146 lines · cpp
1// NOTE: Assertions have been autogenerated by utils/update_cc_test_checks.py UTC_ARGS: --version 22// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -target-feature +fullbf16 -emit-llvm %s -o - | FileCheck %s3// RUN: %clang_cc1 -triple x86_64-unknown-linux-gnu -emit-llvm %s -o - | FileCheck -check-prefix=CHECK-NBF16 %s4 5// CHECK-LABEL: define dso_local void @_Z11test_scalarDF16bDF16b6// CHECK-SAME: (bfloat noundef [[A:%.*]], bfloat noundef [[B:%.*]]) #[[ATTR0:[0-9]+]] {7// CHECK: [[A_ADDR:%.*]] = alloca bfloat, align 28// CHECK-NEXT: [[B_ADDR:%.*]] = alloca bfloat, align 29// CHECK-NEXT: [[C:%.*]] = alloca bfloat, align 210// CHECK-NEXT: store bfloat [[A]], ptr [[A_ADDR]], align 211// CHECK-NEXT: store bfloat [[B]], ptr [[B_ADDR]], align 212// CHECK-NEXT: [[TMP0:%.*]] = load bfloat, ptr [[A_ADDR]], align 213// CHECK-NEXT: [[TMP1:%.*]] = load bfloat, ptr [[B_ADDR]], align 214// CHECK-NEXT: [[ADD:%.*]] = fadd bfloat [[TMP0]], [[TMP1]]15// CHECK-NEXT: store bfloat [[ADD]], ptr [[C]], align 216// CHECK-NEXT: [[TMP2:%.*]] = load bfloat, ptr [[A_ADDR]], align 217// CHECK-NEXT: [[TMP3:%.*]] = load bfloat, ptr [[B_ADDR]], align 218// CHECK-NEXT: [[SUB:%.*]] = fsub bfloat [[TMP2]], [[TMP3]]19// CHECK-NEXT: store bfloat [[SUB]], ptr [[C]], align 220// CHECK-NEXT: [[TMP4:%.*]] = load bfloat, ptr [[A_ADDR]], align 221// CHECK-NEXT: [[TMP5:%.*]] = load bfloat, ptr [[B_ADDR]], align 222// CHECK-NEXT: [[MUL:%.*]] = fmul bfloat [[TMP4]], [[TMP5]]23// CHECK-NEXT: store bfloat [[MUL]], ptr [[C]], align 224// CHECK-NEXT: [[TMP6:%.*]] = load bfloat, ptr [[A_ADDR]], align 225// CHECK-NEXT: [[TMP7:%.*]] = load bfloat, ptr [[B_ADDR]], align 226// CHECK-NEXT: [[DIV:%.*]] = fdiv bfloat [[TMP6]], [[TMP7]]27// CHECK-NEXT: store bfloat [[DIV]], ptr [[C]], align 228// CHECK-NEXT: ret void29//30// CHECK-NBF16-LABEL: define dso_local void @_Z11test_scalarDF16bDF16b31// CHECK-NBF16-SAME: (bfloat noundef [[A:%.*]], bfloat noundef [[B:%.*]]) #[[ATTR0:[0-9]+]] {32// CHECK-NBF16: [[A_ADDR:%.*]] = alloca bfloat, align 233// CHECK-NBF16-NEXT: [[B_ADDR:%.*]] = alloca bfloat, align 234// CHECK-NBF16-NEXT: [[C:%.*]] = alloca bfloat, align 235// CHECK-NBF16-NEXT: store bfloat [[A]], ptr [[A_ADDR]], align 236// CHECK-NBF16-NEXT: store bfloat [[B]], ptr [[B_ADDR]], align 237// CHECK-NBF16-NEXT: [[TMP0:%.*]] = load bfloat, ptr [[A_ADDR]], align 238// CHECK-NBF16-NEXT: [[EXT:%.*]] = fpext bfloat [[TMP0]] to float39// CHECK-NBF16-NEXT: [[TMP1:%.*]] = load bfloat, ptr [[B_ADDR]], align 240// CHECK-NBF16-NEXT: [[EXT1:%.*]] = fpext bfloat [[TMP1]] to float41// CHECK-NBF16-NEXT: [[ADD:%.*]] = fadd float [[EXT]], [[EXT1]]42// CHECK-NBF16-NEXT: [[UNPROMOTION:%.*]] = fptrunc float [[ADD]] to bfloat43// CHECK-NBF16-NEXT: store bfloat [[UNPROMOTION]], ptr [[C]], align 244// CHECK-NBF16-NEXT: [[TMP2:%.*]] = load bfloat, ptr [[A_ADDR]], align 245// CHECK-NBF16-NEXT: [[EXT2:%.*]] = fpext bfloat [[TMP2]] to float46// CHECK-NBF16-NEXT: [[TMP3:%.*]] = load bfloat, ptr [[B_ADDR]], align 247// CHECK-NBF16-NEXT: [[EXT3:%.