Skip to content

Commit a153e83

Browse files
rampitecrampitec
authored
[AMDGPU] gfx1250 v_wmma_scale[16]_f32_16x16x128_f8f6f4 codegen (#152036)
1 parent c598676 commit a153e83

File tree

14 files changed

+2193
-1
lines changed

14 files changed

+2193
-1
lines changed

‎clang/include/clang/Basic/BuiltinsAMDGPU.def

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -790,6 +790,8 @@ TARGET_BUILTIN(__builtin_amdgcn_wmma_f32_16x16x128_fp8_fp8, "V8fV16iV16iIsV8fIbI
790790
TARGET_BUILTIN(__builtin_amdgcn_wmma_f32_16x16x128_fp8_bf8, "V8fV16iV16iIsV8fIbIb", "nc", "gfx1250-insts,wavefrontsize32")
791791
TARGET_BUILTIN(__builtin_amdgcn_wmma_f32_16x16x128_bf8_fp8, "V8fV16iV16iIsV8fIbIb", "nc", "gfx1250-insts,wavefrontsize32")
792792
TARGET_BUILTIN(__builtin_amdgcn_wmma_f32_16x16x128_bf8_bf8, "V8fV16iV16iIsV8fIbIb", "nc", "gfx1250-insts,wavefrontsize32")
793+
TARGET_BUILTIN(__builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4, "V8fIiV16iIiV16iIsV8fIiIiiIiIiiIbIb", "nc", "gfx1250-insts,wavefrontsize32")
794+
TARGET_BUILTIN(__builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4, "V8fIiV16iIiV16iIsV8fIiIiLiIiIiLiIbIb", "nc", "gfx1250-insts,wavefrontsize32")
793795
TARGET_BUILTIN(__builtin_amdgcn_wmma_f32_16x16x32_f16, "V8fIbV16hIbV16hIsV8fIbIb", "nc", "gfx1250-insts,wavefrontsize32")
794796
TARGET_BUILTIN(__builtin_amdgcn_wmma_f16_16x16x32_f16, "V8hIbV16hIbV16hIsV8hIbIb", "nc", "gfx1250-insts,wavefrontsize32")
795797
TARGET_BUILTIN(__builtin_amdgcn_wmma_f32_32x16x128_f4, "V16fV16iV8iIsV16f", "nc", "gfx1250-insts,wavefrontsize32")

‎clang/lib/CodeGen/TargetBuiltins/AMDGPU.cpp

Lines changed: 10 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -892,6 +892,8 @@ Value *CodeGenFunction::EmitAMDGPUBuiltinExpr(unsigned BuiltinID,
892892
case AMDGPU::BI__builtin_amdgcn_wmma_i32_16x16x64_iu8:
893893
case AMDGPU::BI__builtin_amdgcn_wmma_f32_16x16x128_f8f6f4:
894894
case AMDGPU::BI__builtin_amdgcn_wmma_f32_32x16x128_f4:
895+
case AMDGPU::BI__builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4:
896+
case AMDGPU::BI__builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4:
895897
case AMDGPU::BI__builtin_amdgcn_swmmac_f32_16x16x64_f16:
896898
case AMDGPU::BI__builtin_amdgcn_swmmac_f32_16x16x64_bf16:
897899
case AMDGPU::BI__builtin_amdgcn_swmmac_f16_16x16x64_f16:
@@ -1158,6 +1160,14 @@ Value *CodeGenFunction::EmitAMDGPUBuiltinExpr(unsigned BuiltinID,
11581160
ArgsForMatchingMatrixTypes = {5, 1, 3};
11591161
BuiltinWMMAOp = Intrinsic::amdgcn_wmma_f32_16x16x128_f8f6f4;
11601162
break;
1163+
case AMDGPU::BI__builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4:
1164+
ArgsForMatchingMatrixTypes = {5, 1, 3};
1165+
BuiltinWMMAOp = Intrinsic::amdgcn_wmma_scale_f32_16x16x128_f8f6f4;
1166+
break;
1167+
case AMDGPU::BI__builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4:
1168+
ArgsForMatchingMatrixTypes = {5, 1, 3};
1169+
BuiltinWMMAOp = Intrinsic::amdgcn_wmma_scale16_f32_16x16x128_f8f6f4;
1170+
break;
11611171
case AMDGPU::BI__builtin_amdgcn_wmma_f32_32x16x128_f4:
11621172
ArgsForMatchingMatrixTypes = {3, 0, 1};
11631173
BuiltinWMMAOp = Intrinsic::amdgcn_wmma_f32_32x16x128_f4;

