Home | History | Annotate | Download | only in msa
      1 ; Test the MSA intrinsics that are encoded with the 3R instruction format.
      2 ; There are lots of these so this covers those beginning with 'd'
      3 
      4 ; RUN: llc -march=mips -mattr=+msa,+fp64 < %s | FileCheck %s
      5 ; RUN: llc -march=mipsel -mattr=+msa,+fp64 < %s | FileCheck %s
      6 
      7 @llvm_mips_div_s_b_ARG1 = global <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15>, align 16
      8 @llvm_mips_div_s_b_ARG2 = global <16 x i8> <i8 16, i8 17, i8 18, i8 19, i8 20, i8 21, i8 22, i8 23, i8 24, i8 25, i8 26, i8 27, i8 28, i8 29, i8 30, i8 31>, align 16
      9 @llvm_mips_div_s_b_RES  = global <16 x i8> <i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0>, align 16
     10 
     11 define void @llvm_mips_div_s_b_test() nounwind {
     12 entry:
     13   %0 = load <16 x i8>* @llvm_mips_div_s_b_ARG1
     14   %1 = load <16 x i8>* @llvm_mips_div_s_b_ARG2
     15   %2 = tail call <16 x i8> @llvm.mips.div.s.b(<16 x i8> %0, <16 x i8> %1)
     16   store <16 x i8> %2, <16 x i8>* @llvm_mips_div_s_b_RES
     17   ret void
     18 }
     19 
     20 declare <16 x i8> @llvm.mips.div.s.b(<16 x i8>, <16 x i8>) nounwind
     21 
     22 ; CHECK: llvm_mips_div_s_b_test:
     23 ; CHECK: ld.b
     24 ; CHECK: ld.b
     25 ; CHECK: div_s.b
     26 ; CHECK: st.b
     27 ; CHECK: .size llvm_mips_div_s_b_test
     28 ;
     29 @llvm_mips_div_s_h_ARG1 = global <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>, align 16
     30 @llvm_mips_div_s_h_ARG2 = global <8 x i16> <i16 8, i16 9, i16 10, i16 11, i16 12, i16 13, i16 14, i16 15>, align 16
     31 @llvm_mips_div_s_h_RES  = global <8 x i16> <i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0>, align 16
     32 
     33 define void @llvm_mips_div_s_h_test() nounwind {
     34 entry:
     35   %0 = load <8 x i16>* @llvm_mips_div_s_h_ARG1
     36   %1 = load <8 x i16>* @llvm_mips_div_s_h_ARG2
     37   %2 = tail call <8 x i16> @llvm.mips.div.s.h(<8 x i16> %0, <8 x i16> %1)
     38   store <8 x i16> %2, <8 x i16>* @llvm_mips_div_s_h_RES
     39   ret void
     40 }
     41 
     42 declare <8 x i16> @llvm.mips.div.s.h(<8 x i16>, <8 x i16>) nounwind
     43 
     44 ; CHECK: llvm_mips_div_s_h_test:
     45 ; CHECK: ld.h
     46 ; CHECK: ld.h
     47 ; CHECK: div_s.h
     48 ; CHECK: st.h
     49 ; CHECK: .size llvm_mips_div_s_h_test
     50 ;
     51 @llvm_mips_div_s_w_ARG1 = global <4 x i32> <i32 0, i32 1, i32 2, i32 3>, align 16
     52 @llvm_mips_div_s_w_ARG2 = global <4 x i32> <i32 4, i32 5, i32 6, i32 7>, align 16
     53 @llvm_mips_div_s_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
     54 
     55 define void @llvm_mips_div_s_w_test() nounwind {
     56 entry:
     57   %0 = load <4 x i32>* @llvm_mips_div_s_w_ARG1
     58   %1 = load <4 x i32>* @llvm_mips_div_s_w_ARG2
     59   %2 = tail call <4 x i32> @llvm.mips.div.s.w(<4 x i32> %0, <4 x i32> %1)