*]] = fpext bfloat [[TMP3]] to float48// CHECK-NBF16-NEXT: [[SUB:%.*]] = fsub float [[EXT2]], [[EXT3]]49// CHECK-NBF16-NEXT: [[UNPROMOTION4:%.*]] = fptrunc float [[SUB]] to bfloat50// CHECK-NBF16-NEXT: store bfloat [[UNPROMOTION4]], ptr [[C]], align 251// CHECK-NBF16-NEXT: [[TMP4:%.*]] = load bfloat, ptr [[A_ADDR]], align 252// CHECK-NBF16-NEXT: [[EXT5:%.*]] = fpext bfloat [[TMP4]] to float53// CHECK-NBF16-NEXT: [[TMP5:%.*]] = load bfloat, ptr [[B_ADDR]], align 254// CHECK-NBF16-NEXT: [[EXT6:%.*]] = fpext bfloat [[TMP5]] to float55// CHECK-NBF16-NEXT: [[MUL:%.*]] = fmul float [[EXT5]], [[EXT6]]56// CHECK-NBF16-NEXT: [[UNPROMOTION7:%.*]] = fptrunc float [[MUL]] to bfloat57// CHECK-NBF16-NEXT: store bfloat [[UNPROMOTION7]], ptr [[C]], align 258// CHECK-NBF16-NEXT: [[TMP6:%.*]] = load bfloat, ptr [[A_ADDR]], align 259// CHECK-NBF16-NEXT: [[EXT8:%.*]] = fpext bfloat [[TMP6]] to float60// CHECK-NBF16-NEXT: [[TMP7:%.*]] = load bfloat, ptr [[B_ADDR]], align 261// CHECK-NBF16-NEXT: [[EXT9:%.*]] = fpext bfloat [[TMP7]] to float62// CHECK-NBF16-NEXT: [[DIV:%.*]] = fdiv float [[EXT8]], [[EXT9]]63// CHECK-NBF16-NEXT: [[UNPROMOTION10:%.*]] = fptrunc float [[DIV]] to bfloat64// CHECK-NBF16-NEXT: store bfloat [[UNPROMOTION10]], ptr [[C]], align 265// CHECK-NBF16-NEXT: ret void66//67void test_scalar(__bf16 a, __bf16 b) {68 __bf16 c;69 c = a + b;70 c = a - b;71 c = a * b;72 c = a / b;73}74 75typedef __bf16 v8bfloat16 __attribute__((__vector_size__(16)));76 77// CHECK-LABEL: define dso_local void @_Z11test_vectorDv8_DF16bS_78// CHECK-SAME: (<8 x bfloat> noundef [[A:%.*]], <8 x bfloat> noundef [[B:%.*]]) #[[ATTR0:[0-9]+]] {79// CHECK: [[A_ADDR:%.*]] = alloca <8 x bfloat>, align 1680// CHECK-NEXT: [[B_ADDR:%.*]] = alloca <8 x bfloat>, align 1681// CHECK-NEXT: [[C:%.*]] = alloca <8 x bfloat>, align 1682// CHECK-NEXT: store <8 x bfloat> [[A]], ptr [[A_ADDR]], align 1683// CHECK-NEXT: store <8 x bfloat> [[B]], ptr [[B_ADDR]], align 1684// CHECK-NEXT: [[TMP0:%.*]] = load <8 x bfloat>, ptr [[A_ADDR]], align 1685// CHECK-NEXT: [[TMP1:%.*]] = load <8 x bfloat>, ptr [[B_ADDR]], align 1686// CHECK-NEXT: [[ADD:%.*]] = fadd <8 x bfloat> [[TMP0]], [[TMP1]]87// CHECK-NEXT: store <8 x bfloat> [[ADD]], ptr [[C]], align 1688// CHECK-NEXT: [[TMP2:%.*]] = load <8 x bfloat>, ptr [[A_ADDR]], align 1689// CHECK-NEXT: [[TMP3:%.*]] = load <8 x bfloat>, ptr [[B_ADDR]], align 1690// CHECK-NEXT: [[SUB:%.*]] = fsub <8 x bfloat> [[TMP2]], [[TMP3]]91// CHECK-NEXT: store <8 x bfloat> [[SUB]], ptr [[C]], align 1692// CHECK-NEXT: [[TMP4:%.*]] = load <8 x bfloat>, ptr [[A_ADDR]], align 1693// CHECK-NEXT: [[TMP5:%.*]] = load <8 x bfloat>, ptr [[B_ADDR]], align 1694// CHECK-NEXT: [[MUL:%.*]] = fmul <8 x bfloat> [[TMP4]], [[TMP5]]95// CHECK-NEXT: store <8 x bfloat> [[MUL]], ptr [[C]], align 1696// CHECK-NEXT: [[TMP6:%.*]] = load <8 x bfloat>, ptr [[A_ADDR]], align 1697// CHECK-NEXT: [[TMP7:%.*]] = load <8 x bfloat>, ptr [[B_ADDR]], align 1698// CHECK-NEXT: [[DIV:%.