1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
|
; RUN: llvm-as < %s -o %t.bc
; RUN: llvm-spirv -s %t.bc -o %t.regularized.bc
; RUN: llvm-dis %t.regularized.bc -o %t.regularized.ll
; RUN: FileCheck < %t.regularized.ll %s --check-prefix=CHECK-REGULARIZED
; RUN: llvm-spirv %t.bc --spirv-ext=+SPV_INTEL_bfloat16_conversion,+SPV_INTEL_joint_matrix -o %t.spv
; RUN: llvm-spirv %t.spv -to-text -o %t.spt
; RUN: FileCheck < %t.spt %s --check-prefix=CHECK-SPIRV
; RUN: llvm-spirv -r %t.spv -o %t.rev.bc
; RUN: llvm-dis < %t.rev.bc | FileCheck %s --check-prefix=CHECK-LLVM
; CHECK-REGULARIZED: %[[Alloca:.*]] = alloca %"class.cl::sycl::ext::intel::experimental::bfloat16", align 2
; CHECK-REGULARIZED: %[[ASCast:.*]] = addrspacecast %"class.cl::sycl::ext::intel::experimental::bfloat16"* %[[Alloca]] to %"class.cl::sycl::ext::intel::experimental::bfloat16" addrspace(4)*
; CHECK-REGULARIZED: %[[GEP1:.*]] = getelementptr inbounds %"class.cl::sycl::ext::intel::experimental::bfloat16", %"class.cl::sycl::ext::intel::experimental::bfloat16" addrspace(4)* %[[ASCast]], i64 0, i32 0
; CHECK-REGULARIZED: %[[#Extract:]] = call spir_func i16 @_Z28__spirv_VectorExtractDynamicIN2cl4sycl3ext5intel12experimental8bfloat16ELm8ELm16ELN5__spv12MatrixLayoutE0ELNS6_5Scope4FlagE3EET_PNS6_24__spirv_JointMatrixINTELISA_XT0_EXT1_EXT2_EXT3_EEEm(%spirv.JointMatrixINTEL._bfloat16_8_16_0_3 addrspace(4)* align 2 %{{.*}}, i64 noundef %{{.*}})
; CHECK-REGULARIZED: %[[#GEP2:]] = getelementptr inbounds %"class.cl::sycl::ext::intel::experimental::bfloat16", %"class.cl::sycl::ext::intel::experimental::bfloat16" addrspace(4)* %[[ASCast]], i32 0, i32 0
; CHECK-REGULARIZED: store i16 %[[#Extract]], i16 addrspace(4)* %[[#GEP2]], align 2
; CHECK-REGULARIZED: %[[#Load:]] = load i16, i16 addrspace(4)* %[[GEP1]], align 2
; CHECK-REGULARIZED: %[[ConvertVal:.*]] = call spir_func noundef float @_Z27__spirv_ConvertBF16ToFINTELt(i16 noundef zeroext %[[#Load]])
; CHECK-REGULARIZED: %{{.*}} = fadd float %[[ConvertVal]], %{{.*}}
; CHECK-SPIRV: TypeInt [[#TypeI16ID:]] 16 0
; CHECK-SPIRV: TypeFloat [[#TypeFID:]] 32
; CHECK-SPIRV: TypeJointMatrixINTEL [[#TypeJointMID:]] [[#TypeI16ID]] [[#]] [[#]] [[#]] [[#]]
; CHECK-SPIRV: Phi [[#TypeJointMID]] [[#PhiID:]] [[#]] [[#]] [[#]] [[#]]
; CHECK-SPIRV: VectorExtractDynamic [[#TypeI16ID]] [[#ExtractID:]] [[#PhiID]] [[#]]
; CHECK-SPIRV: Store [[#PtrID:]] [[#ExtractID]] [[#]] [[#]]
; CHECK-SPIRV: ConvertBF16ToFINTEL [[#TypeFID]] [[#Conv1ID:]] [[#]]
; CHECK-SPIRV: ConvertBF16ToFINTEL [[#TypeFID]] [[#Conv2ID:]] [[#]]
; CHECK-SPIRV: FAdd [[#TypeFID]] [[#ResId:]] [[#Conv1ID]] [[#Conv2ID]]
; CHECK-SPIRV: ConvertFToBF16INTEL [[#TypeI16ID]] [[#]] [[#ResId]]
; CHECK-SPIRV: Load [[#TypeI16ID]] [[#LoadID:]] [[#]] [[#]] [[#]]
; CHECK-SPIRV: VectorInsertDynamic [[#TypeJointMID]] [[#]] [[#PhiID]] [[#LoadID]] [[#]]
