derek.holloway
7502bf64a4
CodeQL / Analyze (actions) (push) Canceled after 0s
CodeQL / Analyze (python) (push) Canceled after 0s
pre-commit / pre-commit (push) Canceled after 0s
Multi-Arch CI / setup (push) Canceled after 0s
Publish build_manylinux_x86_64 images / publish_build_manylinux_x86_64 (push) Canceled after 0s
Setup (Multi-Arch) / setup (push) Canceled after 0s
Multi-Arch CI / Linux::${{ fromJSON(needs.setup.outputs.linux_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubi10 image / publish_no_rocm_image_ubi10 (push) Canceled after 0s
Multi-Arch CI / Windows::${{ fromJSON(needs.setup.outputs.windows_build_config || '{}').build_variant_label || 'skip' }} (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Multi-Arch CI / Manifest Diff (push) Canceled after 0s
TheRock Manifest Diff Report / Generate Manifest Diff Report (push) Canceled after 0s
Multi-Arch CI / CI Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_ocl_rt images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04_rocgdb images / publish_no_rocm_image_ubuntu24_04_rocgdb (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_ocl_rt (push) Canceled after 0s
Publish TheRock Docker image / build-and-push-image (push) Canceled after 0s
Unit Tests / Unit Tests :: ubuntu-24.04 (push) Canceled after 0s
Unit Tests / Unit Tests :: windows-2022 (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_media (push) Canceled after 0s
Unit Tests / Unit Tests Summary (push) Canceled after 0s
Publish no_rocm_image_ubuntu24_04 images / publish_no_rocm_image_ubuntu24_04_openmpi (push) Canceled after 0s
219 lines
8.6 KiB
C
219 lines
8.6 KiB
C
// RUN: %clang_cc1 -x c -ffreestanding %s -triple=x86_64-unknown-unknown -target-feature +tbm -emit-llvm -o - | FileCheck %s --check-prefixes=CHECK,X64
|
|
// RUN: %clang_cc1 -x c -ffreestanding %s -triple=i386-unknown-unknown -target-feature +tbm -emit-llvm -o - | FileCheck %s --check-prefixes=CHECK
|
|
// RUN: %clang_cc1 -x c++ -ffreestanding %s -triple=x86_64-unknown-unknown -target-feature +tbm -emit-llvm -o - | FileCheck %s --check-prefixes=CHECK,X64
|
|
// RUN: %clang_cc1 -x c++ -ffreestanding %s -triple=i386-unknown-unknown -target-feature +tbm -emit-llvm -o - | FileCheck %s --check-prefixes=CHECK
|
|
|
|
// RUN: %clang_cc1 -x c -ffreestanding %s -triple=x86_64-unknown-unknown -target-feature +tbm -emit-llvm -o - -fexperimental-new-constant-interpreter | FileCheck %s --check-prefixes=CHECK,X64
|
|
// RUN: %clang_cc1 -x c -ffreestanding %s -triple=i386-unknown-unknown -target-feature +tbm -emit-llvm -o - -fexperimental-new-constant-interpreter | FileCheck %s --check-prefixes=CHECK
|
|
// RUN: %clang_cc1 -x c++ -ffreestanding %s -triple=x86_64-unknown-unknown -target-feature +tbm -emit-llvm -o - -fexperimental-new-constant-interpreter | FileCheck %s --check-prefixes=CHECK,X64
|
|