*]] = fdiv <8 x bfloat> [[TMP6]], [[TMP7]]99// CHECK-NEXT: store <8 x bfloat> [[DIV]], ptr [[C]], align 16100// CHECK-NEXT: ret void101//102// CHECK-NBF16-LABEL: define dso_local void @_Z11test_vectorDv8_DF16bS_103// CHECK-NBF16-SAME: (<8 x bfloat> noundef [[A:%.*]], <8 x bfloat> noundef [[B:%.*]]) #[[ATTR0:[0-9]+]] {104// CHECK-NBF16: [[A_ADDR:%.*]] = alloca <8 x bfloat>, align 16105// CHECK-NBF16-NEXT: [[B_ADDR:%.*]] = alloca <8 x bfloat>, align 16106// CHECK-NBF16-NEXT: [[C:%.*]] = alloca <8 x bfloat>, align 16107// CHECK-NBF16-NEXT: store <8 x bfloat> [[A]], ptr [[A_ADDR]], align 16108// CHECK-NBF16-NEXT: store <8 x bfloat> [[B]], ptr [[B_ADDR]], align 16109// CHECK-NBF16-NEXT: [[TMP0:%.*]] = load <8 x bfloat>, ptr [[A_ADDR]], align 16110// CHECK-NBF16-NEXT: [[EXT:%.*]] = fpext <8 x bfloat> [[TMP0]] to <8 x float>111// CHECK-NBF16-NEXT: [[TMP1:%.*]] = load <8 x bfloat>, ptr [[B_ADDR]], align 16112// CHECK-NBF16-NEXT: [[EXT1:%.*]] = fpext <8 x bfloat> [[TMP1]] to <8 x float>113// CHECK-NBF16-NEXT: [[ADD:%.*]] = fadd <8 x float> [[EXT]], [[EXT1]]114// CHECK-NBF16-NEXT: [[UNPROMOTION:%.*]] = fptrunc <8 x float> [[ADD]] to <8 x bfloat>115// CHECK-NBF16-NEXT: store <8 x bfloat> [[UNPROMOTION]], ptr [[C]], align 16116// CHECK-NBF16-NEXT: [[TMP2:%.*]] = load <8 x bfloat>, ptr [[A_ADDR]], align 16117// CHECK-NBF16-NEXT: [[EXT2:%.*]] = fpext <8 x bfloat> [[TMP2]] to <8 x float>118// CHECK-NBF16-NEXT: [[TMP3:%.*]] = load <8 x bfloat>, ptr [[B_ADDR]], align 16119// CHECK-NBF16-NEXT: [[EXT3:%.*]] = fpext <8 x bfloat> [[TMP3]] to <8 x float>120// CHECK-NBF16-NEXT: [[SUB:%.*]] = fsub <8 x float> [[EXT2]], [[EXT3]]121// CHECK-NBF16-NEXT: [[UNPROMOTION4:%.*]] = fptrunc <8 x float> [[SUB]] to <8 x bfloat>122// CHECK-NBF16-NEXT: store <8 x bfloat> [[UNPROMOTION4]], ptr [[C]], align 16123// CHECK-NBF16-NEXT: [[TMP4:%.*]] = load <8 x bfloat>, ptr [[A_ADDR]], align 16124// CHECK-NBF16-NEXT: [[EXT5:%.*]] = fpext <8 x bfloat> [[TMP4]] to <8 x float>125// CHECK-NBF16-NEXT: [[TMP5:%.*]] = load <8 x bfloat>, ptr [[B_ADDR]], align 16126// CHECK-NBF16-NEXT: [[EXT6:%.*]] = fpext <8 x bfloat> [[TMP5]] to <8 x float>127// CHECK-NBF16-NEXT: [[MUL:%.*]] = fmul <8 x float> [[EXT5]], [[EXT6]]128// CHECK-NBF16-NEXT: [[UNPROMOTION7:%.*]] = fptrunc <8 x float> [[MUL]] to <8 x bfloat>129// CHECK-NBF16-NEXT: store <8 x bfloat> [[UNPROMOTION7]], ptr [[C]], align 16130// CHECK-NBF16-NEXT: [[TMP6:%.*]] = load <8 x bfloat>, ptr [[A_ADDR]], align 16131// CHECK-NBF16-NEXT: [[EXT8:%.*]] = fpext <8 x bfloat> [[TMP6]] to <8 x float>132// CHECK-NBF16-NEXT: [[TMP7:%.*]] = load <8 x bfloat>, ptr [[B_ADDR]], align 16133// CHECK-NBF16-NEXT: [[EXT9:%.*]] = fpext <8 x bfloat> [[TMP7]] to <8 x float>134// CHECK-NBF16-NEXT: [[DIV:%.*]] = fdiv <8 x float> [[EXT8]], [[EXT9]]135// CHECK-NBF16-NEXT: [[UNPROMOTION10:%.*]] = fptrunc <8 x float> [[DIV]] to <8 x bfloat>136// CHECK-NBF16-NEXT: store <8 x bfloat> [[UNPROMOTION10]], ptr [[C]], align 16137// CHECK-NBF16-NEXT: ret void138//139void test_vector(v8bfloat16 a, v8bfloat16 b) {140 v8bfloat16 c;141 c = a + b;142 c = a - b;143 c = a * b;144 c = a / b;145}146