maxjhandsome
/
llvm-project
forked from OSchip/llvm-project
1
0
Fork 0
Code Issues Pull Requests Packages Releases Wiki Activity
llvm-project/llvm/test/CodeGen/X86/avx512dq-intrinsics.ll

819 lines
48 KiB
LLVM
Raw Normal View History

[AVX512] Use update_llc_test_checks.py to regenerate a test in preparation for a future commit. llvm-svn: 273252
2016-06-21 15:37:27 +08:00
; NOTE: Assertions have been autogenerated by utils/update_llc_test_checks.py
[X86][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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
AVX-512: Added intrinsics for VCVT* instructions. All SKX forms. All VCVT instructions for float/double/int/long types. Differential Revision: http://reviews.llvm.org/D11343 llvm-svn: 242877
2015-07-22 16:56:00 +08:00
[X86] Add intrinsics for KADD instructions These are intrinsics for supporting kadd builtins in clang. These builtins are already in gcc to implement intrinsics from icc. Though they are missing from the Intel Intrinsics Guide. This instruction adds two mask registers together as if they were scalar rather than a vXi1. We might be able to get away with a bitcast to scalar and a normal add instruction, but that would require DAG combine smarts in the backend to recoqnize add+bitcast. For now I'd prefer to go with the easiest implementation so we can get these builtins in to clang with good codegen. Differential Revision: https://reviews.llvm.org/D51370 llvm-svn: 340869
2018-08-29 03:22:55 +08:00
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>)
[X86] Add intrinsics for KTEST instructions. These intrinsics use the same implementation as PTEST intrinsics, but use vXi1 vectors. New clang builtins will be accompanying them shortly. llvm-svn: 341259
2018-09-01 05:31:53 +08:00
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
AVX-512: Added intrinsics for VCVT* instructions. All SKX forms. All VCVT instructions for float/double/int/long types. Differential Revision: http://reviews.llvm.org/D11343 llvm-svn: 242877
2015-07-22 16:56:00 +08:00
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][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
AVX-512: Added intrinsics for VCVT* instructions. All SKX forms. All VCVT instructions for float/double/int/long types. Differential Revision: http://reviews.llvm.org/D11343 llvm-svn: 242877
2015-07-22 16:56:00 +08:00
%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][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
AVX-512: Added intrinsics for VCVT* instructions. All SKX forms. All VCVT instructions for float/double/int/long types. Differential Revision: http://reviews.llvm.org/D11343 llvm-svn: 242877
2015-07-22 16:56:00 +08:00
%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][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
AVX-512: Added intrinsics for VCVT* instructions. All SKX forms. All VCVT instructions for float/double/int/long types. Differential Revision: http://reviews.llvm.org/D11343 llvm-svn: 242877
2015-07-22 16:56:00 +08:00
%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][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
AVX-512: Added intrinsics for VCVT* instructions. All SKX forms. All VCVT instructions for float/double/int/long types. Differential Revision: http://reviews.llvm.org/D11343 llvm-svn: 242877
2015-07-22 16:56:00 +08:00
%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][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
AVX-512: Added intrinsics for VCVT* instructions. All SKX forms. All VCVT instructions for float/double/int/long types. Differential Revision: http://reviews.llvm.org/D11343 llvm-svn: 242877
2015-07-22 16:56:00 +08:00
%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][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
AVX-512: Added intrinsics for VCVT* instructions. All SKX forms. All VCVT instructions for float/double/int/long types. Differential Revision: http://reviews.llvm.org/D11343 llvm-svn: 242877
2015-07-22 16:56:00 +08:00
%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][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
AVX-512: Added intrinsics for VCVT* instructions. All SKX forms. All VCVT instructions for float/double/int/long types. Differential Revision: http://reviews.llvm.org/D11343 llvm-svn: 242877
