 x float> @llvm.x86.avx512.mask.range.ps.512(<16 x float>, <16 x float>, i32, <16 x float>, i16, i32)

define <16 x float>@test_int_x86_avx512_mask_range_ps_512(<16 x float> %x0, <16 x float> %x1, <16 x float> %x3, i16 %x4) {
; X86-LABEL: test_int_x86_avx512_mask_range_ps_512:
; X86:       # %bb.0:
; X86-NEXT:    kmovw {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf8,0x90,0x4c,0x24,0x04]
; X86-NEXT:    vrangeps $88, %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf3,0x7d,0x49,0x50,0xd1,0x58]
; X86-NEXT:    vrangeps $4, {sae}, %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf3,0x7d,0x18,0x50,0xc1,0x04]
; X86-NEXT:    vaddps %zmm0, %zmm2, %zmm0 # encoding: [0x62,0xf1,0x6c,0x48,0x58,0xc0]
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_range_ps_512:
; X64:       # %bb.0:
; X64-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-NEXT:    vrangeps $88, %zmm1, %zmm0, %zmm2 {%k1} # encoding: [0x62,0xf3,0x7d,0x49,0x50,0xd1,0x58]
; X64-NEXT:    vrangeps $4, {sae}, %zmm1, %zmm0, %zmm0 # encoding: [0x62,0xf3,0x7d,0x18,0x50,0xc1,0x04]
; X64-NEXT:    vaddps %zmm0, %zmm2, %zmm0 # encoding: [0x62,0xf1,0x6c,0x48,0x58,0xc0]
; X64-NEXT:    retq # encoding: [0xc3]
  %res = call <16 x float> @llvm.x86.avx512.mask.range.ps.512(<16 x float> %x0, <16 x float> %x1, i32 88, <16 x float> %x3, i16 %x4, i32 4)
  %res1 = call <16 x float> @llvm.x86.avx512.mask.range.ps.512(<16 x float> %x0, <16 x float> %x1, i32 4, <16 x float> %x3, i16 -1, i32 8)
  %res2 = fadd <16 x float> %res, %res1
  ret <16 x float> %res2
}

declare <4 x float> @llvm.x86.avx512.mask.reduce.ss(<4 x float>, <4 x float>,<4 x float>, i8, i32, i32)

define <4 x float>@test_int_x86_avx512_mask_reduce_ss(<4 x float> %x0, <4 x float> %x1, <4 x float> %x3, i8 %x4) {
; X86-AVX512DQ-LABEL: test_int_x86_avx512_mask_reduce_ss:
; X86-AVX512DQ:       # %bb.0:
; X86-AVX512DQ-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-AVX512DQ-NEXT:    vreducess $4, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0x7d,0x09,0x57,0xd1,0x04]
; X86-AVX512DQ-NEXT:    vreducess $4, {sae}, %xmm1, %xmm0, %xmm0 # encoding: [0x62,0xf3,0x7d,0x18,0x57,0xc1,0x04]
; X86-AVX512DQ-NEXT:    vaddps %xmm0, %xmm2, %xmm0 # encoding: [0xc5,0xe8,0x58,0xc0]
; X86-AVX512DQ-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512DQVL-LABEL: test_int_x86_avx512_mask_reduce_ss:
; X86-AVX512DQVL:       # %bb.0:
; X86-AVX512DQVL-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-AVX512DQVL-NEXT:    vreducess $4, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0x7d,0x09,0x57,0xd1,0x04]
; X86-AVX512DQVL-NEXT:    vreducess $4, {sae}, %xmm1, %xmm0, %xmm0 # encoding: [0x62,0xf3,0x7d,0x18,0x57,0xc1,0x04]
; X86-AVX512DQVL-NEXT:    vaddps %xmm0, %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe8,0x58,0xc0]
; X86-AVX512DQVL-NEXT:    retl # encoding: [0xc3]
;
; X64-AVX512DQ-LABEL: test_int_x86_avx512_mask_reduce_ss:
; X64-AVX512DQ:       # %bb.0:
; X64-AVX512DQ-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-AVX512DQ-NEXT:    vreducess $4, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0x7d,0x09,0x57,0xd1,0x04]
; X64-AVX512DQ-NEXT:    vreducess $4, {sae}, %xmm1, %xmm0, %xmm0 # encoding: [0x62,0xf3,0x7d,0x18,0x57,0xc1,0x04]
; X64-AVX512DQ-NEXT:    vaddps %xmm0, %xmm2, %xmm0 # encoding: [0xc5,0xe8,0x58,0xc0]
; X64-AVX512DQ-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512DQVL-LABEL: test_int_x86_avx512_mask_reduce_ss:
; X64-AVX512DQVL:       # %bb.0:
; X64-AVX512DQVL-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-AVX512DQVL-NEXT:    vreducess $4, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0x7d,0x09,0x57,0xd1,0x04]
; X64-AVX512DQVL-NEXT:    vreducess $4, {sae}, %xmm1, %xmm0, %xmm0 # encoding: [0x62,0xf3,0x7d,0x18,0x57,0xc1,0x04]
; X64-AVX512DQVL-NEXT:    vaddps %xmm0, %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe8,0x58,0xc0]
; X64-AVX512DQVL-NEXT:    retq # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.avx512.mask.reduce.ss(<4 x float> %x0, <4 x float> %x1, <4 x float> %x3, i8 %x4, i32 4, i32 4)
  %res1 = call <4 x float> @llvm.x86.avx512.mask.reduce.ss(<4 x float> %x0, <4 x float> %x1, <4 x float> %x3, i8 -1, i32 4, i32 8)
  %res2 = fadd <4 x float> %res, %res1
  ret <4 x float> %res2
}