     60   store <4 x i32> %2, <4 x i32>* @llvm_mips_div_s_w_RES
     61   ret void
     62 }
     63 
     64 declare <4 x i32> @llvm.mips.div.s.w(<4 x i32>, <4 x i32>) nounwind
     65 
     66 ; CHECK: llvm_mips_div_s_w_test:
     67 ; CHECK: ld.w
     68 ; CHECK: ld.w
     69 ; CHECK: div_s.w
     70 ; CHECK: st.w
     71 ; CHECK: .size llvm_mips_div_s_w_test
     72 ;
     73 @llvm_mips_div_s_d_ARG1 = global <2 x i64> <i64 0, i64 1>, align 16
     74 @llvm_mips_div_s_d_ARG2 = global <2 x i64> <i64 2, i64 3>, align 16
     75 @llvm_mips_div_s_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
     76 
     77 define void @llvm_mips_div_s_d_test() nounwind {
     78 entry:
     79   %0 = load <2 x i64>* @llvm_mips_div_s_d_ARG1
     80   %1 = load <2 x i64>* @llvm_mips_div_s_d_ARG2
     81   %2 = tail call <2 x i64> @llvm.mips.div.s.d(<2 x i64> %0, <2 x i64> %1)
     82   store <2 x i64> %2, <2 x i64>* @llvm_mips_div_s_d_RES
     83   ret void
     84 }
     85 
     86 declare <2 x i64> @llvm.mips.div.s.d(<2 x i64>, <2 x i64>) nounwind
     87 
     88 ; CHECK: llvm_mips_div_s_d_test:
     89 ; CHECK: ld.d
     90 ; CHECK: ld.d
     91 ; CHECK: div_s.d
     92 ; CHECK: st.d
     93 ; CHECK: .size llvm_mips_div_s_d_test
     94 ;
     95 
     96 define void @div_s_b_test() nounwind {
     97 entry:
     98   %0 = load <16 x i8>* @llvm_mips_div_s_b_ARG1
     99   %1 = load <16 x i8>* @llvm_mips_div_s_b_ARG2
    100   %2 = sdiv <16 x i8> %0, %1
    101   store <16 x i8> %2, <16 x i8>* @llvm_mips_div_s_b_RES
    102   ret void
    103 }
    104 
    105 ; CHECK: div_s_b_test:
    106 ; CHECK: ld.b
    107 ; CHECK: ld.b
    108 ; CHECK: div_s.b
    109 ; CHECK: st.b
    110 ; CHECK: .size div_s_b_test
    111 
    112 define void @div_s_h_test() nounwind {
    113 entry:
    114   %0 = load <8 x i16>* @llvm_mips_div_s_h_ARG1
    115   %1 = load <8 x i16>* @llvm_mips_div_s_h_ARG2
    116   %2 = sdiv <8 x i16> %0, %1
    117   store <8 x i16> %2, <8 x i16>* @llvm_mips_div_s_h_RES
    118   ret void
    119 }
    120 
    121 ; CHECK: div_s_h_test:
    122 ; CHECK: ld.h
    123 ; CHECK: ld.h
    124 ; CHECK: div_s.h
    125 ; CHECK: st.h
    126 ; CHECK: .size div_s_h_test
    127 
    128 define void @div_s_w_test() nounwind {
    129 entry:
    130   %0 = load <4 x i32>* @llvm_mips_div_s_w_ARG1
    131   %1 = load <4 x i32>* @llvm_mips_div_s_w_ARG2
    132   %2 = sdiv <4 x i32> %0, %1
    133   store <4 x i32> %2, <4 x i32>* @llvm_mips_div_s_w_RES
    134   ret void
    135 }
    136 
    137 ; CHECK: div_s_w_test:
    138 ; CHECK: ld.w
    139 ; CHECK: ld.w
    140 ; CHECK: div_s.w
    141 ; CHECK: st.w
    142 ; CHECK: .size div_s_w_test
    143 
    144 define void @div_s_d_test() nounwind {
    145 entry:
    146   %0 = load <2 x i64>* @llvm_mips_div_s_d_ARG1
    147   %1 = load <2 x i64>* @llvm_mips_div_s_d_ARG2
    148   %2 = sdiv <2 x i64> %0, %1