2015-07-22 16:56:00 +08:00
%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][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
AVX-512: Added intrinsics for VCVT* instructions. All SKX forms. All VCVT instructions for float/double/int/long types. Differential Revision: http://reviews.llvm.org/D11343 llvm-svn: 242877
2015-07-22 16:56:00 +08:00
%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][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
AVX-512: Added intrinsics for VCVT* instructions. All SKX forms. All VCVT instructions for float/double/int/long types. Differential Revision: http://reviews.llvm.org/D11343 llvm-svn: 242877
2015-07-22 16:56:00 +08:00
%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][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
AVX-512: Added intrinsics for VCVT* instructions. All SKX forms. All VCVT instructions for float/double/int/long types. Differential Revision: http://reviews.llvm.org/D11343 llvm-svn: 242877
2015-07-22 16:56:00 +08:00
%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][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
AVX-512: Added intrinsics for VCVT* instructions. All SKX forms. All VCVT instructions for float/double/int/long types. Differential Revision: http://reviews.llvm.org/D11343 llvm-svn: 242877
2015-07-22 16:56:00 +08:00
%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][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
AVX-512: Added intrinsics for VCVT* instructions. All SKX forms. All VCVT instructions for float/double/int/long types. Differential Revision: http://reviews.llvm.org/D11343 llvm-svn: 242877
2015-07-22 16:56:00 +08:00
%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
}
[X86][AVX512] add reduce/range/scalef/rndScale include encoding and intrinsics Differential Revision: http://reviews.llvm.org/D11222 llvm-svn: 242896
2015-07-22 20:00:43 +08:00
declare <8 x double> @llvm.x86.avx512.mask.reduce.pd.512(<8 x double>, i32, <8 x double>, i8, i32)
[AVX512] Use update_llc_test_checks.py to regenerate a test in preparation for a future commit. llvm-svn: 273252
2016-06-21 15:37:27 +08:00
[X86][AVX512] add reduce/range/scalef/rndScale include encoding and intrinsics Differential Revision: http://reviews.llvm.org/D11222 llvm-svn: 242896
2015-07-22 20:00:43 +08:00
define <8 x double>@test_int_x86_avx512_mask_reduce_pd_512(<8 x double> %x0, <8 x double> %x2, i8 %x3) {
[X86][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
[X86][AVX512] add reduce/range/scalef/rndScale include encoding and intrinsics Differential Revision: http://reviews.llvm.org/D11222 llvm-svn: 242896
2015-07-22 20:00:43 +08:00
%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)
[AVX512] Use update_llc_test_checks.py to regenerate a test in preparation for a future commit. llvm-svn: 273252
2016-06-21 15:37:27 +08:00
[X86][AVX512] add reduce/range/scalef/rndScale include encoding and intrinsics Differential Revision: http://reviews.llvm.org/D11222 llvm-svn: 242896
2015-07-22 20:00:43 +08:00
define <16 x float>@test_int_x86_avx512_mask_reduce_ps_512(<16 x float> %x0, <16 x float> %x2, i16 %x3) {
[X86][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
[X86][AVX512] add reduce/range/scalef/rndScale include encoding and intrinsics Differential Revision: http://reviews.llvm.org/D11222 llvm-svn: 242896
2015-07-22 20:00:43 +08:00
%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)
[AVX512] Use update_llc_test_checks.py to regenerate a test in preparation for a future commit. llvm-svn: 273252
2016-06-21 15:37:27 +08:00
[X86][AVX512] add reduce/range/scalef/rndScale include encoding and intrinsics Differential Revision: http://reviews.llvm.org/D11222 llvm-svn: 242896
2015-07-22 20:00:43 +08:00
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][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
[X86][AVX512] add reduce/range/scalef/rndScale include encoding and intrinsics Differential Revision: http://reviews.llvm.org/D11222 llvm-svn: 242896
2015-07-22 20:00:43 +08:00
%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][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
