r340798 - [X86] Add kortest intrinsics for 8, 32, and 64 bit masks. Add new intrinsic names for 16 bit masks.
Craig Topper via cfe-commits
cfe-commits at lists.llvm.org
Mon Aug 27 23:28:25 PDT 2018
Author: ctopper
Date: Mon Aug 27 23:28:25 2018
New Revision: 340798
URL: http://llvm.org/viewvc/llvm-project?rev=340798&view=rev
Log:
[X86] Add kortest intrinsics for 8, 32, and 64 bit masks. Add new intrinsic names for 16 bit masks.
This matches gcc and icc despite not being documented in the Intel Intrinsics Guide.
Modified:
cfe/trunk/include/clang/Basic/BuiltinsX86.def
cfe/trunk/lib/CodeGen/CGBuiltin.cpp
cfe/trunk/lib/Headers/avx512bwintrin.h
cfe/trunk/lib/Headers/avx512dqintrin.h
cfe/trunk/lib/Headers/avx512fintrin.h
cfe/trunk/test/CodeGen/avx512bw-builtins.c
cfe/trunk/test/CodeGen/avx512dq-builtins.c
cfe/trunk/test/CodeGen/avx512f-builtins.c
Modified: cfe/trunk/include/clang/Basic/BuiltinsX86.def
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/include/clang/Basic/BuiltinsX86.def?rev=340798&r1=340797&r2=340798&view=diff
==============================================================================
--- cfe/trunk/include/clang/Basic/BuiltinsX86.def (original)
+++ cfe/trunk/include/clang/Basic/BuiltinsX86.def Mon Aug 27 23:28:25 2018
@@ -1749,8 +1749,14 @@ TARGET_BUILTIN(__builtin_ia32_korqi, "Uc
TARGET_BUILTIN(__builtin_ia32_korhi, "UsUsUs", "nc", "avx512f")
TARGET_BUILTIN(__builtin_ia32_korsi, "UiUiUi", "nc", "avx512bw")
TARGET_BUILTIN(__builtin_ia32_kordi, "ULLiULLiULLi", "nc", "avx512bw")
+TARGET_BUILTIN(__builtin_ia32_kortestcqi, "iUcUc", "nc", "avx512dq")
+TARGET_BUILTIN(__builtin_ia32_kortestzqi, "iUcUc", "nc", "avx512dq")
TARGET_BUILTIN(__builtin_ia32_kortestchi, "iUsUs", "nc", "avx512f")
TARGET_BUILTIN(__builtin_ia32_kortestzhi, "iUsUs", "nc", "avx512f")
+TARGET_BUILTIN(__builtin_ia32_kortestcsi, "iUiUi", "nc", "avx512bw")
+TARGET_BUILTIN(__builtin_ia32_kortestzsi, "iUiUi", "nc", "avx512bw")
+TARGET_BUILTIN(__builtin_ia32_kortestcdi, "iULLiULLi", "nc", "avx512bw")
+TARGET_BUILTIN(__builtin_ia32_kortestzdi, "iULLiULLi", "nc", "avx512bw")
TARGET_BUILTIN(__builtin_ia32_kunpckhi, "UsUsUs", "nc", "avx512f")
TARGET_BUILTIN(__builtin_ia32_kxnorqi, "UcUcUc", "nc", "avx512dq")
TARGET_BUILTIN(__builtin_ia32_kxnorhi, "UsUsUs", "nc", "avx512f")
Modified: cfe/trunk/lib/CodeGen/CGBuiltin.cpp
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/CodeGen/CGBuiltin.cpp?rev=340798&r1=340797&r2=340798&view=diff
==============================================================================
--- cfe/trunk/lib/CodeGen/CGBuiltin.cpp (original)
+++ cfe/trunk/lib/CodeGen/CGBuiltin.cpp Mon Aug 27 23:28:25 2018