‎clang/test/CodeGenOpenCL/builtins-amdgcn-gfx1250-wmma-w32.cl

Lines changed: 24 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -169,6 +169,30 @@ void test_amdgcn_wmma_f32_16x16x128_f8f6f4(global v8f* out, v16i a, v16i b, v8f
169169
*out = __builtin_amdgcn_wmma_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c);
170170
}
171171

172+
// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_scale_f32_16x16x128_f8f6f4(
173+
// CHECK-GFX1250-NEXT: entry:
174+
// CHECK-GFX1250-NEXT: [[TMP0:%.*]] = shufflevector <16 x i32> [[B:%.*]], <16 x i32> poison, <12 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
175+
// CHECK-GFX1250-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.scale.f32.16x16x128.f8f6f4.v8f32.v16i32.v12i32(i32 1, <16 x i32> [[A:%.*]], i32 2, <12 x i32> [[TMP0]], i16 0, <8 x float> [[C:%.*]], i32 1, i32 2, i32 [[SCALE_SRC0:%.*]], i32 2, i32 1, i32 [[SCALE_SRC1:%.*]], i1 false, i1 true)
176+
// CHECK-GFX1250-NEXT: store <8 x float> [[TMP1]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA4]]
177+
// CHECK-GFX1250-NEXT: ret void
178+
//
179+
void test_amdgcn_wmma_scale_f32_16x16x128_f8f6f4(global v8f* out, v16i a, v16i b, v8f c, int scale_src0, int scale_src1)
180+
{
181+
*out = __builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c, 1, 2, scale_src0, 2, 1, scale_src1, 0, 1);
182+
}
183+
184+
// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4(
185+
// CHECK-GFX1250-NEXT: entry:
186+
// CHECK-GFX1250-NEXT: [[TMP0:%.*]] = shufflevector <16 x i32> [[B:%.*]], <16 x i32> poison, <12 x i32> <i32 0, i32 1, i32 2, i32 3, i32 4, i32 5, i32 6, i32 7, i32 8, i32 9, i32 10, i32 11>
187+
// CHECK-GFX1250-NEXT: [[TMP1:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.scale16.f32.16x16x128.f8f6f4.v8f32.v16i32.v12i32(i32 1, <16 x i32> [[A:%.*]], i32 2, <12 x i32> [[TMP0]], i16 0, <8 x float> [[C:%.*]], i32 1, i32 2, i64 [[SCALE_SRC0:%.*]], i32 2, i32 1, i64 [[SCALE_SRC1:%.*]], i1 false, i1 true)
188+
// CHECK-GFX1250-NEXT: store <8 x float> [[TMP1]], ptr addrspace(1) [[OUT:%.*]], align 32, !tbaa [[TBAA4]]
189+
// CHECK-GFX1250-NEXT: ret void
190+
//
191+
void test_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4(global v8f* out, v16i a, v16i b, v8f c, long scale_src0, long scale_src1)
192+
{
193+
*out = __builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c, 1, 2, scale_src0, 2, 1, scale_src1, 0, 1);
194+
}
195+
172196
// CHECK-GFX1250-LABEL: @test_amdgcn_wmma_f32_16x16x32_f16(
173197
// CHECK-GFX1250-NEXT: entry:
174198
// CHECK-GFX1250-NEXT: [[TMP0:%.*]] = tail call <8 x float> @llvm.amdgcn.wmma.f32.16x16x32.f16.v8f32.v16f16(i1 false, <16 x half> [[A:%.*]], i1 false, <16 x half> [[B:%.*]], i16 0, <8 x float> [[C:%.*]], i1 false, i1 true)

‎clang/test/SemaOpenCL/builtins-amdgcn-error-gfx1250-wmma-w32-param.cl

Lines changed: 30 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -121,6 +121,36 @@ void test_amdgcn_wmma_f32_16x16x128_f8f6f4(global v8f* out, v16i a, v16i b, v8f
121121
*out = __builtin_amdgcn_wmma_f32_16x16x128_f8f6f4(1, a, 2, b, mod, c); // expected-error {{'__builtin_amdgcn_wmma_f32_16x16x128_f8f6f4' must be a constant integer}}
122122
}
123123