    149   store <2 x i64> %2, <2 x i64>* @llvm_mips_div_s_d_RES
    150   ret void
    151 }
    152 
    153 ; CHECK: div_s_d_test:
    154 ; CHECK: ld.d
    155 ; CHECK: ld.d
    156 ; CHECK: div_s.d
    157 ; CHECK: st.d
    158 ; CHECK: .size div_s_d_test
    159 ;
    160 @llvm_mips_div_u_b_ARG1 = global <16 x i8> <i8 0, i8 1, i8 2, i8 3, i8 4, i8 5, i8 6, i8 7, i8 8, i8 9, i8 10, i8 11, i8 12, i8 13, i8 14, i8 15>, align 16
    161 @llvm_mips_div_u_b_ARG2 = global <16 x i8> <i8 16, i8 17, i8 18, i8 19, i8 20, i8 21, i8 22, i8 23, i8 24, i8 25, i8 26, i8 27, i8 28, i8 29, i8 30, i8 31>, align 16
    162 @llvm_mips_div_u_b_RES  = global <16 x i8> <i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0, i8 0>, align 16
    163 
    164 define void @llvm_mips_div_u_b_test() nounwind {
    165 entry:
    166   %0 = load <16 x i8>* @llvm_mips_div_u_b_ARG1
    167   %1 = load <16 x i8>* @llvm_mips_div_u_b_ARG2
    168   %2 = tail call <16 x i8> @llvm.mips.div.u.b(<16 x i8> %0, <16 x i8> %1)
    169   store <16 x i8> %2, <16 x i8>* @llvm_mips_div_u_b_RES
    170   ret void
    171 }
    172 
    173 declare <16 x i8> @llvm.mips.div.u.b(<16 x i8>, <16 x i8>) nounwind
    174 
    175 ; CHECK: llvm_mips_div_u_b_test:
    176 ; CHECK: ld.b
    177 ; CHECK: ld.b
    178 ; CHECK: div_u.b
    179 ; CHECK: st.b
    180 ; CHECK: .size llvm_mips_div_u_b_test
    181 ;
    182 @llvm_mips_div_u_h_ARG1 = global <8 x i16> <i16 0, i16 1, i16 2, i16 3, i16 4, i16 5, i16 6, i16 7>, align 16
    183 @llvm_mips_div_u_h_ARG2 = global <8 x i16> <i16 8, i16 9, i16 10, i16 11, i16 12, i16 13, i16 14, i16 15>, align 16
    184 @llvm_mips_div_u_h_RES  = global <8 x i16> <i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0, i16 0>, align 16
    185 
    186 define void @llvm_mips_div_u_h_test() nounwind {
    187 entry:
    188   %0 = load <8 x i16>* @llvm_mips_div_u_h_ARG1
    189   %1 = load <8 x i16>* @llvm_mips_div_u_h_ARG2
    190   %2 = tail call <8 x i16> @llvm.mips.div.u.h(<8 x i16> %0, <8 x i16> %1)
    191   store <8 x i16> %2, <8 x i16>* @llvm_mips_div_u_h_RES
    192   ret void
    193 }
    194 
    195 declare <8 x i16> @llvm.mips.div.u.h(<8 x i16>, <8 x i16>) nounwind
    196 
    197 ; CHECK: llvm_mips_div_u_h_test:
    198 ; CHECK: ld.h
    199 ; CHECK: ld.h
    200 ; CHECK: div_u.h
    201 ; CHECK: st.h
    202 ; CHECK: .size llvm_mips_div_u_h_test
    203 ;
    204 @llvm_mips_div_u_w_ARG1 = global <4 x i32> <i32 0, i32 1, i32 2, i32 3>, align 16
    205 @llvm_mips_div_u_w_ARG2 = global <4 x i32> <i32 4, i32 5, i32 6, i32 7>, align 16
    206 @llvm_mips_div_u_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>, align 16
    207 
    208 define void @llvm_mips_div_u_w_test() nounwind {
    209 entry:
    210   %0 = load <4 x i32>* @llvm_mips_div_u_w_ARG1
    211   %1 = load <4 x i32>* @llvm_mips_div_u_w_ARG2
    212   %2 = tail call <4 x i32> @llvm.mips.div.u.w(<4 x i32> %0, <4 x i32> %1)