@@ -10012,14 +10012,21 @@ Value *CodeGenFunction::EmitX86BuiltinEx
return EmitX86MaskedCompare(*this, CC, false, Ops);
}
+ case X86::BI__builtin_ia32_kortestcqi:
case X86::BI__builtin_ia32_kortestchi:
- case X86::BI__builtin_ia32_kortestzhi: {
+ case X86::BI__builtin_ia32_kortestcsi:
+ case X86::BI__builtin_ia32_kortestcdi: {
Value *Or = EmitX86MaskLogic(*this, Instruction::Or, Ops);
- Value *C;
- if (BuiltinID == X86::BI__builtin_ia32_kortestchi)
- C = llvm::Constant::getAllOnesValue(Builder.getInt16Ty());
- else
- C = llvm::Constant::getNullValue(Builder.getInt16Ty());
+ Value *C = llvm::Constant::getAllOnesValue(Ops[0]->getType());
+ Value *Cmp = Builder.CreateICmpEQ(Or, C);
+ return Builder.CreateZExt(Cmp, ConvertType(E->getType()));
+ }
+ case X86::BI__builtin_ia32_kortestzqi:
+ case X86::BI__builtin_ia32_kortestzhi:
+ case X86::BI__builtin_ia32_kortestzsi:
+ case X86::BI__builtin_ia32_kortestzdi: {
+ Value *Or = EmitX86MaskLogic(*this, Instruction::Or, Ops);
+ Value *C = llvm::Constant::getNullValue(Ops[0]->getType());
Value *Cmp = Builder.CreateICmpEQ(Or, C);
return Builder.CreateZExt(Cmp, ConvertType(E->getType()));
}
Modified: cfe/trunk/lib/Headers/avx512bwintrin.h
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Headers/avx512bwintrin.h?rev=340798&r1=340797&r2=340798&view=diff
==============================================================================
--- cfe/trunk/lib/Headers/avx512bwintrin.h (original)
+++ cfe/trunk/lib/Headers/avx512bwintrin.h Mon Aug 27 23:28:25 2018
@@ -107,6 +107,42 @@ _kxor_mask64(__mmask64 __A, __mmask64 __
return (__mmask64)__builtin_ia32_kxordi((__mmask64)__A, (__mmask64)__B);
}
+static __inline__ unsigned char __DEFAULT_FN_ATTRS
+_kortestc_mask32_u8(__mmask32 __A, __mmask32 __B)
+{
+ return (unsigned char)__builtin_ia32_kortestcsi(__A, __B);
+}
+
+static __inline__ unsigned char __DEFAULT_FN_ATTRS
+_kortestz_mask32_u8(__mmask32 __A, __mmask32 __B)
+{
+ return (unsigned char)__builtin_ia32_kortestzsi(__A, __B);
+}
+
+static __inline__ unsigned char __DEFAULT_FN_ATTRS
+_kortest_mask32_u8(__mmask32 __A, __mmask32 __B, unsigned char *__C) {
+ *__C = (unsigned char)__builtin_ia32_kortestcsi(__A, __B);
+ return (unsigned char)__builtin_ia32_kortestzsi(__A, __B);
+}
+
+static __inline__ unsigned char __DEFAULT_FN_ATTRS
+_kortestc_mask64_u8(__mmask64 __A, __mmask64 __B)
+{
+ return (unsigned char)__builtin_ia32_kortestcdi(__A, __B);
+}
+
+static __inline__ unsigned char __DEFAULT_FN_ATTRS
+_kortestz_mask64_u8(__mmask64 __A, __mmask64 __B)
+{
+ return (unsigned char)__builtin_ia32_kortestzdi(__A, __B);
+}
+
+static __inline__ unsigned char __DEFAULT_FN_ATTRS
+_kortest_mask64_u8(__mmask64 __A, __mmask64 __B, unsigned char *__C) {