; CHECK-LLVM: %spirv.JointMatrixINTEL._short_8_16_0_3
; CHECK-LLVM: %[[GEP1:.*]] = getelementptr inbounds %"class.cl::sycl::ext::intel::experimental::bfloat16", %"class.cl::sycl::ext::intel::experimental::bfloat16" addrspace(4)* %{{.*}}, i64 0, i32 0
; CHECK-LLVM: %[[GEP2:.*]] = getelementptr inbounds %"class.cl::sycl::ext::intel::experimental::bfloat16", %"class.cl::sycl::ext::intel::experimental::bfloat16" addrspace(4)* %{{.*}}, i64 0, i32 0
; CHECK-LLVM: %[[ConvertConst:.*]] = call spir_func i16 @_Z32intel_convert_bfloat16_as_ushortf(float 2.000000e+00)
; CHECK-LLVM: %[[#LoadGEP:]] = load i16, i16 addrspace(4)* %[[GEP2]], align 2
; CHECK-LLVM: %[[ConvertVal:.*]] = call spir_func float @_Z31intel_convert_as_bfloat16_floats(i16 %[[#LoadGEP]])
; CHECK-LLVM: %[[ConvertConstToF:.*]] = call spir_func float @_Z31intel_convert_as_bfloat16_floats(i16 %[[ConvertConst]])
; CHECK-LLVM: %[[FAddRes:.*]] = fadd float %[[ConvertVal]], %[[ConvertConstToF]]
; CHECK-LLVM: %[[ConvertResToBF:.*]] = call spir_func i16 @_Z32intel_convert_bfloat16_as_ushortf(float %[[FAddRes]])
; CHECK-LLVM: store i16 %[[ConvertResToBF]], i16 addrspace(4)* %[[#]], align 2
; ModuleID = 'joint_matrix_bfloat16_test.bc'
source_filename = "joint_matrix_bfloat16_test.cpp"
target datalayout = "e-i64:64-v16:16-v24:32-v32:32-v48:64-v96:128-v192:256-v256:256-v512:512-v1024:1024-n8:16:32:64"
target triple = "spir64-unknown-unknown"
%class.anon = type { %"class.cl::sycl::accessor" }
%"class.cl::sycl::accessor" = type { %"class.cl::sycl::detail::AccessorImplDevice", %union.anon }
%"class.cl::sycl::detail::AccessorImplDevice" = type { %"class.cl::sycl::id", %"class.cl::sycl::range", %"class.cl::sycl::range" }
%"class.cl::sycl::id" = type { %"class.cl::sycl::detail::array" }
%"class.cl::sycl::detail::array" = type { [2 x i64] }
%"class.cl::sycl::range" = type { %"class.cl::sycl::detail::array" }
%union.anon = type { %"class.cl::sycl::ext::intel::experimental::bfloat16" addrspace(1)* }
%"class.cl::sycl::ext::intel::experimental::bfloat16" = type { i16 }
%"class.cl::sycl::nd_item" = type { %"class.cl::sycl::item", %"class.cl::sycl::item.0", %"class.cl::sycl::group" }
%"class.cl::sycl::item" = type { %"struct.cl::sycl::detail::ItemBase" }
%"struct.cl::sycl::detail::ItemBase" = type { %"class.cl::sycl::range", %"class.cl::sycl::id", %"class.cl::sycl::id" }
%"class.cl::sycl::item.0" = type { %"struct.cl::sycl::detail::ItemBase.1" }
%"struct.cl::sycl::detail::ItemBase.1" = type { %"class.cl::sycl::range", %"class.cl::sycl::id" }
%"class.cl::sycl::group" = type { %"class.cl::sycl::range", %"class.cl::sycl::range", %"class.cl::sycl::range", %"class.cl::sycl::id" }
%spirv.JointMatrixINTEL._bfloat16_8_16_0_3 = type opaque
