Lines Matching full:matrix
1 ; RUN: opt -p lower-matrix-intrinsics -matrix-print-after-transpose-opt -disable-output -S %s 2>&1 | FileCheck %s
12 ; CHECK-NEXT: [[T:%.+]] = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> [[A]], i32 3, i32 2)
16 %a.t = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> %a, i32 3, i32 2)
17 %b.t = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> %b, i32 3, i32 2)
28 ; CHECK-NEXT: [[T:%.+]] = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> [[ADD]], i32 3, i32 2)
34 %a.t = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> %a, i32 3, i32 2)
35 %b.t = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> %b, i32 3, i32 2)
44 ; CHECK-NEXT: [[T:%.+]] = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> [[A]], i32 1, i32 6)
48 %a.t = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> %a, i32 1, i32 6)
49 %b.t = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> %b, i32 3, i32 2)
60 ; CHECK-NEXT: [[T:%.+]] = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> [[ADD]], i32 1, i32 6)
66 %a.t = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> %a, i32 1, i32 6)
67 %b.t = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> %b, i32 3, i32 2)
76 ; CHECK-NEXT: [[T:%.+]] = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> [[A]], i32 3, i32 2)
81 %a.t = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> %a, i32 3, i32 2)
82 %b.t = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> %b, i32 6, i32 1)
93 ; CHECK-NEXT: [[T:%.+]] = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> [[ADD]], i32 3, i32 2)
99 %a.t = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> %a, i32 3, i32 2)
100 %b.t = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> %b, i32 6, i32 1)
108 ; CHECK-NEXT: [[MUL:%.+]] = call <9 x double> @llvm.matrix.multiply.v9f64.v6f64.v6f64(<6 x double> %b, <6 x double> %a, i32 3, i32 2, i32 3)
109 ; CHECK-NEXT: [[T:%.+]] = call <9 x double> @llvm.matrix.transpose.v9f64(<9 x double> [[MUL]], i32 3, i32 3)
113 %a.t = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> %a, i32 3, i32 2)
114 %b.t = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> %b, i32 2, i32 3)
115 %mul = call <9 x double> @llvm.matrix.multiply.v9f64.v6f64(<6 x double> %a.t, <6 x double> %b.t, i32 3, i32 2 , i32 3)
122 ; CHECK-NEXT: [[MUL:%.+]] = call <6 x double> @llvm.matrix.multiply.v6f64.v4f64.v6f64(<4 x double> %b, <6 x double> %a, i32 2, i32 2, i32 3)
123 ; CHECK-NEXT: [[T:%.+]] = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> [[MUL]], i32 2, i32 3)
127 %a.t = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> %a, i32 3, i32 2)
128 %b.t = call <4 x double> @llvm.matrix.transpose.v4f64(<4 x double> %b, i32 2, i32 2)
129 %mul = call <6 x double> @llvm.matrix.multiply.v6f64.v6f64.v4f64(<6 x double> %a.t, <4 x double> %b.t, i32 3, i32 2 , i32 2)
136 ; CHECK-NEXT: [[MUL:%.+]] = call <6 x double> @llvm.matrix.multiply.v6f64.v6f64.v4f64(<6 x double> %b, <4 x double> %a, i32 3, i32 2, i32 2)
137 ; CHECK-NEXT: [[T:%.+]] = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> [[MUL]], i32 3, i32 2)
141 %a.t = call <4 x double> @llvm.matrix.transpose.v4f64(<4 x double> %a, i32 2, i32 2)
142 %b.t = call <6 x double> @llvm.matrix.transpose.v6f64(<6 x double> %b, i32 2, i32 3)
143 %mul = call <6 x double> @llvm.matrix.multiply.v6f64.v4f64.v6f64(<4 x double> %a.t, <6 x double> %b.t, i32 2, i32 2 , i32 3)
151 ; CHECK-NEXT: [[TMP0:%.*]] = call <6 x float> @llvm.matrix.multiply.v6f32.v3f32.v2f32(<3 x float> [[A]], <2 x float> [[B]], i32 3, i32 1, i32 2)
153 ; CHECK-NEXT: [[MFADD_T:%.*]] = call <6 x float> @llvm.matrix.transpose.v6f32(<6 x float> [[MFADD]], i32 3, i32 2)
158 %m = tail call <6 x float> @llvm.matrix.multiply.v6f32.v3f32.v2f32(<3 x float> %a, <2 x float> %b, i32 3, i32 1, i32 2)
160 %t = tail call <6 x float> @llvm.matrix.transpose.v6f32(<6 x float> %add, i32 3, i32 2)
165 declare <6 x double> @llvm.matrix.transpose.v6f64.v6f64(<6 x double>, i32, i32)
166 declare <4 x double> @llvm.matrix.transpose.v4f64.v4f64(<4 x double>, i32, i32)
167 declare <9 x double> @llvm.matrix.multiply.v9f64.v6f64(<6 x double>, <6 x double>, i32, i32, i32)
168 declare <6 x double> @llvm.matrix.multiply.v6f64.v6f64.v4f64(<6 x double>, <4 x double>, i32, i32, i32)
169 declare <6 x double> @llvm.matrix.multiply.v6f64.v6f64.v6f64(<6 x double>, <4 x double>, i32, i32, i32)
170 declare <6 x float> @llvm.matrix.transpose.v6f32(<6 x float>, i32 immarg, i32 immarg)
171 declare <6 x float> @llvm.matrix.multiply.v6f32.v3f32.v2f32(<3 x float>, <2 x float>, i32 immarg, i32 immarg, i32 immarg)