+ *__C = (unsigned char)__builtin_ia32_kortestcdi(__A, __B);
+ return (unsigned char)__builtin_ia32_kortestzdi(__A, __B);
+}
+
/* Integer compare */
#define _mm512_cmp_epi8_mask(a, b, p) \
Modified: cfe/trunk/lib/Headers/avx512dqintrin.h
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Headers/avx512dqintrin.h?rev=340798&r1=340797&r2=340798&view=diff
==============================================================================
--- cfe/trunk/lib/Headers/avx512dqintrin.h (original)
+++ cfe/trunk/lib/Headers/avx512dqintrin.h Mon Aug 27 23:28:25 2018
@@ -68,6 +68,24 @@ _kxor_mask8(__mmask8 __A, __mmask8 __B)
return (__mmask8)__builtin_ia32_kxorqi((__mmask8)__A, (__mmask8)__B);
}
+static __inline__ unsigned char __DEFAULT_FN_ATTRS
+_kortestc_mask8_u8(__mmask8 __A, __mmask8 __B)
+{
+ return (unsigned char)__builtin_ia32_kortestcqi(__A, __B);
+}
+
+static __inline__ unsigned char __DEFAULT_FN_ATTRS
+_kortestz_mask8_u8(__mmask8 __A, __mmask8 __B)
+{
+ return (unsigned char)__builtin_ia32_kortestzqi(__A, __B);
+}
+
+static __inline__ unsigned char __DEFAULT_FN_ATTRS
+_kortest_mask8_u8(__mmask8 __A, __mmask8 __B, unsigned char *__C) {
+ *__C = (unsigned char)__builtin_ia32_kortestcqi(__A, __B);
+ return (unsigned char)__builtin_ia32_kortestzqi(__A, __B);
+}
+
static __inline__ __m512i __DEFAULT_FN_ATTRS512
_mm512_mullo_epi64 (__m512i __A, __m512i __B) {
return (__m512i) ((__v8du) __A * (__v8du) __B);
Modified: cfe/trunk/lib/Headers/avx512fintrin.h
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/lib/Headers/avx512fintrin.h?rev=340798&r1=340797&r2=340798&view=diff
==============================================================================
--- cfe/trunk/lib/Headers/avx512fintrin.h (original)
+++ cfe/trunk/lib/Headers/avx512fintrin.h Mon Aug 27 23:28:25 2018
@@ -8351,6 +8351,24 @@ _mm512_kortestz (__mmask16 __A, __mmask1
return __builtin_ia32_kortestzhi ((__mmask16) __A, (__mmask16) __B);
}
+static __inline__ unsigned char __DEFAULT_FN_ATTRS
+_kortestc_mask16_u8(__mmask16 __A, __mmask16 __B)
+{
+ return (unsigned char)__builtin_ia32_kortestchi(__A, __B);
+}
+
+static __inline__ unsigned char __DEFAULT_FN_ATTRS
+_kortestz_mask16_u8(__mmask16 __A, __mmask16 __B)
+{
+ return (unsigned char)__builtin_ia32_kortestzhi(__A, __B);
+}
+
+static __inline__ unsigned char __DEFAULT_FN_ATTRS
+_kortest_mask16_u8(__mmask16 __A, __mmask16 __B, unsigned char *__C) {
+ *__C = (unsigned char)__builtin_ia32_kortestchi(__A, __B);
+ return (unsigned char)__builtin_ia32_kortestzhi(__A, __B);
+}
+
static __inline__ __mmask16 __DEFAULT_FN_ATTRS
_mm512_kunpackb (__mmask16 __A, __mmask16 __B)
{
