1 ; Test the MSA intrinsics that are encoded with the 3RF instruction format and
2 ; use the result as a third operand and perform fixed-point operations.
4 ; RUN: llc -march=mips -mattr=+msa,+fp64,+mips32r2 < %s | FileCheck %s
5 ; RUN: llc -march=mipsel -mattr=+msa,+fp64,+mips32r2 < %s | FileCheck %s
7 @llvm_mips_madd_q_h_ARG1 = global <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>, align 16
8 @llvm_mips_madd_q_h_ARG2 = global <8 x i16> <i16 8, i16 9, i16 10, i16 11, i16 12, i16 13, i16 14, i16 15>, align 16
9 @llvm_mips_madd_q_h_ARG3 = global <8 x i16> <i16 16, i16 17, i16 18, i16 19, i16 20, i16 21, i16 22, i16 23>, align 16
10 @llvm_mips_madd_q_h_RES = global <8 x i16> <i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0>, align 16
12 define void @llvm_mips_madd_q_h_test() nounwind {
14 %0 = load <8 x i16>, ptr @llvm_mips_madd_q_h_ARG1
15 %1 = load <8 x i16>, ptr @llvm_mips_madd_q_h_ARG2
16 %2 = load <8 x i16>, ptr @llvm_mips_madd_q_h_ARG3
17 %3 = tail call <8 x i16> @llvm.mips.madd.q.h(<8 x i16> %0, <8 x i16> %1, <8 x i16> %2)
18 store <8 x i16> %3, ptr @llvm_mips_madd_q_h_RES
22 declare <8 x i16> @llvm.mips.madd.q.h(<8 x i16>, <8 x i16>, <8 x i16>) nounwind
24 ; CHECK: llvm_mips_madd_q_h_test:
30 ; CHECK: .size llvm_mips_madd_q_h_test
32 @llvm_mips_madd_q_w_ARG1 = global <4 x i32> <i32 0, i32 1, i32 2, i32 3>, align 16
33 @llvm_mips_madd_q_w_ARG2 = global <4 x i32> <i32 4, i32 5, i32 6, i32 7>, align 16
34 @llvm_mips_madd_q_w_ARG3 = global <4 x i32> <i32 8, i32 9, i32 10, i32 11>, align 16
35 @llvm_mips_madd_q_w_RES = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
37 define void @llvm_mips_madd_q_w_test() nounwind {
39 %0 = load <4 x i32>, ptr @llvm_mips_madd_q_w_ARG1
40 %1 = load <4 x i32>, ptr @llvm_mips_madd_q_w_ARG2
41 %2 = load <4 x i32>, ptr @llvm_mips_madd_q_w_ARG3
42 %3 = tail call <4 x i32> @llvm.mips.madd.q.w(<4 x i32> %0, <4 x i32> %1, <4 x i32> %2)
43 store <4 x i32> %3, ptr @llvm_mips_madd_q_w_RES
47 declare <4 x i32> @llvm.mips.madd.q.w(<4 x i32>, <4 x i32>, <4 x i32>) nounwind
49 ; CHECK: llvm_mips_madd_q_w_test:
55 ; CHECK: .size llvm_mips_madd_q_w_test
57 @llvm_mips_maddr_q_h_ARG1 = global <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>, align 16
58 @llvm_mips_maddr_q_h_ARG2 = global <8 x i16> <i16 8, i16 9, i16 10, i16 11, i16 12, i16 13, i16 14, i16 15>, align 16
59 @llvm_mips_maddr_q_h_ARG3 = global <8 x i16> <i16 16, i16 17, i16 18, i16 19, i16 20, i16 21, i16 22, i16 23>, align 16
60 @llvm_mips_maddr_q_h_RES = global <8 x i16> <i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0>, align 16