    213   store <4 x i32> %2, <4 x i32>* @llvm_mips_div_u_w_RES
    214   ret void
    215 }
    216 
    217 declare <4 x i32> @llvm.mips.div.u.w(<4 x i32>, <4 x i32>) nounwind
    218 
    219 ; CHECK: llvm_mips_div_u_w_test:
    220 ; CHECK: ld.w
    221 ; CHECK: ld.w
    222 ; CHECK: div_u.w
    223 ; CHECK: st.w
    224 ; CHECK: .size llvm_mips_div_u_w_test
    225 ;
    226 @llvm_mips_div_u_d_ARG1 = global <2 x i64> <i64 0, i64 1>, align 16
    227 @llvm_mips_div_u_d_ARG2 = global <2 x i64> <i64 2, i64 3>, align 16
    228 @llvm_mips_div_u_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
    229 
    230 define void @llvm_mips_div_u_d_test() nounwind {
    231 entry:
    232   %0 = load <2 x i64>* @llvm_mips_div_u_d_ARG1
    233   %1 = load <2 x i64>* @llvm_mips_div_u_d_ARG2
    234   %2 = tail call <2 x i64> @llvm.mips.div.u.d(<2 x i64> %0, <2 x i64> %1)
    235   store <2 x i64> %2, <2 x i64>* @llvm_mips_div_u_d_RES
    236   ret void
    237 }
    238 
    239 declare <2 x i64> @llvm.mips.div.u.d(<2 x i64>, <2 x i64>) nounwind
    240 
    241 ; CHECK: llvm_mips_div_u_d_test:
    242 ; CHECK: ld.d
    243 ; CHECK: ld.d
    244 ; CHECK: div_u.d
    245 ; CHECK: st.d
    246 ; CHECK: .size llvm_mips_div_u_d_test
    247 ;
    248 
    249 define void @div_u_b_test() nounwind {
    250 entry:
    251   %0 = load <16 x i8>* @llvm_mips_div_u_b_ARG1
    252   %1 = load <16 x i8>* @llvm_mips_div_u_b_ARG2
    253   %2 = udiv <16 x i8> %0, %1
    254   store <16 x i8> %2, <16 x i8>* @llvm_mips_div_u_b_RES
    255   ret void
    256 }
    257 
    258 ; CHECK: div_u_b_test:
    259 ; CHECK: ld.b
    260 ; CHECK: ld.b
    261 ; CHECK: div_u.b
    262 ; CHECK: st.b
    263 ; CHECK: .size div_u_b_test
    264 
    265 define void @div_u_h_test() nounwind {
    266 entry:
    267   %0 = load <8 x i16>* @llvm_mips_div_u_h_ARG1
    268   %1 = load <8 x i16>* @llvm_mips_div_u_h_ARG2
    269   %2 = udiv <8 x i16> %0, %1
    270   store <8 x i16> %2, <8 x i16>* @llvm_mips_div_u_h_RES
    271   ret void
    272 }
    273 
    274 ; CHECK: div_u_h_test:
    275 ; CHECK: ld.h
    276 ; CHECK: ld.h
    277 ; CHECK: div_u.h
    278 ; CHECK: st.h
    279 ; CHECK: .size div_u_h_test
    280 
    281 define void @div_u_w_test() nounwind {
    282 entry:
    283   %0 = load <4 x i32>* @llvm_mips_div_u_w_ARG1
    284   %1 = load <4 x i32>* @llvm_mips_div_u_w_ARG2
    285   %2 = udiv <4 x i32> %0, %1
    286   store <4 x i32> %2, <4 x i32>* @llvm_mips_div_u_w_RES
    287   ret void
    288 }
    289 
    290 ; CHECK: div_u_w_test:
    291 ; CHECK: ld.w
    292 ; CHECK: ld.w
    293 ; CHECK: div_u.w
    294 ; CHECK: st.w
    295 ; CHECK: .size div_u_w_test
    296 
    297 define void @div_u_d_test() nounwind {
    298 entry:
    299   %0 = load <2 x i64>* @llvm_mips_div_u_d_ARG1
    300   %1 = load <2 x i64>* @llvm_mips_div_u_d_ARG2
    301   %2 = udiv <2 x i64> %0, %1