$_ZZZ17matrix_verify_addIN2cl4sycl3ext5intel12experimental8bfloat16ELm16ELm16EEvNS1_5queueER10big_matrixIT_XT0_EXT1_EERNS1_8nd_rangeILi2EEEfENKUlRNS1_7handlerEE_clESF_ENKUlNS1_7nd_itemILi2EEEE_clESI_ = comdat any
; Function Attrs: argmemonly nofree nosync nounwind willreturn
declare void @llvm.lifetime.start.p0i8(i64 immarg, i8* nocapture) #0
; Function Attrs: convergent inlinehint norecurse
define linkonce_odr dso_local spir_func void @_ZZZ17matrix_verify_addIN2cl4sycl3ext5intel12experimental8bfloat16ELm16ELm16EEvNS1_5queueER10big_matrixIT_XT0_EXT1_EERNS1_8nd_rangeILi2EEEfENKUlRNS1_7handlerEE_clESF_ENKUlNS1_7nd_itemILi2EEEE_clESI_(%class.anon addrspace(4)* noundef align 8 dereferenceable_or_null(56) %this, %"class.cl::sycl::nd_item"* noundef byval(%"class.cl::sycl::nd_item") align 8 %spmd_item) local_unnamed_addr #1 comdat align 2 {
entry:
%ref.tmp.i = alloca %"class.cl::sycl::ext::intel::experimental::bfloat16", align 2
%agg.tmp.i54 = alloca %"class.cl::sycl::ext::intel::experimental::bfloat16", align 2
%agg.tmp.i = alloca %"class.cl::sycl::ext::intel::experimental::bfloat16", align 2
%spmd_item.ascast = addrspacecast %"class.cl::sycl::nd_item"* %spmd_item to %"class.cl::sycl::nd_item" addrspace(4)*
%arrayidx.i.i.i = getelementptr inbounds %"class.cl::sycl::nd_item", %"class.cl::sycl::nd_item" addrspace(4)* %spmd_item.ascast, i64 0, i32 0, i32 0, i32 1, i32 0, i32 0, i64 0
%0 = load i64, i64 addrspace(4)* %arrayidx.i.i.i, align 8, !tbaa !5
%cmp.i = icmp ult i64 %0, 2147483648
tail call void @llvm.assume(i1 %cmp.i)
%arrayidx.i.i.i29 = getelementptr inbounds %"class.cl::sycl::nd_item", %"class.cl::sycl::nd_item" addrspace(4)* %spmd_item.ascast, i64 0, i32 0, i32 0, i32 1, i32 0, i32 0, i64 1
%1 = load i64, i64 addrspace(4)* %arrayidx.i.i.i29, align 8, !tbaa !5
%cmp.i30 = icmp ult i64 %1, 2147483648
tail call void @llvm.assume(i1 %cmp.i30)
%arrayidx.i.i.i31 = getelementptr inbounds %"class.cl::sycl::nd_item", %"class.cl::sycl::nd_item" addrspace(4)* %spmd_item.ascast, i64 0, i32 1, i32 0, i32 1, i32 0, i32 0, i64 0
%2 = load i64, i64 addrspace(4)* %arrayidx.i.i.i31, align 8, !tbaa !5
%cmp.i32 = icmp ult i64 %2, 2147483648
tail call void @llvm.assume(i1 %cmp.i32)
%arrayidx.i.i.i33 = getelementptr inbounds %"class.cl::sycl::nd_item", %"class.cl::sycl::nd_item" addrspace(4)* %spmd_item.ascast, i64 0, i32 1, i32 0, i32 1, i32 0, i32 0, i64 1
%3 = load i64, i64 addrspace(4)* %arrayidx.i.i.i33, align 8, !tbaa !5
%cmp.i34 = icmp ult i64 %3, 2147483648
tail call void @llvm.assume(i1 %cmp.i34)
%4 = bitcast %"class.cl::sycl::ext::intel::experimental::bfloat16"* %agg.tmp.i to i8*
call void @llvm.lifetime.start.p0i8(i64 2, i8* nonnull %4)
%agg.tmp.ascast.i = addrspacecast %"class.cl::sycl::ext::intel::experimental::bfloat16"* %agg.tmp.i to %"class.cl::sycl::ext::intel::experimental::bfloat16" addrspace(4)*