62 define void @llvm_mips_maddr_q_h_test() nounwind {
64 %0 = load <8 x i16>, ptr @llvm_mips_maddr_q_h_ARG1
65 %1 = load <8 x i16>, ptr @llvm_mips_maddr_q_h_ARG2
66 %2 = load <8 x i16>, ptr @llvm_mips_maddr_q_h_ARG3
67 %3 = tail call <8 x i16> @llvm.mips.maddr.q.h(<8 x i16> %0, <8 x i16> %1, <8 x i16> %2)
68 store <8 x i16> %3, ptr @llvm_mips_maddr_q_h_RES
72 declare <8 x i16> @llvm.mips.maddr.q.h(<8 x i16>, <8 x i16>, <8 x i16>) nounwind
74 ; CHECK: llvm_mips_maddr_q_h_test:
80 ; CHECK: .size llvm_mips_maddr_q_h_test
82 @llvm_mips_maddr_q_w_ARG1 = global <4 x i32> <i32 0, i32 1, i32 2, i32 3>, align 16
83 @llvm_mips_maddr_q_w_ARG2 = global <4 x i32> <i32 4, i32 5, i32 6, i32 7>, align 16
84 @llvm_mips_maddr_q_w_ARG3 = global <4 x i32> <i32 8, i32 9, i32 10, i32 11>, align 16
85 @llvm_mips_maddr_q_w_RES = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
87 define void @llvm_mips_maddr_q_w_test() nounwind {
89 %0 = load <4 x i32>, ptr @llvm_mips_maddr_q_w_ARG1
90 %1 = load <4 x i32>, ptr @llvm_mips_maddr_q_w_ARG2
91 %2 = load <4 x i32>, ptr @llvm_mips_maddr_q_w_ARG3
92 %3 = tail call <4 x i32> @llvm.mips.maddr.q.w(<4 x i32> %0, <4 x i32> %1, <4 x i32> %2)
93 store <4 x i32> %3, ptr @llvm_mips_maddr_q_w_RES
97 declare <4 x i32> @llvm.mips.maddr.q.w(<4 x i32>, <4 x i32>, <4 x i32>) nounwind
99 ; CHECK: llvm_mips_maddr_q_w_test:
105 ; CHECK: .size llvm_mips_maddr_q_w_test
107 @llvm_mips_msub_q_h_ARG1 = global <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>, align 16
108 @llvm_mips_msub_q_h_ARG2 = global <8 x i16> <i16 8, i16 9, i16 10, i16 11, i16 12, i16 13, i16 14, i16 15>, align 16
109 @llvm_mips_msub_q_h_ARG3 = global <8 x i16> <i16 16, i16 17, i16 18, i16 19, i16 20, i16 21, i16 22, i16 23>, align 16
110 @llvm_mips_msub_q_h_RES = global <8 x i16> <i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0>, align 16
112 define void @llvm_mips_msub_q_h_test() nounwind {
114 %0 = load <8 x i16>, ptr @llvm_mips_msub_q_h_ARG1
115 %1 = load <8 x i16>, ptr @llvm_mips_msub_q_h_ARG2
116 %2 = load <8 x i16>, ptr @llvm_mips_msub_q_h_ARG3
117 %3 = tail call <8 x i16> @llvm.mips.msub.q.h(<8 x i16> %0, <8 x i16> %1, <8 x i16> %2)
118 store <8 x i16> %3, ptr @llvm_mips_msub_q_h_RES
122 declare <8 x i16> @llvm.mips.msub.q.h(<8 x i16>, <8 x i16>, <8 x i16>) nounwind
124 ; CHECK: llvm_mips_msub_q_h_test:
130 ; CHECK: .size llvm_mips_msub_q_h_test
132 @llvm_mips_msub_q_w_ARG1 = global <4 x i32> <i32 0, i32 1, i32 2, i32 3>, align 16
133 @llvm_mips_msub_q_w_ARG2 = global <4 x i32> <i32 4, i32 5, i32 6, i32 7>, align 16