124+
void test_amdgcn_wmma_scale_f32_16x16x128_f8f6f4(global v8f* out, v16i a, v16i b, v8f c, int mod, int scale_src0, int scale_src1, bool reuse)
125+
{
126+
*out = __builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4(mod, a, 2, b, 0, c, 1, 0, scale_src0, 2, 0, scale_src1, 1, 0); // expected-error {{'__builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4' must be a constant integer}}
127+
*out = __builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4(1, a, mod, b, 0, c, 1, 0, scale_src0, 2, 0, scale_src1, 0, 1); // expected-error {{'__builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4' must be a constant integer}}
128+
*out = __builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4(1, a, 2, b, mod, c, 1, 0, scale_src0, 2, 0, scale_src1, 1, 0); // expected-error {{'__builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4' must be a constant integer}}
129+
*out = __builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c, mod, 0, scale_src0, 2, 0, scale_src1, 0, 0); // expected-error {{'__builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4' must be a constant integer}}
130+
*out = __builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c, 1, 0, scale_src0, mod, 0, scale_src1, 0, 0); // expected-error {{'__builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4' must be a constant integer}}
131+
*out = __builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c, 1, 0, scale_src0, 2, 0, scale_src1, reuse, 0); // expected-error {{'__builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4' must be a constant integer}}
132+
*out = __builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c, 1, 0, scale_src0, 2, 0, scale_src1, 0, reuse); // expected-error {{'__builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4' must be a constant integer}}
133+
*out = __builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c, 1, mod, scale_src0, 2, 0, scale_src1, 1, 0); // expected-error {{'__builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4' must be a constant integer}}
134+
*out = __builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c, 1, 0, scale_src0, 2, mod, scale_src1, 1, 0); // expected-error {{'__builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4' must be a constant integer}}
135+
*out = __builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c, 1, 0, scale_src0, 2, 0, scale_src1, mod, 0); // expected-error {{'__builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4' must be a constant integer}}
136+
*out = __builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c, 1, 0, scale_src0, 2, 0, scale_src1, 1, mod); // expected-error {{'__builtin_amdgcn_wmma_scale_f32_16x16x128_f8f6f4' must be a constant integer}}
137+
}
138+
139+
void test_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4(global v8f* out, v16i a, v16i b, v8f c, int mod, long scale_src0, long scale_src1, bool reuse)
140+
{
141+
*out = __builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4(mod, a, 2, b, 0, c, 1, 0, scale_src0, 2, 0, scale_src1, 1, 0); // expected-error {{'__builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4' must be a constant integer}}
142+
*out = __builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4(1, a, mod, b, 0, c, 1, 0, scale_src0, 2, 0, scale_src1, 0, 1); // expected-error {{'__builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4' must be a constant integer}}
143+
*out = __builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4(1, a, 2, b, mod, c, 1, 0, scale_src0, 2, 0, scale_src1, 1, 0); // expected-error {{'__builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4' must be a constant integer}}
144+
*out = __builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c, mod, 0, scale_src0, 2, 0, scale_src1, 0, 0); // expected-error {{'__builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4' must be a constant integer}}
145+
*out = __builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c, 1, 0, scale_src0, mod, 0, scale_src1, 0, 0); // expected-error {{'__builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4' must be a constant integer}}
146+
*out = __builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c, 1, 0, scale_src0, 2, 0, scale_src1, reuse, 0); // expected-error {{'__builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4' must be a constant integer}}
147+
*out = __builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c, 1, 0, scale_src0, 2, 0, scale_src1, 0, reuse); // expected-error {{'__builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4' must be a constant integer}}
148+
*out = __builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c, 1, mod, scale_src0, 2, 0, scale_src1, 1, 0); // expected-error {{'__builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4' must be a constant integer}}
149+
*out = __builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c, 1, 0, scale_src0, 2, mod, scale_src1, 1, 0); // expected-error {{'__builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4' must be a constant integer}}
150+
*out = __builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c, 1, 0, scale_src0, 2, 0, scale_src1, mod, 0); // expected-error {{'__builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4' must be a constant integer}}
151+
*out = __builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4(1, a, 2, b, 0, c, 1, 0, scale_src0, 2, 0, scale_src1, 1, mod); // expected-error {{'__builtin_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4' must be a constant integer}}
152+
}
153+
124154
void test_amdgcn_wmma_f32_16x16x32_f16(global v8f* out, v16h a, v16h b, v8f c, int mod)
125155
{
126156
*out = __builtin_amdgcn_wmma_f32_16x16x32_f16(mod, a, 0, b, 0, c, false, false); // expected-error {{'__builtin_amdgcn_wmma_f32_16x16x32_f16' must be a constant integer}}