[X86][AVX512] add reduce/range/scalef/rndScale include encoding and intrinsics Differential Revision: http://reviews.llvm.org/D11222 llvm-svn: 242896
2015-07-22 20:00:43 +08:00
%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][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
[X86][AVX512] add reduce/range/scalef/rndScale include encoding and intrinsics Differential Revision: http://reviews.llvm.org/D11222 llvm-svn: 242896
2015-07-22 20:00:43 +08:00
%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)
[AVX512] Use update_llc_test_checks.py to regenerate a test in preparation for a future commit. llvm-svn: 273252
2016-06-21 15:37:27 +08:00
[X86][AVX512] add reduce/range/scalef/rndScale include encoding and intrinsics Differential Revision: http://reviews.llvm.org/D11222 llvm-svn: 242896
2015-07-22 20:00:43 +08:00
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][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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] Merge the FR128 and VR128 regclass since they have identical spill and alignment characteristics. This unfortunately requires a bunch of bitcasts to be added added to SUBREG_TO_REG, COPY_TO_REGCLASS, and instructions in output patterns. Otherwise tablegen seems to default to picking f128 and then we fail when something tries to get the register class for f128 which isn't always valid. The test changes are because we were previously mixing fr128 and vr128 due to contrainRegClass finding FR128 first and passes like live range shrinking weren't handling that well. llvm-svn: 337147
2018-07-16 14:56:09 +08:00
; X86-AVX512DQ-NEXT: vaddps %xmm3, %xmm2, %xmm2 # encoding: [0xc5,0xe8,0x58,0xd3]
[X86][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; X86-AVX512DQ-NEXT: vrangess $4, %xmm1, %xmm0, %xmm0 # encoding: [0x62,0xf3,0x7d,0x08,0x51,0xc1,0x04]
[X86] Merge the FR128 and VR128 regclass since they have identical spill and alignment characteristics. This unfortunately requires a bunch of bitcasts to be added added to SUBREG_TO_REG, COPY_TO_REGCLASS, and instructions in output patterns. Otherwise tablegen seems to default to picking f128 and then we fail when something tries to get the register class for f128 which isn't always valid. The test changes are because we were previously mixing fr128 and vr128 due to contrainRegClass finding FR128 first and passes like live range shrinking weren't handling that well. llvm-svn: 337147
2018-07-16 14:56:09 +08:00
; X86-AVX512DQ-NEXT: vaddps %xmm2, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x58,0xc2]
[X86][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
[X86] Merge the FR128 and VR128 regclass since they have identical spill and alignment characteristics. This unfortunately requires a bunch of bitcasts to be added added to SUBREG_TO_REG, COPY_TO_REGCLASS, and instructions in output patterns. Otherwise tablegen seems to default to picking f128 and then we fail when something tries to get the register class for f128 which isn't always valid. The test changes are because we were previously mixing fr128 and vr128 due to contrainRegClass finding FR128 first and passes like live range shrinking weren't handling that well. llvm-svn: 337147
2018-07-16 14:56:09 +08:00
; X64-AVX512DQ-NEXT: vaddps %xmm3, %xmm2, %xmm2 # encoding: [0xc5,0xe8,0x58,0xd3]
[X86][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; X64-AVX512DQ-NEXT: vrangess $4, %xmm1, %xmm0, %xmm0 # encoding: [0x62,0xf3,0x7d,0x08,0x51,0xc1,0x04]
[X86] Merge the FR128 and VR128 regclass since they have identical spill and alignment characteristics. This unfortunately requires a bunch of bitcasts to be added added to SUBREG_TO_REG, COPY_TO_REGCLASS, and instructions in output patterns. Otherwise tablegen seems to default to picking f128 and then we fail when something tries to get the register class for f128 which isn't always valid. The test changes are because we were previously mixing fr128 and vr128 due to contrainRegClass finding FR128 first and passes like live range shrinking weren't handling that well. llvm-svn: 337147
2018-07-16 14:56:09 +08:00
; X64-AVX512DQ-NEXT: vaddps %xmm2, %xmm0, %xmm0 # encoding: [0xc5,0xf8,0x58,0xc2]
[X86][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
[X86][AVX512] add reduce/range/scalef/rndScale include encoding and intrinsics Differential Revision: http://reviews.llvm.org/D11222 llvm-svn: 242896