Modified: cfe/trunk/test/CodeGen/avx512bw-builtins.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/avx512bw-builtins.c?rev=340798&r1=340797&r2=340798&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/avx512bw-builtins.c (original)
+++ cfe/trunk/test/CodeGen/avx512bw-builtins.c Mon Aug 27 23:28:25 2018
@@ -134,6 +134,98 @@ __mmask64 test_kxor_mask64(__m512i __A,
__E, __F);
}
+unsigned char test_kortestz_mask32_u8(__m512i __A, __m512i __B, __m512i __C, __m512i __D) {
+ // CHECK-LABEL: @test_kortestz_mask32_u8
+ // CHECK: [[LHS:%.*]] = bitcast i32 %{{.*}} to <32 x i1>
+ // CHECK: [[RHS:%.*]] = bitcast i32 %{{.*}} to <32 x i1>
+ // CHECK: [[OR:%.*]] = or <32 x i1> [[LHS]], [[RHS]]
+ // CHECK: [[CAST:%.*]] = bitcast <32 x i1> [[OR]] to i32
+ // CHECK: [[CMP:%.*]] = icmp eq i32 [[CAST]], 0
+ // CHECK: [[ZEXT:%.*]] = zext i1 [[CMP]] to i32
+ // CHECK: trunc i32 [[ZEXT]] to i8
+ return _kortestz_mask32_u8(_mm512_cmpneq_epu16_mask(__A, __B),
+ _mm512_cmpneq_epu16_mask(__C, __D));
+}
+
+unsigned char test_kortestc_mask32_u8(__m512i __A, __m512i __B, __m512i __C, __m512i __D) {
+ // CHECK-LABEL: @test_kortestc_mask32_u8
+ // CHECK: [[LHS:%.*]] = bitcast i32 %{{.*}} to <32 x i1>
+ // CHECK: [[RHS:%.*]] = bitcast i32 %{{.*}} to <32 x i1>
+ // CHECK: [[OR:%.*]] = or <32 x i1> [[LHS]], [[RHS]]
+ // CHECK: [[CAST:%.*]] = bitcast <32 x i1> [[OR]] to i32
+ // CHECK: [[CMP:%.*]] = icmp eq i32 [[CAST]], -1
+ // CHECK: [[ZEXT:%.*]] = zext i1 [[CMP]] to i32
+ // CHECK: trunc i32 [[ZEXT]] to i8
+ return _kortestc_mask32_u8(_mm512_cmpneq_epu16_mask(__A, __B),
+ _mm512_cmpneq_epu16_mask(__C, __D));
+}
+
+unsigned char test_kortest_mask32_u8(__m512i __A, __m512i __B, __m512i __C, __m512i __D, unsigned char *CF) {
+ // CHECK-LABEL: @test_kortest_mask32_u8
+ // CHECK: [[LHS:%.*]] = bitcast i32 %{{.*}} to <32 x i1>
+ // CHECK: [[RHS:%.*]] = bitcast i32 %{{.*}} to <32 x i1>
+ // CHECK: [[OR:%.*]] = or <32 x i1> [[LHS]], [[RHS]]
+ // CHECK: [[CAST:%.*]] = bitcast <32 x i1> [[OR]] to i32
+ // CHECK: [[CMP:%.*]] = icmp eq i32 [[CAST]], -1
+ // CHECK: [[ZEXT:%.*]] = zext i1 [[CMP]] to i32
+ // CHECK: trunc i32 [[ZEXT]] to i8
+ // CHECK: [[LHS2:%.*]] = bitcast i32 %{{.*}} to <32 x i1>
+ // CHECK: [[RHS2:%.*]] = bitcast i32 %{{.*}} to <32 x i1>
+ // CHECK: [[OR2:%.*]] = or <32 x i1> [[LHS2]], [[RHS2]]
+ // CHECK: [[CAST2:%.*]] = bitcast <32 x i1> [[OR2]] to i32
+ // CHECK: [[CMP2:%.*]] = icmp eq i32 [[CAST2]], 0
+ // CHECK: [[ZEXT2:%.*]] = zext i1 [[CMP2]] to i32
+ // CHECK: trunc i32 [[ZEXT2]] to i8
+ return _kortest_mask32_u8(_mm512_cmpneq_epu16_mask(__A, __B),
+ _mm512_cmpneq_epu16_mask(__C, __D), CF);
+}
+
+unsigned char test_kortestz_mask64_u8(__m512i __A, __m512i __B, __m512i __C, __m512i __D) {