‎llvm/include/llvm/IR/IntrinsicsAMDGPU.td

Lines changed: 26 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -3932,6 +3932,30 @@ class AMDGPUWmmaIntrinsicModsC_MatrixFMT :
39323932
[IntrNoMem, IntrConvergent, ImmArg<ArgIndex<0>>, ImmArg<ArgIndex<2>>, ImmArg<ArgIndex<4>>, IntrWillReturn, IntrNoCallback, IntrNoFree]
39333933
>;
39343934

3935+
class AMDGPUWmmaScaleIntrinsicModsC<LLVMType scale_ty> :
3936+
Intrinsic<
3937+
[llvm_anyfloat_ty], // %D
3938+
[
3939+
llvm_i32_ty, // matrix_a_fmt
3940+
llvm_anyint_ty, // %A
3941+
llvm_i32_ty, // matrix_b_fmt
3942+
llvm_anyint_ty, // %B
3943+
llvm_i16_ty, // %C_mod: 0 - none, 1 - neg, 2 - abs, 3 - neg(abs)
3944+
LLVMMatchType<0>, // %C
3945+
llvm_i32_ty, // matrix_a_scale
3946+
llvm_i32_ty, // matrix_a_scale_fmt
3947+
scale_ty, // matrix a scale exponential
3948+
llvm_i32_ty, // matrix_b_scale
3949+
llvm_i32_ty, // matrix_b_scale_fmt
3950+
scale_ty, // matrix b scale exponential
3951+
llvm_i1_ty, // matrix_a_reuse
3952+
llvm_i1_ty, // matrix_b_reuse
3953+
],
3954+
[IntrNoMem, IntrConvergent, ImmArg<ArgIndex<0>>, ImmArg<ArgIndex<2>>, ImmArg<ArgIndex<4>>, ImmArg<ArgIndex<6>>,
3955+
ImmArg<ArgIndex<7>>, ImmArg<ArgIndex<9>>, ImmArg<ArgIndex<10>>, ImmArg<ArgIndex<12>>, ImmArg<ArgIndex<13>>,
3956+
IntrWillReturn, IntrNoCallback, IntrNoFree]
3957+
>;
3958+
39353959
defset list<Intrinsic> AMDGPUWMMAIntrinsicsGFX1250 = {
39363960
def int_amdgcn_wmma_f32_16x16x4_f32 : AMDGPUWmmaIntrinsicModsAllReuse<llvm_anyfloat_ty, llvm_anyfloat_ty>;
39373961
def int_amdgcn_wmma_f32_16x16x32_bf16 : AMDGPUWmmaIntrinsicModsAllReuse<llvm_anyfloat_ty, llvm_anyfloat_ty>;
@@ -3957,6 +3981,8 @@ def int_amdgcn_wmma_f32_16x16x128_bf8_fp8 : AMDGPUWmmaIntrinsicModsC<llvm_anyint
39573981
def int_amdgcn_wmma_f32_16x16x128_bf8_bf8 : AMDGPUWmmaIntrinsicModsC<llvm_anyint_ty, llvm_anyfloat_ty>;
39583982
def int_amdgcn_wmma_i32_16x16x64_iu8 : AMDGPUWmmaIntrinsicModsAB<llvm_anyint_ty, llvm_anyint_ty>;
39593983
def int_amdgcn_wmma_f32_16x16x128_f8f6f4 : AMDGPUWmmaIntrinsicModsC_MatrixFMT;
3984+
def int_amdgcn_wmma_scale_f32_16x16x128_f8f6f4 : AMDGPUWmmaScaleIntrinsicModsC<llvm_i32_ty>;
3985+
def int_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4 : AMDGPUWmmaScaleIntrinsicModsC<llvm_i64_ty>;
39603986
def int_amdgcn_wmma_f32_32x16x128_f4 : AMDGPUWmmaIntrinsicF4ModsC<llvm_anyint_ty, llvm_anyint_ty, llvm_anyfloat_ty>;
39613987
}
39623988