declare <4 x float> @llvm.x86.avx512.mask.range.ss(<4 x float>, <4 x float>,<4 x float>, i8, i32, i32)

define <4 x float>@test_int_x86_avx512_mask_range_ss(<4 x float> %x0, <4 x float> %x1, <4 x float> %x3, i8 %x4) {
; X86-AVX512DQ-LABEL: test_int_x86_avx512_mask_range_ss:
; X86-AVX512DQ:       # %bb.0:
; X86-AVX512DQ-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-AVX512DQ-NEXT:    vrangess $4, {sae}, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0x7d,0x19,0x51,0xd1,0x04]
; X86-AVX512DQ-NEXT:    vrangess $4, {sae}, %xmm1, %xmm0, %xmm3 # encoding: [0x62,0xf3,0x7d,0x18,0x51,0xd9,0x04]
; X86-AVX512DQ-NEXT:    vaddps %xmm3, %xmm2, %xmm2 # encoding: [0xc5,0xe8,0x58,0xd3]
; X86-AVX512DQ-NEXT:    vrangess $4, %xmm1, %xmm0, %xmm0 # encoding: [0x62,0xf3,0x7d,0x08,0x51,0xc1,0x04]
; X86-AVX512DQ-NEXT:    vaddps %xmm2, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x58,0xc2]
; X86-AVX512DQ-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512DQVL-LABEL: test_int_x86_avx512_mask_range_ss:
; X86-AVX512DQVL:       # %bb.0:
; X86-AVX512DQVL-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-AVX512DQVL-NEXT:    vrangess $4, {sae}, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0x7d,0x19,0x51,0xd1,0x04]
; X86-AVX512DQVL-NEXT:    vrangess $4, {sae}, %xmm1, %xmm0, %xmm3 # encoding: [0x62,0xf3,0x7d,0x18,0x51,0xd9,0x04]
; X86-AVX512DQVL-NEXT:    vaddps %xmm3, %xmm2, %xmm2 # EVEX TO VEX Compression encoding: [0xc5,0xe8,0x58,0xd3]
; X86-AVX512DQVL-NEXT:    vrangess $4, %xmm1, %xmm0, %xmm0 # encoding: [0x62,0xf3,0x7d,0x08,0x51,0xc1,0x04]
; X86-AVX512DQVL-NEXT:    vaddps %xmm2, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x58,0xc2]
; X86-AVX512DQVL-NEXT:    retl # encoding: [0xc3]
;
; X64-AVX512DQ-LABEL: test_int_x86_avx512_mask_range_ss:
; X64-AVX512DQ:       # %bb.0:
; X64-AVX512DQ-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-AVX512DQ-NEXT:    vrangess $4, {sae}, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0x7d,0x19,0x51,0xd1,0x04]
; X64-AVX512DQ-NEXT:    vrangess $4, {sae}, %xmm1, %xmm0, %xmm3 # encoding: [0x62,0xf3,0x7d,0x18,0x51,0xd9,0x04]
; X64-AVX512DQ-NEXT:    vaddps %xmm3, %xmm2, %xmm2 # encoding: [0xc5,0xe8,0x58,0xd3]
; X64-AVX512DQ-NEXT:    vrangess $4, %xmm1, %xmm0, %xmm0 # encoding: [0x62,0xf3,0x7d,0x08,0x51,0xc1,0x04]
; X64-AVX512DQ-NEXT:    vaddps %xmm2, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x58,0xc2]
; X64-AVX512DQ-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512DQVL-LABEL: test_int_x86_avx512_mask_range_ss:
; X64-AVX512DQVL:       # %bb.0:
; X64-AVX512DQVL-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-AVX512DQVL-NEXT:    vrangess $4, {sae}, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0x7d,0x19,0x51,0xd1,0x04]
; X64-AVX512DQVL-NEXT:    vrangess $4, {sae}, %xmm1, %xmm0, %xmm3 # encoding: [0x62,0xf3,0x7d,0x18,0x51,0xd9,0x04]
; X64-AVX512DQVL-NEXT:    vaddps %xmm3, %xmm2, %xmm2 # EVEX TO VEX Compression encoding: [0xc5,0xe8,0x58,0xd3]
; X64-AVX512DQVL-NEXT:    vrangess $4, %xmm1, %xmm0, %xmm0 # encoding: [0x62,0xf3,0x7d,0x08,0x51,0xc1,0x04]
; X64-AVX512DQVL-NEXT:    vaddps %xmm2, %xmm0, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xf8,0x58,0xc2]
; X64-AVX512DQVL-NEXT:    retq # encoding: [0xc3]
  %res = call <4 x float> @llvm.x86.avx512.mask.range.ss(<4 x float> %x0, <4 x float> %x1, <4 x float> %x3, i8 %x4, i32 4, i32 8)
  %res1 = call <4 x float> @llvm.x86.avx512.mask.range.ss(<4 x float> %x0, <4 x float> %x1, <4 x float> %x3, i8 -1, i32 4, i32 8)
  %res2 = call <4 x float> @llvm.x86.avx512.mask.range.ss(<4 x float> %x0, <4 x float> %x1, <4 x float> %x3, i8 -1, i32 4, i32 4)
  %res3 = fadd <4 x float> %res, %res1
  %res4 = fadd <4 x float> %res2, %res3
  ret <4 x float> %res4
}

declare <2 x double> @llvm.x86.avx512.mask.reduce.sd(<2 x double>, <2 x double>,<2 x double>, i8, i32, i32)