+ // CHECK-LABEL: @test_kortestz_mask64_u8
+ // CHECK: [[LHS:%.*]] = bitcast i64 %{{.*}} to <64 x i1>
+ // CHECK: [[RHS:%.*]] = bitcast i64 %{{.*}} to <64 x i1>
+ // CHECK: [[OR:%.*]] = or <64 x i1> [[LHS]], [[RHS]]
+ // CHECK: [[CAST:%.*]] = bitcast <64 x i1> [[OR]] to i64
+ // CHECK: [[CMP:%.*]] = icmp eq i64 [[CAST]], 0
+ // CHECK: [[ZEXT:%.*]] = zext i1 [[CMP]] to i32
+ // CHECK: trunc i32 [[ZEXT]] to i8
+ return _kortestz_mask64_u8(_mm512_cmpneq_epu8_mask(__A, __B),
+ _mm512_cmpneq_epu8_mask(__C, __D));
+}
+
+unsigned char test_kortestc_mask64_u8(__m512i __A, __m512i __B, __m512i __C, __m512i __D) {
+ // CHECK-LABEL: @test_kortestc_mask64_u8
+ // CHECK: [[LHS:%.*]] = bitcast i64 %{{.*}} to <64 x i1>
+ // CHECK: [[RHS:%.*]] = bitcast i64 %{{.*}} to <64 x i1>
+ // CHECK: [[OR:%.*]] = or <64 x i1> [[LHS]], [[RHS]]
+ // CHECK: [[CAST:%.*]] = bitcast <64 x i1> [[OR]] to i64
+ // CHECK: [[CMP:%.*]] = icmp eq i64 [[CAST]], -1
+ // CHECK: [[ZEXT:%.*]] = zext i1 [[CMP]] to i32
+ // CHECK: trunc i32 [[ZEXT]] to i8
+ return _kortestc_mask64_u8(_mm512_cmpneq_epu8_mask(__A, __B),
+ _mm512_cmpneq_epu8_mask(__C, __D));
+}
+
+unsigned char test_kortest_mask64_u8(__m512i __A, __m512i __B, __m512i __C, __m512i __D, unsigned char *CF) {
+ // CHECK-LABEL: @test_kortest_mask64_u8
+ // CHECK: [[LHS:%.*]] = bitcast i64 %{{.*}} to <64 x i1>
+ // CHECK: [[RHS:%.*]] = bitcast i64 %{{.*}} to <64 x i1>
+ // CHECK: [[OR:%.*]] = or <64 x i1> [[LHS]], [[RHS]]
+ // CHECK: [[CAST:%.*]] = bitcast <64 x i1> [[OR]] to i64
+ // CHECK: [[CMP:%.*]] = icmp eq i64 [[CAST]], -1
+ // CHECK: [[ZEXT:%.*]] = zext i1 [[CMP]] to i32
+ // CHECK: trunc i32 [[ZEXT]] to i8
+ // CHECK: [[LHS2:%.*]] = bitcast i64 %{{.*}} to <64 x i1>
+ // CHECK: [[RHS2:%.*]] = bitcast i64 %{{.*}} to <64 x i1>
+ // CHECK: [[OR2:%.*]] = or <64 x i1> [[LHS2]], [[RHS2]]
+ // CHECK: [[CAST2:%.*]] = bitcast <64 x i1> [[OR2]] to i64
+ // CHECK: [[CMP2:%.*]] = icmp eq i64 [[CAST2]], 0
+ // CHECK: [[ZEXT2:%.*]] = zext i1 [[CMP2]] to i32
+ // CHECK: trunc i32 [[ZEXT2]] to i8
+ return _kortest_mask64_u8(_mm512_cmpneq_epu8_mask(__A, __B),
+ _mm512_cmpneq_epu8_mask(__C, __D), CF);
+}
+
__mmask64 test_mm512_cmpeq_epi8_mask(__m512i __a, __m512i __b) {
// CHECK-LABEL: @test_mm512_cmpeq_epi8_mask
// CHECK: icmp eq <64 x i8> %{{.*}}, %{{.*}}
Modified: cfe/trunk/test/CodeGen/avx512dq-builtins.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/avx512dq-builtins.c?rev=340798&r1=340797&r2=340798&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/avx512dq-builtins.c (original)
+++ cfe/trunk/test/CodeGen/avx512dq-builtins.c Mon Aug 27 23:28:25 2018
@@ -68,6 +68,52 @@ __mmask8 test_kxor_mask8(__m512i __A, __