%call.i.i.i = tail call spir_func noundef zeroext i16 @_Z27__spirv_ConvertFToBF16INTELf(float noundef 5.000000e+00) #6
%value.i.i = getelementptr inbounds %"class.cl::sycl::ext::intel::experimental::bfloat16", %"class.cl::sycl::ext::intel::experimental::bfloat16" addrspace(4)* %agg.tmp.ascast.i, i64 0, i32 0
store i16 %call.i.i.i, i16 addrspace(4)* %value.i.i, align 2, !tbaa !9
%call.i = tail call spir_func noundef %spirv.JointMatrixINTEL._bfloat16_8_16_0_3 addrspace(4)* @_Z26__spirv_CompositeConstructIN2cl4sycl3ext5intel12experimental8bfloat16ELm8ELm16ELN5__spv12MatrixLayoutE0ELNS6_5Scope4FlagE3EEPNS6_24__spirv_JointMatrixINTELIT_XT0_EXT1_EXT2_EXT3_EEESB_(%"class.cl::sycl::ext::intel::experimental::bfloat16"* noundef nonnull byval(%"class.cl::sycl::ext::intel::experimental::bfloat16") align 2 %agg.tmp.i) #7
call void @llvm.lifetime.end.p0i8(i64 2, i8* nonnull %4)
%ref.tmp.ascast.i = addrspacecast %"class.cl::sycl::ext::intel::experimental::bfloat16"* %ref.tmp.i to %"class.cl::sycl::ext::intel::experimental::bfloat16" addrspace(4)*
%5 = bitcast %"class.cl::sycl::ext::intel::experimental::bfloat16"* %ref.tmp.i to i8*
%value.i.i.i = getelementptr inbounds %"class.cl::sycl::ext::intel::experimental::bfloat16", %"class.cl::sycl::ext::intel::experimental::bfloat16" addrspace(4)* %ref.tmp.ascast.i, i64 0, i32 0
%6 = bitcast %"class.cl::sycl::ext::intel::experimental::bfloat16"* %agg.tmp.i54 to i8*
%7 = getelementptr inbounds %"class.cl::sycl::ext::intel::experimental::bfloat16", %"class.cl::sycl::ext::intel::experimental::bfloat16"* %agg.tmp.i54, i64 0, i32 0
%8 = addrspacecast i16* %7 to i16 addrspace(4)*
br label %for.cond
for.cond: ; preds = %for.body, %entry
%sub_a.sroa.0.0 = phi %spirv.JointMatrixINTEL._bfloat16_8_16_0_3 addrspace(4)* [ %call.i, %entry ], [ %call.i58, %for.body ]
%i.0 = phi i32 [ 0, %entry ], [ %inc, %for.body ]
%conv = zext i32 %i.0 to i64
%call.i41 = call spir_func noundef i64 @_Z38__spirv_JointMatrixWorkItemLengthINTELIN2cl4sycl3ext5intel12experimental8bfloat16ELm8ELm16ELN5__spv12MatrixLayoutE0ELNS6_5Scope4FlagE3EEmPNS6_24__spirv_JointMatrixINTELIT_XT0_EXT1_EXT2_EXT3_EEE(%spirv.JointMatrixINTEL._bfloat16_8_16_0_3 addrspace(4)* noundef %sub_a.sroa.0.0) #7
%cmp = icmp ugt i64 %call.i41, %conv
br i1 %cmp, label %for.body, label %for.cond.cleanup
for.cond.cleanup: ; preds = %for.cond
%sub5 = sub nsw i64 %1, %3
%sub = sub nsw i64 %0, %2
%MData.i.i.i = getelementptr inbounds %class.anon, %class.anon addrspace(4)* %this, i64 0, i32 0, i32 1, i32 0
%9 = load %"class.cl::sycl::ext::intel::experimental::bfloat16" addrspace(1)*, %"class.cl::sycl::ext::intel::experimental::bfloat16" addrspace(1)* addrspace(4)* %MData.i.i.i, align 8, !tbaa !12, !noalias !13
%mul19 = shl nsw i64 %sub, 7
%add.ptr.i = getelementptr inbounds %"class.cl::sycl::ext::intel::experimental::bfloat16", %"class.cl::sycl::ext::intel::experimental::bfloat16" addrspace(1)* %9, i64 %mul19