2015-07-22 20:00:43 +08:00
%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)
[X86] Add test cases and command lines demonstrating how we accidentally select vrangeps/vrangepd from vrangess/vrangesd instrinsics when the rounding mode is CUR_DIRECTION llvm-svn: 317998
2017-11-13 02:51:08 +08:00
%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
[X86][AVX512] add reduce/range/scalef/rndScale include encoding and intrinsics Differential Revision: http://reviews.llvm.org/D11222 llvm-svn: 242896
2015-07-22 20:00:43 +08:00
}
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][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
[X86][AVX512] add reduce/range/scalef/rndScale include encoding and intrinsics Differential Revision: http://reviews.llvm.org/D11222 llvm-svn: 242896
2015-07-22 20:00:43 +08:00
%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)
[AVX512] Use update_llc_test_checks.py to regenerate a test in preparation for a future commit. llvm-svn: 273252
2016-06-21 15:37:27 +08:00
[X86][AVX512] add reduce/range/scalef/rndScale include encoding and intrinsics Differential Revision: http://reviews.llvm.org/D11222 llvm-svn: 242896
2015-07-22 20:00:43 +08:00
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][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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] In getScalarMaskingNode, replace scalar_to_vector with a bitcast to v8i1 and an extract_subvector to convert i8 to v1i1. The bitcast can be nicely merged with any i8 loads that exist for argument passing in 32 mode for example. llvm-svn: 347380
2018-11-21 15:01:22 +08:00
; X86-AVX512DQ-NEXT: vrangesd $4, %xmm1, %xmm0, %xmm3 # encoding: [0x62,0xf3,0xfd,0x08,0x51,0xd9,0x04]
[X86][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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] In getScalarMaskingNode, replace scalar_to_vector with a bitcast to v8i1 and an extract_subvector to convert i8 to v1i1. The bitcast can be nicely merged with any i8 loads that exist for argument passing in 32 mode for example. llvm-svn: 347380
2018-11-21 15:01:22 +08:00
; X86-AVX512DQVL-NEXT: vrangesd $4, %xmm1, %xmm0, %xmm3 # encoding: [0x62,0xf3,0xfd,0x08,0x51,0xd9,0x04]
[X86][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
[X86] In getScalarMaskingNode, replace scalar_to_vector with a bitcast to v8i1 and an extract_subvector to convert i8 to v1i1. The bitcast can be nicely merged with any i8 loads that exist for argument passing in 32 mode for example. llvm-svn: 347380
2018-11-21 15:01:22 +08:00
; X64-AVX512DQ-NEXT: vrangesd $4, %xmm1, %xmm0, %xmm3 # encoding: [0x62,0xf3,0xfd,0x08,0x51,0xd9,0x04]
[X86][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
[X86] In getScalarMaskingNode, replace scalar_to_vector with a bitcast to v8i1 and an extract_subvector to convert i8 to v1i1. The bitcast can be nicely merged with any i8 loads that exist for argument passing in 32 mode for example. llvm-svn: 347380
2018-11-21 15:01:22 +08:00
; X64-AVX512DQVL-NEXT: vrangesd $4, %xmm1, %xmm0, %xmm3 # encoding: [0x62,0xf3,0xfd,0x08,0x51,0xd9,0x04]
[X86][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
[X86][AVX512] add reduce/range/scalef/rndScale include encoding and intrinsics Differential Revision: http://reviews.llvm.org/D11222 llvm-svn: 242896
2015-07-22 20:00:43 +08:00
%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)
[X86] Add test cases and command lines demonstrating how we accidentally select vrangeps/vrangepd from vrangess/vrangesd instrinsics when the rounding mode is CUR_DIRECTION llvm-svn: 317998
2017-11-13 02:51:08 +08:00
%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
[X86][AVX512] add reduce/range/scalef/rndScale include encoding and intrinsics Differential Revision: http://reviews.llvm.org/D11222 llvm-svn: 242896
2015-07-22 20:00:43 +08:00
}
AVX512: Implemented encoding and intrinsics for vextracti64x4 ,vextracti64x2, vextracti32x8, vextracti32x4, vextractf64x4, vextractf64x2, vextractf32x8, vextractf32x4 Added tests for intrinsics and encoding. Differential Revision: http://reviews.llvm.org/D11802 llvm-svn: 247276
2015-09-10 20:54:54 +08:00