134 @llvm_mips_msub_q_w_ARG3 = global <4 x i32> <i32 8, i32 9, i32 10, i32 11>, align 16
135 @llvm_mips_msub_q_w_RES = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
137 define void @llvm_mips_msub_q_w_test() nounwind {
139 %0 = load <4 x i32>, ptr @llvm_mips_msub_q_w_ARG1
140 %1 = load <4 x i32>, ptr @llvm_mips_msub_q_w_ARG2
141 %2 = load <4 x i32>, ptr @llvm_mips_msub_q_w_ARG3
142 %3 = tail call <4 x i32> @llvm.mips.msub.q.w(<4 x i32> %0, <4 x i32> %1, <4 x i32> %2)
143 store <4 x i32> %3, ptr @llvm_mips_msub_q_w_RES
147 declare <4 x i32> @llvm.mips.msub.q.w(<4 x i32>, <4 x i32>, <4 x i32>) nounwind
149 ; CHECK: llvm_mips_msub_q_w_test:
155 ; CHECK: .size llvm_mips_msub_q_w_test
157 @llvm_mips_msubr_q_h_ARG1 = global <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>, align 16
158 @llvm_mips_msubr_q_h_ARG2 = global <8 x i16> <i16 8, i16 9, i16 10, i16 11, i16 12, i16 13, i16 14, i16 15>, align 16
159 @llvm_mips_msubr_q_h_ARG3 = global <8 x i16> <i16 16, i16 17, i16 18, i16 19, i16 20, i16 21, i16 22, i16 23>, align 16
160 @llvm_mips_msubr_q_h_RES = global <8 x i16> <i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0>, align 16
162 define void @llvm_mips_msubr_q_h_test() nounwind {
164 %0 = load <8 x i16>, ptr @llvm_mips_msubr_q_h_ARG1
165 %1 = load <8 x i16>, ptr @llvm_mips_msubr_q_h_ARG2
166 %2 = load <8 x i16>, ptr @llvm_mips_msubr_q_h_ARG3
167 %3 = tail call <8 x i16> @llvm.mips.msubr.q.h(<8 x i16> %0, <8 x i16> %1, <8 x i16> %2)
168 store <8 x i16> %3, ptr @llvm_mips_msubr_q_h_RES
172 declare <8 x i16> @llvm.mips.msubr.q.h(<8 x i16>, <8 x i16>, <8 x i16>) nounwind
174 ; CHECK: llvm_mips_msubr_q_h_test:
180 ; CHECK: .size llvm_mips_msubr_q_h_test
182 @llvm_mips_msubr_q_w_ARG1 = global <4 x i32> <i32 0, i32 1, i32 2, i32 3>, align 16
183 @llvm_mips_msubr_q_w_ARG2 = global <4 x i32> <i32 4, i32 5, i32 6, i32 7>, align 16
184 @llvm_mips_msubr_q_w_ARG3 = global <4 x i32> <i32 8, i32 9, i32 10, i32 11>, align 16
185 @llvm_mips_msubr_q_w_RES = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
187 define void @llvm_mips_msubr_q_w_test() nounwind {
189 %0 = load <4 x i32>, ptr @llvm_mips_msubr_q_w_ARG1
190 %1 = load <4 x i32>, ptr @llvm_mips_msubr_q_w_ARG2
191 %2 = load <4 x i32>, ptr @llvm_mips_msubr_q_w_ARG3
192 %3 = tail call <4 x i32> @llvm.mips.msubr.q.w(<4 x i32> %0, <4 x i32> %1, <4 x i32> %2)
193 store <4 x i32> %3, ptr @llvm_mips_msubr_q_w_RES
197 declare <4 x i32> @llvm.mips.msubr.q.w(<4 x i32>, <4 x i32>, <4 x i32>) nounwind
199 ; CHECK: llvm_mips_msubr_q_w_test:
205 ; CHECK: .size llvm_mips_msubr_q_w_test