define <2 x double>@test_int_x86_avx512_mask_reduce_sd(<2 x double> %x0, <2 x double> %x1, <2 x double> %x3, i8 %x4) {
; X86-AVX512DQ-LABEL: test_int_x86_avx512_mask_reduce_sd:
; X86-AVX512DQ:       # %bb.0:
; X86-AVX512DQ-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-AVX512DQ-NEXT:    vreducesd $4, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x09,0x57,0xd1,0x04]
; X86-AVX512DQ-NEXT:    vreducesd $4, {sae}, %xmm1, %xmm0, %xmm0 # encoding: [0x62,0xf3,0xfd,0x18,0x57,0xc1,0x04]
; X86-AVX512DQ-NEXT:    vaddpd %xmm0, %xmm2, %xmm0 # encoding: [0xc5,0xe9,0x58,0xc0]
; X86-AVX512DQ-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512DQVL-LABEL: test_int_x86_avx512_mask_reduce_sd:
; X86-AVX512DQVL:       # %bb.0:
; X86-AVX512DQVL-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-AVX512DQVL-NEXT:    vreducesd $4, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x09,0x57,0xd1,0x04]
; X86-AVX512DQVL-NEXT:    vreducesd $4, {sae}, %xmm1, %xmm0, %xmm0 # encoding: [0x62,0xf3,0xfd,0x18,0x57,0xc1,0x04]
; X86-AVX512DQVL-NEXT:    vaddpd %xmm0, %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe9,0x58,0xc0]
; X86-AVX512DQVL-NEXT:    retl # encoding: [0xc3]
;
; X64-AVX512DQ-LABEL: test_int_x86_avx512_mask_reduce_sd:
; X64-AVX512DQ:       # %bb.0:
; X64-AVX512DQ-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-AVX512DQ-NEXT:    vreducesd $4, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x09,0x57,0xd1,0x04]
; X64-AVX512DQ-NEXT:    vreducesd $4, {sae}, %xmm1, %xmm0, %xmm0 # encoding: [0x62,0xf3,0xfd,0x18,0x57,0xc1,0x04]
; X64-AVX512DQ-NEXT:    vaddpd %xmm0, %xmm2, %xmm0 # encoding: [0xc5,0xe9,0x58,0xc0]
; X64-AVX512DQ-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512DQVL-LABEL: test_int_x86_avx512_mask_reduce_sd:
; X64-AVX512DQVL:       # %bb.0:
; X64-AVX512DQVL-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-AVX512DQVL-NEXT:    vreducesd $4, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x09,0x57,0xd1,0x04]
; X64-AVX512DQVL-NEXT:    vreducesd $4, {sae}, %xmm1, %xmm0, %xmm0 # encoding: [0x62,0xf3,0xfd,0x18,0x57,0xc1,0x04]
; X64-AVX512DQVL-NEXT:    vaddpd %xmm0, %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe9,0x58,0xc0]
; X64-AVX512DQVL-NEXT:    retq # encoding: [0xc3]
  %res = call <2 x double> @llvm.x86.avx512.mask.reduce.sd(<2 x double> %x0, <2 x double> %x1, <2 x double> %x3, i8 %x4, i32 4, i32 4)
  %res1 = call <2 x double> @llvm.x86.avx512.mask.reduce.sd(<2 x double> %x0, <2 x double> %x1, <2 x double> %x3, i8 -1, i32 4, i32 8)
  %res2 = fadd <2 x double> %res, %res1
  ret <2 x double> %res2
}

declare <2 x double> @llvm.x86.avx512.mask.range.sd(<2 x double>, <2 x double>,<2 x double>, i8, i32, i32)