[X86] Rename the autoupgraded of packed fp compare and fpclass intrinsics that don't take a mask as input to exclude '.mask.' from their name. I think the intrinsics named 'avx512.mask.' should refer to the previous behavior of taking a mask argument in the intrinsic instead of using a 'select' or 'and' instruction in IR to accomplish the masking. This is more consistent with the goal that eventually we will have no intrinsics that have masking builtin. When we reach that goal, we should have no intrinsics named "avx512.mask". llvm-svn: 335744
2018-06-27 23:57:53 +08:00
declare <8 x i1> @llvm.x86.avx512.fpclass.pd.512(<8 x double>, i32)
[X86][AVX512DQ] Add fpclass instruction Differential Revision: http://reviews.llvm.org/D12931 llvm-svn: 248115
2015-09-20 16:46:07 +08:00
[X86] Rename the autoupgraded of packed fp compare and fpclass intrinsics that don't take a mask as input to exclude '.mask.' from their name. I think the intrinsics named 'avx512.mask.' should refer to the previous behavior of taking a mask argument in the intrinsic instead of using a 'select' or 'and' instruction in IR to accomplish the masking. This is more consistent with the goal that eventually we will have no intrinsics that have masking builtin. When we reach that goal, we should have no intrinsics named "avx512.mask". llvm-svn: 335744
2018-06-27 23:57:53 +08:00
define i8 @test_int_x86_avx512_fpclass_pd_512(<8 x double> %x0) {
; CHECK-LABEL: test_int_x86_avx512_fpclass_pd_512:
[X86] Update fpclass intrinsic tests to chain their calls to the intrinsic rather than joining them with add. The test cases try to test masked and unmasked isntructions at the same time. Previously the masked version relies on an extra fucntion parameter. Then the two results were combined with 'add'. This patch gets rid of the second parameter and just passes the result of the first intrinsic into the mask argument of the second call. Then there's no need for an 'add'. This configuration works a lot better with an upcoming patch to redefine the intrinsics to use vXi1 types for the output and mask argument. llvm-svn: 335551
2018-06-26 07:29:47 +08:00
; 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]
[X86] Rename the autoupgraded of packed fp compare and fpclass intrinsics that don't take a mask as input to exclude '.mask.' from their name. I think the intrinsics named 'avx512.mask.' should refer to the previous behavior of taking a mask argument in the intrinsic instead of using a 'select' or 'and' instruction in IR to accomplish the masking. This is more consistent with the goal that eventually we will have no intrinsics that have masking builtin. When we reach that goal, we should have no intrinsics named "avx512.mask". llvm-svn: 335744
2018-06-27 23:57:53 +08:00
%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)
[X86] Redefine avx512 packed fpclass intrinsics to return a vXi1 mask and implement the mask input argument using an 'and' IR instruction. This recommits r335562 and 335563 as a single commit. The frontend will surround the intrinsic with the appropriate marshalling to/from a scalar type to match the sigature of the builtin that software expects. By exposing the vXi1 type directly in the llvm intrinsic we make it available to optimizers much earlier. This can enable the scalar marshalling code to be optimized away. llvm-svn: 335568
2018-06-26 09:37:02 +08:00
%1 = and <8 x i1> %res1, %res
%2 = bitcast <8 x i1> %1 to i8
ret i8 %2
[X86][AVX512DQ] Add fpclass instruction Differential Revision: http://reviews.llvm.org/D12931 llvm-svn: 248115
2015-09-20 16:46:07 +08:00
}
[X86] Rename the autoupgraded of packed fp compare and fpclass intrinsics that don't take a mask as input to exclude '.mask.' from their name. I think the intrinsics named 'avx512.mask.' should refer to the previous behavior of taking a mask argument in the intrinsic instead of using a 'select' or 'and' instruction in IR to accomplish the masking. This is more consistent with the goal that eventually we will have no intrinsics that have masking builtin. When we reach that goal, we should have no intrinsics named "avx512.mask". llvm-svn: 335744
2018-06-27 23:57:53 +08:00
declare <16 x i1> @llvm.x86.avx512.fpclass.ps.512(<16 x float>, i32)
[X86][AVX512DQ] Add fpclass instruction Differential Revision: http://reviews.llvm.org/D12931 llvm-svn: 248115