%div = and i64 %sub5, -8
%add.ptr.i45 = getelementptr inbounds %"class.cl::sycl::ext::intel::experimental::bfloat16", %"class.cl::sycl::ext::intel::experimental::bfloat16" addrspace(1)* %add.ptr.i, i64 %div
%call.ascast.i = addrspacecast %"class.cl::sycl::ext::intel::experimental::bfloat16" addrspace(1)* %add.ptr.i45 to %"class.cl::sycl::ext::intel::experimental::bfloat16" addrspace(4)*
call spir_func void @_Z29__spirv_JointMatrixStoreINTELIN2cl4sycl3ext5intel12experimental8bfloat16ELm8ELm16ELN5__spv12MatrixLayoutE0ELNS6_5Scope4FlagE3EEvPT_PNS6_24__spirv_JointMatrixINTELISA_XT0_EXT1_EXT2_EXT3_EEEmS7_S9_i(%"class.cl::sycl::ext::intel::experimental::bfloat16" addrspace(4)* noundef %call.ascast.i, %spirv.JointMatrixINTEL._bfloat16_8_16_0_3 addrspace(4)* noundef %sub_a.sroa.0.0, i64 noundef 16, i32 noundef 0, i32 noundef 3, i32 noundef 0) #7
ret void
for.body: ; preds = %for.cond
%call.i.i = call spir_func noundef zeroext i16 @_Z27__spirv_ConvertFToBF16INTELf(float noundef 2.000000e+00) #6
call void @llvm.lifetime.start.p0i8(i64 2, i8* nonnull %5) #8, !noalias !16
call spir_func void @_Z28__spirv_VectorExtractDynamicIN2cl4sycl3ext5intel12experimental8bfloat16ELm8ELm16ELN5__spv12MatrixLayoutE0ELNS6_5Scope4FlagE3EET_PNS6_24__spirv_JointMatrixINTELISA_XT0_EXT1_EXT2_EXT3_EEEm(%"class.cl::sycl::ext::intel::experimental::bfloat16" addrspace(4)* sret(%"class.cl::sycl::ext::intel::experimental::bfloat16") align 2 %ref.tmp.ascast.i, %spirv.JointMatrixINTEL._bfloat16_8_16_0_3 addrspace(4)* noundef %sub_a.sroa.0.0, i64 noundef %conv) #7, !noalias !16
%10 = load i16, i16 addrspace(4)* %value.i.i.i, align 2, !tbaa !19, !noalias !20
%call.i.i.i.i = call spir_func noundef float @_Z27__spirv_ConvertBF16ToFINTELt(i16 noundef zeroext %10) #6, !noalias !20
%call.i.i3.i.i = call spir_func noundef float @_Z27__spirv_ConvertBF16ToFINTELt(i16 noundef zeroext %call.i.i) #6, !noalias !20
%add.i.i = fadd float %call.i.i.i.i, %call.i.i3.i.i
%call.i.i4.i.i = call spir_func noundef zeroext i16 @_Z27__spirv_ConvertFToBF16INTELf(float noundef %add.i.i) #6, !noalias !20
call void @llvm.lifetime.end.p0i8(i64 2, i8* nonnull %5) #8, !noalias !16
call void @llvm.lifetime.start.p0i8(i64 2, i8* nonnull %6)
store i16 %call.i.i4.i.i, i16 addrspace(4)* %8, align 2, !tbaa !19
%call.i58 = call spir_func noundef %spirv.JointMatrixINTEL._bfloat16_8_16_0_3 addrspace(4)* @_Z27__spirv_VectorInsertDynamicIN2cl4sycl3ext5intel12experimental8bfloat16ELm8ELm16ELN5__spv12MatrixLayoutE0ELNS6_5Scope4FlagE3EEPNS6_24__spirv_JointMatrixINTELIT_XT0_EXT1_EXT2_EXT3_EEESD_SB_m(%spirv.JointMatrixINTEL._bfloat16_8_16_0_3 addrspace(4)* noundef %sub_a.sroa.0.0, %"class.cl::sycl::ext::intel::experimental::bfloat16"* noundef nonnull byval(%"class.cl::sycl::ext::intel::experimental::bfloat16") align 2 %agg.tmp.i54, i64 noundef %conv) #7