define <2 x double>@test_int_x86_avx512_mask_range_sd(<2 x double> %x0, <2 x double> %x1, <2 x double> %x3, i8 %x4) {
; X86-AVX512DQ-LABEL: test_int_x86_avx512_mask_range_sd:
; X86-AVX512DQ:       # %bb.0:
; X86-AVX512DQ-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-AVX512DQ-NEXT:    vrangesd $4, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x09,0x51,0xd1,0x04]
; X86-AVX512DQ-NEXT:    vrangesd $4, %xmm1, %xmm0, %xmm3 # encoding: [0x62,0xf3,0xfd,0x08,0x51,0xd9,0x04]
; X86-AVX512DQ-NEXT:    vrangesd $4, {sae}, %xmm1, %xmm0, %xmm0 # encoding: [0x62,0xf3,0xfd,0x18,0x51,0xc1,0x04]
; X86-AVX512DQ-NEXT:    vaddpd %xmm0, %xmm2, %xmm0 # encoding: [0xc5,0xe9,0x58,0xc0]
; X86-AVX512DQ-NEXT:    vaddpd %xmm0, %xmm3, %xmm0 # encoding: [0xc5,0xe1,0x58,0xc0]
; X86-AVX512DQ-NEXT:    retl # encoding: [0xc3]
;
; X86-AVX512DQVL-LABEL: test_int_x86_avx512_mask_range_sd:
; X86-AVX512DQVL:       # %bb.0:
; X86-AVX512DQVL-NEXT:    kmovb {{[0-9]+}}(%esp), %k1 # encoding: [0xc5,0xf9,0x90,0x4c,0x24,0x04]
; X86-AVX512DQVL-NEXT:    vrangesd $4, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x09,0x51,0xd1,0x04]
; X86-AVX512DQVL-NEXT:    vrangesd $4, %xmm1, %xmm0, %xmm3 # encoding: [0x62,0xf3,0xfd,0x08,0x51,0xd9,0x04]
; X86-AVX512DQVL-NEXT:    vrangesd $4, {sae}, %xmm1, %xmm0, %xmm0 # encoding: [0x62,0xf3,0xfd,0x18,0x51,0xc1,0x04]
; X86-AVX512DQVL-NEXT:    vaddpd %xmm0, %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe9,0x58,0xc0]
; X86-AVX512DQVL-NEXT:    vaddpd %xmm0, %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe1,0x58,0xc0]
; X86-AVX512DQVL-NEXT:    retl # encoding: [0xc3]
;
; X64-AVX512DQ-LABEL: test_int_x86_avx512_mask_range_sd:
; X64-AVX512DQ:       # %bb.0:
; X64-AVX512DQ-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-AVX512DQ-NEXT:    vrangesd $4, %xmm1, %xmm0, %xmm3 # encoding: [0x62,0xf3,0xfd,0x08,0x51,0xd9,0x04]