2015-09-20 16:46:07 +08:00
[X86] Rename the autoupgraded of packed fp compare and fpclass intrinsics that don't take a mask as input to exclude '.mask.' from their name. I think the intrinsics named 'avx512.mask.' should refer to the previous behavior of taking a mask argument in the intrinsic instead of using a 'select' or 'and' instruction in IR to accomplish the masking. This is more consistent with the goal that eventually we will have no intrinsics that have masking builtin. When we reach that goal, we should have no intrinsics named "avx512.mask". llvm-svn: 335744
2018-06-27 23:57:53 +08:00
define i16@test_int_x86_avx512_fpclass_ps_512(<16 x float> %x0) {
; CHECK-LABEL: test_int_x86_avx512_fpclass_ps_512:
[X86] Update fpclass intrinsic tests to chain their calls to the intrinsic rather than joining them with add. The test cases try to test masked and unmasked isntructions at the same time. Previously the masked version relies on an extra fucntion parameter. Then the two results were combined with 'add'. This patch gets rid of the second parameter and just passes the result of the first intrinsic into the mask argument of the second call. Then there's no need for an 'add'. This configuration works a lot better with an upcoming patch to redefine the intrinsics to use vXi1 types for the output and mask argument. llvm-svn: 335551
2018-06-26 07:29:47 +08:00
; 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]
[X86] Rename the autoupgraded of packed fp compare and fpclass intrinsics that don't take a mask as input to exclude '.mask.' from their name. I think the intrinsics named 'avx512.mask.' should refer to the previous behavior of taking a mask argument in the intrinsic instead of using a 'select' or 'and' instruction in IR to accomplish the masking. This is more consistent with the goal that eventually we will have no intrinsics that have masking builtin. When we reach that goal, we should have no intrinsics named "avx512.mask". llvm-svn: 335744
2018-06-27 23:57:53 +08:00
%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)
[X86] Redefine avx512 packed fpclass intrinsics to return a vXi1 mask and implement the mask input argument using an 'and' IR instruction. This recommits r335562 and 335563 as a single commit. The frontend will surround the intrinsic with the appropriate marshalling to/from a scalar type to match the sigature of the builtin that software expects. By exposing the vXi1 type directly in the llvm intrinsic we make it available to optimizers much earlier. This can enable the scalar marshalling code to be optimized away. llvm-svn: 335568
2018-06-26 09:37:02 +08:00
%1 = and <16 x i1> %res1, %res
%2 = bitcast <16 x i1> %1 to i16
ret i16 %2
[X86][AVX512DQ] Add fpclass instruction Differential Revision: http://reviews.llvm.org/D12931 llvm-svn: 248115
2015-09-20 16:46:07 +08:00
}
[X86][AVX512DQ] add scalar fpclass Differential Revision: http://reviews.llvm.org/D13769 llvm-svn: 250650
2015-10-18 19:04:38 +08:00
declare i8 @llvm.x86.avx512.mask.fpclass.sd(<2 x double>, i32, i8)
[X86] Update fpclass intrinsic tests to chain their calls to the intrinsic rather than joining them with add. The test cases try to test masked and unmasked isntructions at the same time. Previously the masked version relies on an extra fucntion parameter. Then the two results were combined with 'add'. This patch gets rid of the second parameter and just passes the result of the first intrinsic into the mask argument of the second call. Then there's no need for an 'add'. This configuration works a lot better with an upcoming patch to redefine the intrinsics to use vXi1 types for the output and mask argument. llvm-svn: 335551
2018-06-26 07:29:47 +08:00
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:
[X86] In getScalarMaskingNode, replace scalar_to_vector with a bitcast to v8i1 and an extract_subvector to convert i8 to v1i1. The bitcast can be nicely merged with any i8 loads that exist for argument passing in 32 mode for example. llvm-svn: 347380
2018-11-21 15:01:22 +08:00
; 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]