call void @llvm.lifetime.end.p0i8(i64 2, i8* nonnull %6)
%inc = add nuw nsw i32 %i.0, 1
br label %for.cond, !llvm.loop !23
}
; Function Attrs: argmemonly nofree nosync nounwind willreturn
declare void @llvm.lifetime.end.p0i8(i64 immarg, i8* nocapture) #0
; Function Attrs: inaccessiblememonly nofree nosync nounwind willreturn
declare void @llvm.assume(i1 noundef) #3
; Function Attrs: convergent
declare dso_local spir_func noundef %spirv.JointMatrixINTEL._bfloat16_8_16_0_3 addrspace(4)* @_Z26__spirv_CompositeConstructIN2cl4sycl3ext5intel12experimental8bfloat16ELm8ELm16ELN5__spv12MatrixLayoutE0ELNS6_5Scope4FlagE3EEPNS6_24__spirv_JointMatrixINTELIT_XT0_EXT1_EXT2_EXT3_EEESB_(%"class.cl::sycl::ext::intel::experimental::bfloat16"* noundef byval(%"class.cl::sycl::ext::intel::experimental::bfloat16") align 2) local_unnamed_addr #4
; Function Attrs: convergent
declare dso_local spir_func noundef i64 @_Z38__spirv_JointMatrixWorkItemLengthINTELIN2cl4sycl3ext5intel12experimental8bfloat16ELm8ELm16ELN5__spv12MatrixLayoutE0ELNS6_5Scope4FlagE3EEmPNS6_24__spirv_JointMatrixINTELIT_XT0_EXT1_EXT2_EXT3_EEE(%spirv.JointMatrixINTEL._bfloat16_8_16_0_3 addrspace(4)* noundef) local_unnamed_addr #4
; Function Attrs: convergent
declare dso_local spir_func void @_Z28__spirv_VectorExtractDynamicIN2cl4sycl3ext5intel12experimental8bfloat16ELm8ELm16ELN5__spv12MatrixLayoutE0ELNS6_5Scope4FlagE3EET_PNS6_24__spirv_JointMatrixINTELISA_XT0_EXT1_EXT2_EXT3_EEEm(%"class.cl::sycl::ext::intel::experimental::bfloat16" addrspace(4)* sret(%"class.cl::sycl::ext::intel::experimental::bfloat16") align 2, %spirv.JointMatrixINTEL._bfloat16_8_16_0_3 addrspace(4)* noundef, i64 noundef) local_unnamed_addr #4
; Function Attrs: convergent nounwind
declare dso_local spir_func noundef float @_Z27__spirv_ConvertBF16ToFINTELt(i16 noundef zeroext) local_unnamed_addr #5
; Function Attrs: convergent nounwind
declare dso_local spir_func noundef zeroext i16 @_Z27__spirv_ConvertFToBF16INTELf(float noundef) local_unnamed_addr #5
; Function Attrs: convergent
declare dso_local spir_func noundef %spirv.JointMatrixINTEL._bfloat16_8_16_0_3 addrspace(4)* @_Z27__spirv_VectorInsertDynamicIN2cl4sycl3ext5intel12experimental8bfloat16ELm8ELm16ELN5__spv12MatrixLayoutE0ELNS6_5Scope4FlagE3EEPNS6_24__spirv_JointMatrixINTELIT_XT0_EXT1_EXT2_EXT3_EEESD_SB_m(%spirv.JointMatrixINTEL._bfloat16_8_16_0_3 addrspace(4)* noundef, %"class.cl::sycl::ext::intel::experimental::bfloat16"* noundef byval(%"class.cl::sycl::ext::intel::experimental::bfloat16") align 2, i64 noundef) local_unnamed_addr #4
; Function Attrs: convergent