; X64-AVX512DQ-NEXT:    vrangesd $4, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x09,0x51,0xd1,0x04]
; X64-AVX512DQ-NEXT:    vrangesd $4, {sae}, %xmm1, %xmm0, %xmm0 # encoding: [0x62,0xf3,0xfd,0x18,0x51,0xc1,0x04]
; X64-AVX512DQ-NEXT:    vaddpd %xmm0, %xmm2, %xmm0 # encoding: [0xc5,0xe9,0x58,0xc0]
; X64-AVX512DQ-NEXT:    vaddpd %xmm0, %xmm3, %xmm0 # encoding: [0xc5,0xe1,0x58,0xc0]
; X64-AVX512DQ-NEXT:    retq # encoding: [0xc3]
;
; X64-AVX512DQVL-LABEL: test_int_x86_avx512_mask_range_sd:
; X64-AVX512DQVL:       # %bb.0:
; X64-AVX512DQVL-NEXT:    kmovw %edi, %k1 # encoding: [0xc5,0xf8,0x92,0xcf]
; X64-AVX512DQVL-NEXT:    vrangesd $4, %xmm1, %xmm0, %xmm3 # encoding: [0x62,0xf3,0xfd,0x08,0x51,0xd9,0x04]
; X64-AVX512DQVL-NEXT:    vrangesd $4, %xmm1, %xmm0, %xmm2 {%k1} # encoding: [0x62,0xf3,0xfd,0x09,0x51,0xd1,0x04]
; X64-AVX512DQVL-NEXT:    vrangesd $4, {sae}, %xmm1, %xmm0, %xmm0 # encoding: [0x62,0xf3,0xfd,0x18,0x51,0xc1,0x04]
; X64-AVX512DQVL-NEXT:    vaddpd %xmm0, %xmm2, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe9,0x58,0xc0]
; X64-AVX512DQVL-NEXT:    vaddpd %xmm0, %xmm3, %xmm0 # EVEX TO VEX Compression encoding: [0xc5,0xe1,0x58,0xc0]
; X64-AVX512DQVL-NEXT:    retq # encoding: [0xc3]
  %res = call <2 x double> @llvm.x86.avx512.mask.range.sd(<2 x double> %x0, <2 x double> %x1, <2 x double> %x3, i8 %x4, i32 4, i32 4)
  %res1 = call <2 x double> @llvm.x86.avx512.mask.range.sd(<2 x double> %x0, <2 x double> %x1, <2 x double> %x3, i8 -1, i32 4, i32 8)
  %res2 = call <2 x double> @llvm.x86.avx512.mask.range.sd(<2 x double> %x0, <2 x double> %x1, <2 x double> %x3, i8 -1, i32 4, i32 4)
  %res3 = fadd <2 x double> %res, %res1
  %res4 = fadd <2 x double> %res2, %res3
  ret <2 x double> %res4
}