[X86] Update fpclass intrinsic tests to chain their calls to the intrinsic rather than joining them with add. The test cases try to test masked and unmasked isntructions at the same time. Previously the masked version relies on an extra fucntion parameter. Then the two results were combined with 'add'. This patch gets rid of the second parameter and just passes the result of the first intrinsic into the mask argument of the second call. Then there's no need for an 'add'. This configuration works a lot better with an upcoming patch to redefine the intrinsics to use vXi1 types for the output and mask argument. llvm-svn: 335551
2018-06-26 07:29:47 +08:00
; 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
[X86][AVX512DQ] add scalar fpclass Differential Revision: http://reviews.llvm.org/D13769 llvm-svn: 250650
2015-10-18 19:04:38 +08:00
}
[X86] Add tests for missed opportunities to fold a 128-bit vector load into vfpclassss and vpfpclasssd. llvm-svn: 318018
2017-11-13 14:46:46 +08:00
define i8 @test_int_x86_avx512_mask_fpclass_sd_load(<2 x double>* %x0ptr) {
[X86][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
[X86] Add tests for missed opportunities to fold a 128-bit vector load into vfpclassss and vpfpclasssd. llvm-svn: 318018
2017-11-13 14:46:46 +08:00
%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
}
[X86][AVX512DQ] add scalar fpclass Differential Revision: http://reviews.llvm.org/D13769 llvm-svn: 250650
2015-10-18 19:04:38 +08:00
declare i8 @llvm.x86.avx512.mask.fpclass.ss(<4 x float>, i32, i8)
[X86] Update fpclass intrinsic tests to chain their calls to the intrinsic rather than joining them with add. The test cases try to test masked and unmasked isntructions at the same time. Previously the masked version relies on an extra fucntion parameter. Then the two results were combined with 'add'. This patch gets rid of the second parameter and just passes the result of the first intrinsic into the mask argument of the second call. Then there's no need for an 'add'. This configuration works a lot better with an upcoming patch to redefine the intrinsics to use vXi1 types for the output and mask argument. llvm-svn: 335551
2018-06-26 07:29:47 +08:00
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:
[X86] In getScalarMaskingNode, replace scalar_to_vector with a bitcast to v8i1 and an extract_subvector to convert i8 to v1i1. The bitcast can be nicely merged with any i8 loads that exist for argument passing in 32 mode for example. llvm-svn: 347380
2018-11-21 15:01:22 +08:00
; 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]
[X86] Update fpclass intrinsic tests to chain their calls to the intrinsic rather than joining them with add. The test cases try to test masked and unmasked isntructions at the same time. Previously the masked version relies on an extra fucntion parameter. Then the two results were combined with 'add'. This patch gets rid of the second parameter and just passes the result of the first intrinsic into the mask argument of the second call. Then there's no need for an 'add'. This configuration works a lot better with an upcoming patch to redefine the intrinsics to use vXi1 types for the output and mask argument. llvm-svn: 335551
2018-06-26 07:29:47 +08:00
; 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
[X86][AVX512DQ] add scalar fpclass Differential Revision: http://reviews.llvm.org/D13769 llvm-svn: 250650
2015-10-18 19:04:38 +08:00
}
AVX512: Implemented encoding and intrinsics for VBROADCASTI32x2 and VBROADCASTF32x2 instructions. Differential Revision: http://reviews.llvm.org/D14216 llvm-svn: 251781
2015-11-02 15:39:36 +08:00
[X86] Add tests for missed opportunities to fold a 128-bit vector load into vfpclassss and vpfpclasssd. llvm-svn: 318018
2017-11-13 14:46:46 +08:00
define i8 @test_int_x86_avx512_mask_fpclass_ss_load(<4 x float>* %x0ptr, i8 %x1) {
[X86][AVX512] Cleanup intrinsics tests Ensure we test on 32-bit and 64-bit targets, and strip -mcpu usage. Part of ongoing work to ensure we test all intrinsic style tests on 32 and 64 bit targets where possible. llvm-svn: 333843
2018-06-03 22:56:04 +08:00
; 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]
[X86] Add tests for missed opportunities to fold a 128-bit vector load into vfpclassss and vpfpclasssd. llvm-svn: 318018
2017-11-13 14:46:46 +08:00
%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
}