‎llvm/lib/Target/AMDGPU/AMDGPUInstCombineIntrinsic.cpp

Lines changed: 3 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -1694,7 +1694,9 @@ GCNTTIImpl::instCombineIntrinsic(InstCombiner &IC, IntrinsicInst &II) const {
16941694
NewII->takeName(&II);
16951695
return IC.replaceInstUsesWith(II, NewII);
16961696
}
1697-
case Intrinsic::amdgcn_wmma_f32_16x16x128_f8f6f4: {
1697+
case Intrinsic::amdgcn_wmma_f32_16x16x128_f8f6f4:
1698+
case Intrinsic::amdgcn_wmma_scale_f32_16x16x128_f8f6f4:
1699+
case Intrinsic::amdgcn_wmma_scale16_f32_16x16x128_f8f6f4: {
16981700
Value *Src0 = II.getArgOperand(1);
16991701
Value *Src1 = II.getArgOperand(3);
17001702
unsigned FmtA = cast<ConstantInt>(II.getArgOperand(0))->getZExtValue();

‎llvm/lib/Target/AMDGPU/AMDGPURegisterBankInfo.cpp

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -4798,6 +4798,8 @@ AMDGPURegisterBankInfo::getInstrMapping(const MachineInstr &MI) const {
47984798
case Intrinsic::amdgcn_wmma_f32_16x16x128_bf8_bf8:
47994799
case Intrinsic::amdgcn_wmma_i32_16x16x64_iu8:
48004800
case Intrinsic::amdgcn_wmma_f32_16x16x128_f8f6f4:
4801+
case Intrinsic::amdgcn_wmma_scale_f32_16x16x128_f8f6f4:
4802+
case Intrinsic::amdgcn_wmma_scale16_f32_16x16x128_f8f6f4:
48014803
case Intrinsic::amdgcn_wmma_f32_32x16x128_f4:
48024804
case Intrinsic::amdgcn_swmmac_f16_16x16x64_f16:
48034805
case Intrinsic::amdgcn_swmmac_bf16_16x16x64_bf16:

‎llvm/lib/Target/AMDGPU/VOP3PInstructions.td

Lines changed: 2 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -2009,6 +2009,8 @@ let SubtargetPredicate = isGFX125xOnly in {
20092009

20102010
foreach I = ["f8_f8", "f8_f6", "f8_f4", "f6_f8", "f6_f6", "f6_f4", "f4_f8", "f4_f6", "f4_f4"] in {
20112011
defm : WMMAPat<"V_WMMA_F32_16X16X128_F8F6F4_" # I # "_w32", int_amdgcn_wmma_f32_16x16x128_f8f6f4, !cast<VOP3PWMMA_Profile>("F32_16X16X128_F8F6F4_" # I # "_w32")>;
2012+
defm : WMMAPat<"V_WMMA_SCALE_F32_16X16X128_F8F6F4_" # I # "_w32", int_amdgcn_wmma_scale_f32_16x16x128_f8f6f4, !cast<VOP3PWMMA_Profile>("F32_16X16X128_F8F6F4_SCALE_" # I # "_w32")>;
2013+
defm : WMMAPat<"V_WMMA_SCALE16_F32_16X16X128_F8F6F4_" # I # "_w32", int_amdgcn_wmma_scale16_f32_16x16x128_f8f6f4, !cast<VOP3PWMMA_Profile>("F32_16X16X128_F8F6F4_SCALE16_" # I # "_w32")>;
20122014
}
20132015

20142016
def : SWMMACPat<V_SWMMAC_F32_16X16X64_BF16_w32_twoaddr, int_amdgcn_swmmac_f32_16x16x64_bf16, F32_BF16X64_SWMMAC_w32>;

‎llvm/test/Analysis/UniformityAnalysis/AMDGPU/intrinsics.ll

Lines changed: 18 additions & 0 deletions
Original file line numberDiff line numberDiff line change
@@ -310,6 +310,22 @@ bb:
310310
ret void
311311
}
312312