declare <8 x i1> @llvm.x86.avx512.fpclass.pd.512(<8 x double>, i32)

define i8 @test_int_x86_avx512_fpclass_pd_512(<8 x double> %x0) {
; CHECK-LABEL: test_int_x86_avx512_fpclass_pd_512:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vfpclasspd $2, %zmm0, %k1 # encoding: [0x62,0xf3,0xfd,0x48,0x66,0xc8,0x02]
; CHECK-NEXT:    vfpclasspd $4, %zmm0, %k0 {%k1} # encoding: [0x62,0xf3,0xfd,0x49,0x66,0xc0,0x04]
; CHECK-NEXT:    kmovw %k0, %eax # encoding: [0xc5,0xf8,0x93,0xc0]
; CHECK-NEXT:    # kill: def $al killed $al killed $eax
; CHECK-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <8 x i1> @llvm.x86.avx512.fpclass.pd.512(<8 x double> %x0, i32 4)
  %res1 = call <8 x i1> @llvm.x86.avx512.fpclass.pd.512(<8 x double> %x0, i32 2)
  %1 = and <8 x i1> %res1, %res
  %2 = bitcast <8 x i1> %1 to i8
  ret i8 %2
}
declare <16 x i1> @llvm.x86.avx512.fpclass.ps.512(<16 x float>, i32)

define i16@test_int_x86_avx512_fpclass_ps_512(<16 x float> %x0) {
; CHECK-LABEL: test_int_x86_avx512_fpclass_ps_512:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vfpclassps $2, %zmm0, %k1 # encoding: [0x62,0xf3,0x7d,0x48,0x66,0xc8,0x02]
; CHECK-NEXT:    vfpclassps $4, %zmm0, %k0 {%k1} # encoding: [0x62,0xf3,0x7d,0x49,0x66,0xc0,0x04]
; CHECK-NEXT:    kmovw %k0, %eax # encoding: [0xc5,0xf8,0x93,0xc0]
; CHECK-NEXT:    # kill: def $ax killed $ax killed $eax
; CHECK-NEXT:    vzeroupper # encoding: [0xc5,0xf8,0x77]
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call <16 x i1> @llvm.x86.avx512.fpclass.ps.512(<16 x float> %x0, i32 4)
  %res1 = call <16 x i1> @llvm.x86.avx512.fpclass.ps.512(<16 x float> %x0, i32 2)
  %1 = and <16 x i1> %res1, %res
  %2 = bitcast <16 x i1> %1 to i16
  ret i16 %2
}

declare i8 @llvm.x86.avx512.mask.fpclass.sd(<2 x double>, i32, i8)

define i8 @test_int_x86_avx512_mask_fpclass_sd(<2 x double> %x0) {
; CHECK-LABEL: test_int_x86_avx512_mask_fpclass_sd:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vfpclasssd $4, %xmm0, %k1 # encoding: [0x62,0xf3,0xfd,0x08,0x67,0xc8,0x04]
; CHECK-NEXT:    vfpclasssd $2, %xmm0, %k0 {%k1} # encoding: [0x62,0xf3,0xfd,0x09,0x67,0xc0,0x02]
; CHECK-NEXT:    kmovw %k0, %eax # encoding: [0xc5,0xf8,0x93,0xc0]
; CHECK-NEXT:    # kill: def $al killed $al killed $eax
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i8 @llvm.x86.avx512.mask.fpclass.sd(<2 x double> %x0, i32 2, i8 -1)
  %res1 = call i8 @llvm.x86.avx512.mask.fpclass.sd(<2 x double> %x0, i32 4, i8 %res)
  ret i8 %res1
}