// RUN: %clang_cc1 -x c++ -ffreestanding %s -triple=i386-unknown-unknown -target-feature +tbm -emit-llvm -o - -fexperimental-new-constant-interpreter | FileCheck %s --check-prefixes=CHECK
|
|
|
|
#include <x86intrin.h>
|
|
|
|
// NOTE: This should match the tests in llvm/test/CodeGen/X86/tbm-intrinsics-fast-isel.ll
|
|
|
|
unsigned int test__bextri_u32(unsigned int a) {
|
|
// CHECK-LABEL: test__bextri_u32
|
|
// CHECK: call i32 @llvm.x86.tbm.bextri.u32(i32 %{{.*}}, i32 1)
|
|
return __bextri_u32(a, 1);
|
|
}
|
|
|
|
#ifdef __x86_64__
|
|
unsigned long long test__bextri_u64(unsigned long long a) {
|
|
// X64-LABEL: test__bextri_u64
|
|
// X64: call i64 @llvm.x86.tbm.bextri.u64(i64 %{{.*}}, i64 2)
|
|
return __bextri_u64(a, 2);
|
|
}
|
|
|
|
unsigned long long test__bextri_u64_bigint(unsigned long long a) {
|
|
// X64-LABEL: test__bextri_u64_bigint
|
|
// X64: call i64 @llvm.x86.tbm.bextri.u64(i64 %{{.*}}, i64 549755813887)
|
|
return __bextri_u64(a, 0x7fffffffffLL);
|
|
}
|
|
#endif
|
|
|
|
unsigned int test__blcfill_u32(unsigned int a) {
|
|
// CHECK-LABEL: test__blcfill_u32
|
|
// CHECK: [[TMP:%.*]] = add i32 %{{.*}}, 1
|
|
// CHECK: %{{.*}} = and i32 %{{.*}}, [[TMP]]
|
|
return __blcfill_u32(a);
|
|
}
|
|
|
|
#ifdef __x86_64__
|
|
unsigned long long test__blcfill_u64(unsigned long long a) {
|
|
// X64-LABEL: test__blcfill_u64
|
|
// X64: [[TMP:%.*]] = add i64 %{{.*}}, 1
|
|
// X64: %{{.*}} = and i64 %{{.*}}, [[TMP]]
|
|
return __blcfill_u64(a);
|
|
}
|
|
#endif
|
|
|
|
unsigned int test__blci_u32(unsigned int a) {
|
|
// CHECK-LABEL: test__blci_u32
|
|
// CHECK: [[TMP1:%.*]] = add i32 %{{.*}}, 1
|
|
// CHECK: [[TMP2:%.*]] = xor i32 [[TMP1]], -1
|
|
// CHECK: %{{.*}} = or i32 %{{.*}}, [[TMP2]]
|
|
return __blci_u32(a);
|
|
}
|
|
|
|
#ifdef __x86_64__
|
|
unsigned long long test__blci_u64(unsigned long long a) {
|
|
// X64-LABEL: test__blci_u64
|
|
// X64: [[TMP1:%.*]] = add i64 %{{.*}}, 1
|
|
// X64: [[TMP2:%.*]] = xor i64 [[TMP1]], -1
|
|
// X64: %{{.*}} = or i64 %{{.*}}, [[TMP2]]
|
|
return __blci_u64(a);
|
|
}
|
|
#endif
|
|
|
|
unsigned int test__blcic_u32(unsigned int a) {
|
|
// CHECK-LABEL: test__blcic_u32
|
|
// CHECK: [[TMP1:%.*]] = xor i32 %{{.*}}, -1
|
|
// CHECK: [[TMP2:%.*]] = add i32 %{{.*}}, 1
|
|
// CHECK-NEXT: {{.*}} = and i32 [[TMP1]], [[TMP2]]
|
|
return __blcic_u32(a);
|
|
}
|
|
|
|
#ifdef __x86_64__
|
|
unsigned long long test__blcic_u64(unsigned long long a) {
|
|
// X64-LABEL: test__blcic_u64
|
|
// X64: [[TMP1:%.*]] = xor i64 %{{.*}}, -1
|
|
// X64: [[TMP2:%.*]] = add i64 %{{.*}}, 1
|
|
// X64-NEXT: {{.*}} = and i64 [[TMP1]], [[TMP2]]
|
|
return __blcic_u64(a);
|
|
}
|
|
#endif
|
|
|
|
unsigned int test__blcmsk_u32(unsigned int a) {
|
|
// CHECK-LABEL: test__blcmsk_u32
|
|
// CHECK: [[TMP:%.*]] = add i32 %{{.*}}, 1
|
|
// CHECK-NEXT: {{.*}} = xor i32 %{{.*}}, [[TMP]]
|
|
return __blcmsk_u32(a);
|
|
}
|
|
|
|
#ifdef __x86_64__
|
|
unsigned long long test__blcmsk_u64(unsigned long long a) {
|
|
// X64-LABEL: test__blcmsk_u64
|