declare dso_local spir_func void @_Z29__spirv_JointMatrixStoreINTELIN2cl4sycl3ext5intel12experimental8bfloat16ELm8ELm16ELN5__spv12MatrixLayoutE0ELNS6_5Scope4FlagE3EEvPT_PNS6_24__spirv_JointMatrixINTELISA_XT0_EXT1_EXT2_EXT3_EEEmS7_S9_i(%"class.cl::sycl::ext::intel::experimental::bfloat16" addrspace(4)* noundef, %spirv.JointMatrixINTEL._bfloat16_8_16_0_3 addrspace(4)* noundef, i64 noundef, i32 noundef, i32 noundef, i32 noundef) local_unnamed_addr #4
attributes #0 = { argmemonly nofree nosync nounwind willreturn }
attributes #1 = { convergent inlinehint norecurse "frame-pointer"="all" "min-legal-vector-width"="0" "no-trapping-math"="true" "stack-protector-buffer-size"="8" }
attributes #2 = { argmemonly nofree nounwind willreturn writeonly }
attributes #3 = { inaccessiblememonly nofree nosync nounwind willreturn }
attributes #4 = { convergent "frame-pointer"="all" "no-trapping-math"="true" "stack-protector-buffer-size"="8" }
attributes #5 = { convergent nounwind "frame-pointer"="all" "no-trapping-math"="true" "stack-protector-buffer-size"="8" }
attributes #6 = { convergent nounwind }
attributes #7 = { convergent }
attributes #8 = { nounwind }
!llvm.module.flags = !{!0, !1}
!opencl.spir.version = !{!2}
!spirv.Source = !{!3}
!llvm.ident = !{!4}
!0 = !{i32 1, !"wchar_size", i32 4}
!1 = !{i32 7, !"frame-pointer", i32 2}
!2 = !{i32 1, i32 2}
!3 = !{i32 4, i32 100000}
!4 = !{!"clang version 15.0.0 (https://github.com/pauzinl/llvm.git fb27c655023f19ff91f09413a0c51f0a37071cff)"}
!5 = !{!6, !6, i64 0}
!6 = !{!"long", !7, i64 0}
!7 = !{!"omnipotent char", !8, i64 0}
!8 = !{!"Simple C++ TBAA"}
!9 = !{!10, !11, i64 0}
!10 = !{!"_ZTSN2cl4sycl3ext5intel12experimental8bfloat16E", !11, i64 0}
!11 = !{!"short", !7, i64 0}
!12 = !{!7, !7, i64 0}
!13 = !{!14}
!14 = distinct !{!14, !15, !"_ZNK2cl4sycl8accessorINS0_3ext5intel12experimental8bfloat16ELi2ELNS0_6access4modeE1026ELNS6_6targetE2014ELNS6_11placeholderE0ENS2_6oneapi22accessor_property_listIJEEEE11get_pointerILS8_2014EvEENS0_9multi_ptrIS5_LNS6_13address_spaceE1EEEv: %agg.result"}
!15 = distinct !{!15, !"_ZNK2cl4sycl8accessorINS0_3ext5intel12experimental8bfloat16ELi2ELNS0_6access4modeE1026ELNS6_6targetE2014ELNS6_11placeholderE0ENS2_6oneapi22accessor_property_listIJEEEE11get_pointerILS8_2014EvEENS0_9multi_ptrIS5_LNS6_13address_spaceE1EEEv"}
!16 = !{!17}
!17 = distinct !{!17, !18, !"_ZN2cl4sycl3ext6oneapi12experimental6matrixplERKNS4_10wi_elementINS1_5intel12experimental8bfloat16ELm8ELm16ELNS4_13matrix_layoutE0ENS2_9sub_groupEEERKS8_: %agg.result"}
!18 = distinct !{!18, !"_ZN2cl4sycl3ext6oneapi12experimental6matrixplERKNS4_10wi_elementINS1_5intel12experimental8bfloat16ELm8ELm16ELNS4_13matrix_layoutE0ENS2_9sub_groupEEERKS8_"}
!19 = !{!11, !11, i64 0}
!20 = !{!21, !17}
!21 = distinct !{!21, !22, !"_ZN2cl4sycl3ext5intel12experimentalplERKNS3_8bfloat16ES6_: %agg.result"}
!22 = distinct !{!22, !"_ZN2cl4sycl3ext5intel12experimentalplERKNS3_8bfloat16ES6_"}
!23 = distinct !{!23, !24}
!24 = !{!"llvm.loop.mustprogress"}
|