define i8 @test_int_x86_avx512_mask_fpclass_sd_load(<2 x double>* %x0ptr) {
; X86-LABEL: test_int_x86_avx512_mask_fpclass_sd_load:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vfpclasssd $4, (%eax), %k0 # encoding: [0x62,0xf3,0xfd,0x08,0x67,0x00,0x04]
; X86-NEXT:    kmovw %k0, %eax # encoding: [0xc5,0xf8,0x93,0xc0]
; X86-NEXT:    # kill: def $al killed $al killed $eax
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_fpclass_sd_load:
; X64:       # %bb.0:
; X64-NEXT:    vfpclasssd $4, (%rdi), %k0 # encoding: [0x62,0xf3,0xfd,0x08,0x67,0x07,0x04]
; X64-NEXT:    kmovw %k0, %eax # encoding: [0xc5,0xf8,0x93,0xc0]
; X64-NEXT:    # kill: def $al killed $al killed $eax
; X64-NEXT:    retq # encoding: [0xc3]
  %x0 = load <2 x double>, <2 x double>* %x0ptr
  %res = call i8 @llvm.x86.avx512.mask.fpclass.sd(<2 x double> %x0, i32 4, i8 -1)
  ret i8 %res
}

declare i8 @llvm.x86.avx512.mask.fpclass.ss(<4 x float>, i32, i8)

define i8 @test_int_x86_avx512_mask_fpclass_ss(<4 x float> %x0) {
; CHECK-LABEL: test_int_x86_avx512_mask_fpclass_ss:
; CHECK:       # %bb.0:
; CHECK-NEXT:    vfpclassss $4, %xmm0, %k1 # encoding: [0x62,0xf3,0x7d,0x08,0x67,0xc8,0x04]
; CHECK-NEXT:    vfpclassss $2, %xmm0, %k0 {%k1} # encoding: [0x62,0xf3,0x7d,0x09,0x67,0xc0,0x02]
; CHECK-NEXT:    kmovw %k0, %eax # encoding: [0xc5,0xf8,0x93,0xc0]
; CHECK-NEXT:    # kill: def $al killed $al killed $eax
; CHECK-NEXT:    ret{{[l|q]}} # encoding: [0xc3]
  %res = call i8 @llvm.x86.avx512.mask.fpclass.ss(<4 x float> %x0, i32 2, i8 -1)
  %res1 = call i8 @llvm.x86.avx512.mask.fpclass.ss(<4 x float> %x0, i32 4, i8 %res)
  ret i8 %res1
}

define i8 @test_int_x86_avx512_mask_fpclass_ss_load(<4 x float>* %x0ptr, i8 %x1) {
; X86-LABEL: test_int_x86_avx512_mask_fpclass_ss_load:
; X86:       # %bb.0:
; X86-NEXT:    movl {{[0-9]+}}(%esp), %eax # encoding: [0x8b,0x44,0x24,0x04]
; X86-NEXT:    vfpclassss $4, (%eax), %k0 # encoding: [0x62,0xf3,0x7d,0x08,0x67,0x00,0x04]
; X86-NEXT:    kmovw %k0, %eax # encoding: [0xc5,0xf8,0x93,0xc0]
; X86-NEXT:    # kill: def $al killed $al killed $eax
; X86-NEXT:    retl # encoding: [0xc3]
;
; X64-LABEL: test_int_x86_avx512_mask_fpclass_ss_load:
; X64:       # %bb.0:
; X64-NEXT:    vfpclassss $4, (%rdi), %k0 # encoding: [0x62,0xf3,0x7d,0x08,0x67,0x07,0x04]
; X64-NEXT:    kmovw %k0, %eax # encoding: [0xc5,0xf8,0x93,0xc0]
; X64-NEXT:    # kill: def $al killed $al killed $eax
; X64-NEXT:    retq # encoding: [0xc3]
  %x0 = load <4 x float>, <4 x float>* %x0ptr
  %res = call i8 @llvm.x86.avx512.mask.fpclass.ss(<4 x float> %x0, i32 4, i8 -1)
  ret i8 %res
}