|
// X64: [[TMP:%.*]] = add i64 %{{.*}}, 1
|
|
// X64-NEXT: {{.*}} = xor i64 %{{.*}}, [[TMP]]
|
|
return __blcmsk_u64(a);
|
|
}
|
|
#endif
|
|
|
|
unsigned int test__blcs_u32(unsigned int a) {
|
|
// CHECK-LABEL: test__blcs_u32
|
|
// CHECK: [[TMP:%.*]] = add i32 %{{.*}}, 1
|
|
// CHECK-NEXT: {{.*}} = or i32 %{{.*}}, [[TMP]]
|
|
return __blcs_u32(a);
|
|
}
|
|
|
|
#ifdef __x86_64__
|
|
unsigned long long test__blcs_u64(unsigned long long a) {
|
|
// X64-LABEL: test__blcs_u64
|
|
// X64: [[TMP:%.*]] = add i64 %{{.*}}, 1
|
|
// X64-NEXT: {{.*}} = or i64 %{{.*}}, [[TMP]]
|
|
return __blcs_u64(a);
|
|
}
|
|
#endif
|
|
|
|
unsigned int test__blsfill_u32(unsigned int a) {
|
|
// CHECK-LABEL: test__blsfill_u32
|
|
// CHECK: [[TMP:%.*]] = sub i32 %{{.*}}, 1
|
|
// CHECK-NEXT: {{.*}} = or i32 %{{.*}}, [[TMP]]
|
|
return __blsfill_u32(a);
|
|
}
|
|
|
|
#ifdef __x86_64__
|
|
unsigned long long test__blsfill_u64(unsigned long long a) {
|
|
// X64-LABEL: test__blsfill_u64
|
|
// X64: [[TMP:%.*]] = sub i64 %{{.*}}, 1
|
|
// X64-NEXT: {{.*}} = or i64 %{{.*}}, [[TMP]]
|
|
return __blsfill_u64(a);
|
|
}
|
|
#endif
|
|
|
|
unsigned int test__blsic_u32(unsigned int a) {
|
|
// CHECK-LABEL: test__blsic_u32
|
|
// CHECK: [[TMP1:%.*]] = xor i32 %{{.*}}, -1
|
|
// CHECK: [[TMP2:%.*]] = sub i32 %{{.*}}, 1
|
|
// CHECK-NEXT: {{.*}} = or i32 [[TMP1]], [[TMP2]]
|
|
return __blsic_u32(a);
|
|
}
|
|
|
|
#ifdef __x86_64__
|
|
unsigned long long test__blsic_u64(unsigned long long a) {
|
|
// X64-LABEL: test__blsic_u64
|
|
// X64: [[TMP1:%.*]] = xor i64 %{{.*}}, -1
|
|
// X64: [[TMP2:%.*]] = sub i64 %{{.*}}, 1
|
|
// X64-NEXT: {{.*}} = or i64 [[TMP1]], [[TMP2]]
|
|
return __blsic_u64(a);
|
|
}
|
|
#endif
|
|
|
|
unsigned int test__t1mskc_u32(unsigned int a) {
|
|
// CHECK-LABEL: test__t1mskc_u32
|
|
// CHECK: [[TMP1:%.*]] = xor i32 %{{.*}}, -1
|
|
// CHECK: [[TMP2:%.*]] = add i32 %{{.*}}, 1
|
|
// CHECK-NEXT: {{.*}} = or i32 [[TMP1]], [[TMP2]]
|
|
return __t1mskc_u32(a);
|
|
}
|
|
|
|
#ifdef __x86_64__
|
|
unsigned long long test__t1mskc_u64(unsigned long long a) {
|
|
// X64-LABEL: test__t1mskc_u64
|
|
// X64: [[TMP1:%.*]] = xor i64 %{{.*}}, -1
|
|
// X64: [[TMP2:%.*]] = add i64 %{{.*}}, 1
|
|
// X64-NEXT: {{.*}} = or i64 [[TMP1]], [[TMP2]]
|
|
return __t1mskc_u64(a);
|
|
}
|
|
#endif
|
|
|
|
unsigned int test__tzmsk_u32(unsigned int a) {
|
|
// CHECK-LABEL: test__tzmsk_u32
|
|
// CHECK: [[TMP1:%.*]] = xor i32 %{{.*}}, -1
|
|
// CHECK: [[TMP2:%.*]] = sub i32 %{{.*}}, 1
|
|
// CHECK-NEXT: {{.*}} = and i32 [[TMP1]], [[TMP2]]
|
|
return __tzmsk_u32(a);
|
|
}
|
|
|
|
#ifdef __x86_64__
|
|
unsigned long long test__tzmsk_u64(unsigned long long a) {
|
|
// X64-LABEL: test__tzmsk_u64
|
|
// X64: [[TMP1:%.*]] = xor i64 %{{.*}}, -1
|
|
// X64: [[TMP2:%.*]] = sub i64 %{{.*}}, 1
|
|
// X64-NEXT: {{.*}} = and i64 [[TMP1]], [[TMP2]]
|
|
return __tzmsk_u64(a);
|
|
}
|
|
#endif
|
|
|
|
// Test constexpr handling.