313+
; CHECK: DIVERGENT: %tmp0 = call <8 x float> @llvm.amdgcn.wmma.scale.f32.16x16x128.f8f6f4.v8f32.v16i32.v16i32(i32 0, <16 x i32> %A, i32 0, <16 x i32> %B, i16 0, <8 x float> %C, i32 1, i32 0, i32 1, i32 1, i32 0, i32 1, i1 false, i1 false)
314+
define amdgpu_ps void @wmma_scale_f32_16x16x128_f8f6f4(<16 x i32> %A, <16 x i32> %B, <8 x float> %C, ptr addrspace(1) %out) {
315+
bb:
316+
%tmp0 = call <8 x float> @llvm.amdgcn.wmma.scale.f32.16x16x128.f8f6f4.v8f32.v16i32.v16i32(i32 0, <16 x i32> %A, i32 0, <16 x i32> %B, i16 0, <8 x float> %C, i32 1, i32 0, i32 1, i32 1, i32 0, i32 1, i1 false, i1 false)
317+
store <8 x float> %tmp0, ptr addrspace(1) %out
318+
ret void
319+
}
320+
321+
; CHECK: DIVERGENT: %tmp0 = call <8 x float> @llvm.amdgcn.wmma.scale16.f32.16x16x128.f8f6f4.v8f32.v16i32.v16i32(i32 0, <16 x i32> %A, i32 0, <16 x i32> %B, i16 0, <8 x float> %C, i32 1, i32 0, i64 1, i32 1, i32 0, i64 1, i1 false, i1 false)
322+
define amdgpu_ps void @wmma_scale16_f32_16x16x128_f8f6f4(<16 x i32> %A, <16 x i32> %B, <8 x float> %C, ptr addrspace(1) %out) {
323+
bb:
324+
%tmp0 = call <8 x float> @llvm.amdgcn.wmma.scale16.f32.16x16x128.f8f6f4.v8f32.v16i32.v16i32(i32 0, <16 x i32> %A, i32 0, <16 x i32> %B, i16 0, <8 x float> %C, i32 1, i32 0, i64 1, i32 1, i32 0, i64 1, i1 false, i1 false)
325+
store <8 x float> %tmp0, ptr addrspace(1) %out
326+
ret void
327+
}
328+
313329
; CHRCK: DIVERGENT: %tmp0 = call <8 x float> @llvm.amdgcn.swmmac.f32.16x16x64.f16.v8f32.v16f16.v32f16.i16(i1 false, <16 x half> %A, i1 false, <32 x half> %B, <8 x float> %C, i16 %Index, i1 false, i1 false)
314330
define amdgpu_ps void @swmmac_f32_16x16x64_f16(<16 x half> %A, <32 x half> %B, <8 x float> %C, i16 %Index, ptr addrspace(1) %out) {
315331
%tmp0 = call <8 x float> @llvm.amdgcn.swmmac.f32.16x16x64.f16.v8f32.v16f16.v32f16.i16(i1 0, <16 x half> %A, i1 0, <32 x half> %B, <8 x float> %C, i16 %Index, i1 false, i1 false)
@@ -880,6 +896,8 @@ declare <8 x half> @llvm.amdgcn.wmma.f16.16x16x64.bf8.fp8.v8f16.v8i32(<8 x i32>,
880896
declare <8 x half> @llvm.amdgcn.wmma.f16.16x16x64.bf8.bf8.v8f16.v8i32(<8 x i32>, <8 x i32>, i16, <8 x half>, i1, i1)
881897
declare <8 x i32> @llvm.amdgcn.wmma.i32.16x16x64.iu8.v8i32.v8i32(i1 immarg, <8 x i32>, i1 immarg, <8 x i32>, <8 x i32>, i1, i1)
882898
declare <8 x float> @llvm.amdgcn.wmma.f32.16x16x128.f8f6f4.v8f32.v16i32.v16i32(i32, <16 x i32>, i32, <16 x i32>, i16, <8 x float>)
899+
declare <8 x float> @llvm.amdgcn.wmma.scale.f32.16x16x128.f8f6f4.v8f32.v16i32.v16i32(i32, <16 x i32>, i32, <16 x i32>, i16, <8 x float>, i32, i32, i32, i32, i32, i32, i1, i1)
900+
declare <8 x float> @llvm.amdgcn.wmma.scale16.f32.16x16x128.f8f6f4.v8f32.v16i32.v16i32(i32, <16 x i32>, i32, <16 x i32>, i16, <8 x float>, i32, i32, i64, i32, i32, i64, i1, i1)
883901
declare <8 x float> @llvm.amdgcn.swmmac.f32.16x16x64.f16.v8f32.v16f16.v32f16.i16(i1, <16 x half>, i1, <32 x half>, <8 x float>, i16, i1, i1)
884902
declare <8 x float> @llvm.amdgcn.swmmac.f32.16x16x64.bf16.v8f32.v16bf16.v32bf16.i16(i1, <16 x bfloat>, i1, <32 x bfloat>, <8 x float>, i16, i1, i1)
885903
declare <8 x half> @llvm.amdgcn.swmmac.f16.16x16x64.f16.v8f16.v16f16.v32f16.i16(i1, <16 x half>, i1, <32 x half>, <8 x half>, i16, i1, i1)

0 commit comments

Comments
 (0)