__E, __F);
}
+unsigned char test_kortestz_mask8_u8(__m512i __A, __m512i __B, __m512i __C, __m512i __D) {
+ // CHECK-LABEL: @test_kortestz_mask8_u8
+ // CHECK: [[LHS:%.*]] = bitcast i8 %{{.*}} to <8 x i1>
+ // CHECK: [[RHS:%.*]] = bitcast i8 %{{.*}} to <8 x i1>
+ // CHECK: [[OR:%.*]] = or <8 x i1> [[LHS]], [[RHS]]
+ // CHECK: [[CAST:%.*]] = bitcast <8 x i1> [[OR]] to i8
+ // CHECK: [[CMP:%.*]] = icmp eq i8 [[CAST]], 0
+ // CHECK: [[ZEXT:%.*]] = zext i1 [[CMP]] to i32
+ // CHECK: trunc i32 [[ZEXT]] to i8
+ return _kortestz_mask8_u8(_mm512_cmpneq_epu64_mask(__A, __B),
+ _mm512_cmpneq_epu64_mask(__C, __D));
+}
+
+unsigned char test_kortestc_mask8_u8(__m512i __A, __m512i __B, __m512i __C, __m512i __D) {
+ // CHECK-LABEL: @test_kortestc_mask8_u8
+ // CHECK: [[LHS:%.*]] = bitcast i8 %{{.*}} to <8 x i1>
+ // CHECK: [[RHS:%.*]] = bitcast i8 %{{.*}} to <8 x i1>
+ // CHECK: [[OR:%.*]] = or <8 x i1> [[LHS]], [[RHS]]
+ // CHECK: [[CAST:%.*]] = bitcast <8 x i1> [[OR]] to i8
+ // CHECK: [[CMP:%.*]] = icmp eq i8 [[CAST]], -1
+ // CHECK: [[ZEXT:%.*]] = zext i1 [[CMP]] to i32
+ // CHECK: trunc i32 [[ZEXT]] to i8
+ return _kortestc_mask8_u8(_mm512_cmpneq_epu64_mask(__A, __B),
+ _mm512_cmpneq_epu64_mask(__C, __D));
+}
+
+unsigned char test_kortest_mask8_u8(__m512i __A, __m512i __B, __m512i __C, __m512i __D, unsigned char *CF) {
+ // CHECK-LABEL: @test_kortest_mask8_u8
+ // CHECK: [[LHS:%.*]] = bitcast i8 %{{.*}} to <8 x i1>
+ // CHECK: [[RHS:%.*]] = bitcast i8 %{{.*}} to <8 x i1>
+ // CHECK: [[OR:%.*]] = or <8 x i1> [[LHS]], [[RHS]]
+ // CHECK: [[CAST:%.*]] = bitcast <8 x i1> [[OR]] to i8
+ // CHECK: [[CMP:%.*]] = icmp eq i8 [[CAST]], -1
+ // CHECK: [[ZEXT:%.*]] = zext i1 [[CMP]] to i32
+ // CHECK: trunc i32 [[ZEXT]] to i8
+ // CHECK: [[LHS2:%.*]] = bitcast i8 %{{.*}} to <8 x i1>
+ // CHECK: [[RHS2:%.*]] = bitcast i8 %{{.*}} to <8 x i1>
+ // CHECK: [[OR2:%.*]] = or <8 x i1> [[LHS2]], [[RHS2]]
+ // CHECK: [[CAST2:%.*]] = bitcast <8 x i1> [[OR2]] to i8
+ // CHECK: [[CMP2:%.*]] = icmp eq i8 [[CAST2]], 0
+ // CHECK: [[ZEXT2:%.*]] = zext i1 [[CMP2]] to i32
+ // CHECK: trunc i32 [[ZEXT2]] to i8
+ return _kortest_mask8_u8(_mm512_cmpneq_epu64_mask(__A, __B),
+ _mm512_cmpneq_epu64_mask(__C, __D), CF);
+}
+
__m512i test_mm512_mullo_epi64 (__m512i __A, __m512i __B) {
// CHECK-LABEL: @test_mm512_mullo_epi64
// CHECK: mul <8 x i64>
Modified: cfe/trunk/test/CodeGen/avx512f-builtins.c
URL: http://llvm.org/viewvc/llvm-project/cfe/trunk/test/CodeGen/avx512f-builtins.c?rev=340798&r1=340797&r2=340798&view=diff
==============================================================================
--- cfe/trunk/test/CodeGen/avx512f-builtins.c (original)