|
|
#if defined(__cplusplus) && (__cplusplus >= 201103L)
|
|
char bextri32_0[__bextri_u32(0x00000000, 0x00000000) == 0x00000000 ? 1 : -1];
|
|
char bextri32_1[__bextri_u32(0x000003F0, 0xFFFF1004) == 0x0000003F ? 1 : -1];
|
|
char bextri32_2[__bextri_u32(0x000003F0, 0xFFFF3008) == 0x00000003 ? 1 : -1];
|
|
|
|
char blcfill32[__blcfill_u32(0x89ABCDEF) == (0x89ABCDEF & (0x89ABCDEF + 1)) ? 1 : -1];
|
|
char blci32[__blci_u32(0x89ABCDEF) == (0x89ABCDEF | ~(0x89ABCDEF + 1)) ? 1 : -1];
|
|
char blcmsk32[__blcmsk_u32(0x89ABCDEF) == (0x89ABCDEF ^ (0x89ABCDEF + 1)) ? 1 : -1];
|
|
char blcs32[__blcs_u32(0x89ABCDEF) == (0x89ABCDEF | (0x89ABCDEF + 1)) ? 1 : -1];
|
|
char blsfill32[__blsfill_u32(0x89ABCDEF) == (0x89ABCDEF | (0x89ABCDEF - 1)) ? 1 : -1];
|
|
char blsic32[__blsic_u32(0x89ABCDEF) == (~0x89ABCDEF | (0x89ABCDEF - 1)) ? 1 : -1];
|
|
char t1mskc32[__t1mskc_u32(0x89ABCDEF) == (~0x89ABCDEF | (0x89ABCDEF + 1)) ? 1 : -1];
|
|
char tzmsk32[__tzmsk_u32(0x89ABCDEF) == (~0x89ABCDEF & (0x89ABCDEF - 1)) ? 1 : -1];
|
|
|
|
#ifdef __x86_64__
|
|
char bextri64_0[__bextri_u64(0x0000000000000000ULL, 0x0000000000000000ULL) == 0x0000000000000000ULL ? 1 : -1];
|
|
char bextri64_1[__bextri_u64(0xF000000000000001ULL, 0x0000000000004001ULL) == 0x7800000000000000ULL ? 1 : -1];
|
|
char bextri64_2[__bextri_u64(0xF000000000000001ULL, 0xFFFFFFFFFFFF1001ULL) == 0x0000000000000000ULL ? 1 : -1];
|
|
|
|
char blcfill64[__blcfill_u64(0xFEDCBA9876543210) == (0xFEDCBA9876543210 & (0xFEDCBA9876543210 + 1)) ? 1 : -1];
|
|
char blci64[__blci_u64(0xFEDCBA9876543210) == (0xFEDCBA9876543210 | ~(0xFEDCBA9876543210 + 1)) ? 1 : -1];
|
|
char blcmsk64[__blcmsk_u64(0xFEDCBA9876543210) == (0xFEDCBA9876543210 ^ (0xFEDCBA9876543210 + 1)) ? 1 : -1];
|
|
char blcs64[__blcs_u64(0xFEDCBA9876543210) == (0xFEDCBA9876543210 | (0xFEDCBA9876543210 + 1)) ? 1 : -1];
|
|
char blsfill64[__blsfill_u64(0xFEDCBA9876543210) == (0xFEDCBA9876543210 | (0xFEDCBA9876543210 - 1)) ? 1 : -1];
|
|
char blsic64[__blsic_u64(0xFEDCBA9876543210) == (~0xFEDCBA9876543210 | (0xFEDCBA9876543210 - 1)) ? 1 : -1];
|
|
char t1mskc64[__t1mskc_u64(0xFEDCBA9876543210) == (~0xFEDCBA9876543210 | (0xFEDCBA9876543210 + 1)) ? 1 : -1];
|
|
char tzmsk64[__tzmsk_u64(0xFEDCBA9876543210) == (~0xFEDCBA9876543210 & (0xFEDCBA9876543210 - 1)) ? 1 : -1];
|
|
#endif
|
|
#endif
|