+++ cfe/trunk/test/CodeGen/avx512f-builtins.c Mon Aug 27 23:28:25 2018
@@ -8149,6 +8149,52 @@ int test_mm512_kortestz(__m512i __A, __m
_mm512_cmpneq_epu32_mask(__C, __D));
}
+unsigned char test_kortestz_mask16_u8(__m512i __A, __m512i __B, __m512i __C, __m512i __D) {
+ // CHECK-LABEL: @test_kortestz_mask16_u8
+ // CHECK: [[LHS:%.*]] = bitcast i16 %{{.*}} to <16 x i1>
+ // CHECK: [[RHS:%.*]] = bitcast i16 %{{.*}} to <16 x i1>
+ // CHECK: [[OR:%.*]] = or <16 x i1> [[LHS]], [[RHS]]
+ // CHECK: [[CAST:%.*]] = bitcast <16 x i1> [[OR]] to i16
+ // CHECK: [[CMP:%.*]] = icmp eq i16 [[CAST]], 0
+ // CHECK: [[ZEXT:%.*]] = zext i1 [[CMP]] to i32
+ // CHECK: trunc i32 [[ZEXT]] to i8
+ return _kortestz_mask16_u8(_mm512_cmpneq_epu32_mask(__A, __B),
+ _mm512_cmpneq_epu32_mask(__C, __D));
+}
+
+unsigned char test_kortestc_mask16_u8(__m512i __A, __m512i __B, __m512i __C, __m512i __D) {
+ // CHECK-LABEL: @test_kortestc_mask16_u8
+ // CHECK: [[LHS:%.*]] = bitcast i16 %{{.*}} to <16 x i1>
+ // CHECK: [[RHS:%.*]] = bitcast i16 %{{.*}} to <16 x i1>
+ // CHECK: [[OR:%.*]] = or <16 x i1> [[LHS]], [[RHS]]
+ // CHECK: [[CAST:%.*]] = bitcast <16 x i1> [[OR]] to i16
+ // CHECK: [[CMP:%.*]] = icmp eq i16 [[CAST]], -1
+ // CHECK: [[ZEXT:%.*]] = zext i1 [[CMP]] to i32
+ // CHECK: trunc i32 [[ZEXT]] to i8
+ return _kortestc_mask16_u8(_mm512_cmpneq_epu32_mask(__A, __B),
+ _mm512_cmpneq_epu32_mask(__C, __D));
+}
+
+unsigned char test_kortest_mask16_u8(__m512i __A, __m512i __B, __m512i __C, __m512i __D, unsigned char *CF) {
+ // CHECK-LABEL: @test_kortest_mask16_u8
+ // CHECK: [[LHS:%.*]] = bitcast i16 %{{.*}} to <16 x i1>
+ // CHECK: [[RHS:%.*]] = bitcast i16 %{{.*}} to <16 x i1>
+ // CHECK: [[OR:%.*]] = or <16 x i1> [[LHS]], [[RHS]]
+ // CHECK: [[CAST:%.*]] = bitcast <16 x i1> [[OR]] to i16
+ // CHECK: [[CMP:%.*]] = icmp eq i16 [[CAST]], -1
+ // CHECK: [[ZEXT:%.*]] = zext i1 [[CMP]] to i32
+ // CHECK: trunc i32 [[ZEXT]] to i8
+ // CHECK: [[LHS2:%.*]] = bitcast i16 %{{.*}} to <16 x i1>
+ // CHECK: [[RHS2:%.*]] = bitcast i16 %{{.*}} to <16 x i1>
+ // CHECK: [[OR2:%.*]] = or <16 x i1> [[LHS2]], [[RHS2]]
+ // CHECK: [[CAST2:%.*]] = bitcast <16 x i1> [[OR2]] to i16
+ // CHECK: [[CMP2:%.*]] = icmp eq i16 [[CAST2]], 0
+ // CHECK: [[ZEXT2:%.*]] = zext i1 [[CMP2]] to i32
+ // CHECK: trunc i32 [[ZEXT2]] to i8
+ return _kortest_mask16_u8(_mm512_cmpneq_epu32_mask(__A, __B),
+ _mm512_cmpneq_epu32_mask(__C, __D), CF);
+}
+
__mmask16 test_mm512_kunpackb(__m512i __A, __m512i __B, __m512i __C, __m512i __D, __m512i __E, __m512i __F) {
// CHECK-LABEL: @test_mm512_kunpackb
// CHECK: [[LHS:%.*]] = bitcast i16 %{{.*}} to <16 x i1>
More information about the cfe-commits
mailing list