    302   store <2 x i64> %2, <2 x i64>* @llvm_mips_div_u_d_RES
    303   ret void
    304 }
    305 
    306 ; CHECK: div_u_d_test:
    307 ; CHECK: ld.d
    308 ; CHECK: ld.d
    309 ; CHECK: div_u.d
    310 ; CHECK: st.d
    311 ; CHECK: .size div_u_d_test
    312 ;
    313 @llvm_mips_dotp_s_h_ARG1 = global <16 x i8> <i8  0, i8  1, i8  2, i8  3,
    314                                              i8  4, i8  5, i8  6, i8  7,
    315                                              i8  8, i8  9, i8 10, i8 11,
    316                                              i8 12, i8 13, i8 14, i8 15>,
    317                                             align 16
    318 @llvm_mips_dotp_s_h_ARG2 = global <16 x i8> <i8 16, i8 17, i8 18, i8 19,
    319                                              i8 20, i8 21, i8 22, i8 23,
    320                                              i8 24, i8 25, i8 26, i8 27,
    321                                              i8 28, i8 29, i8 30, i8 31>,
    322                                             align 16
    323 @llvm_mips_dotp_s_h_RES  = global <8 x i16> <i16 0, i16 0, i16 0, i16 0,
    324                                              i16 0, i16 0, i16 0, i16 0>,
    325                                             align 16
    326 
    327 define void @llvm_mips_dotp_s_h_test() nounwind {
    328 entry:
    329   %0 = load <16 x i8>* @llvm_mips_dotp_s_h_ARG1
    330   %1 = load <16 x i8>* @llvm_mips_dotp_s_h_ARG2
    331   %2 = tail call <8 x i16> @llvm.mips.dotp.s.h(<16 x i8> %0, <16 x i8> %1)
    332   store <8 x i16> %2, <8 x i16>* @llvm_mips_dotp_s_h_RES
    333   ret void
    334 }
    335 
    336 declare <8 x i16> @llvm.mips.dotp.s.h(<16 x i8>, <16 x i8>) nounwind
    337 
    338 ; CHECK: llvm_mips_dotp_s_h_test:
    339 ; CHECK: ld.b
    340 ; CHECK: ld.b
    341 ; CHECK: dotp_s.h
    342 ; CHECK: st.h
    343 ; CHECK: .size llvm_mips_dotp_s_h_test
    344 ;
    345 @llvm_mips_dotp_s_w_ARG1 = global <8 x i16> <i16 0, i16 1, i16 2, i16 3,
    346                                              i16 4, i16 5, i16 6, i16 7>,
    347                                             align 16
    348 @llvm_mips_dotp_s_w_ARG2 = global <8 x i16> <i16  4, i16  5, i16  6, i16  7,
    349                                              i16  8, i16  9, i16 10, i16 11>,
    350                                             align 16
    351 @llvm_mips_dotp_s_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>,
    352                                             align 16
    353 
    354 define void @llvm_mips_dotp_s_w_test() nounwind {
    355 entry:
    356   %0 = load <8 x i16>* @llvm_mips_dotp_s_w_ARG1
    357   %1 = load <8 x i16>* @llvm_mips_dotp_s_w_ARG2
    358   %2 = tail call <4 x i32> @llvm.mips.dotp.s.w(<8 x i16> %0, <8 x i16> %1)
    359   store <4 x i32> %2, <4 x i32>* @llvm_mips_dotp_s_w_RES
    360   ret void
    361 }
    362 
    363 declare <4 x i32> @llvm.mips.dotp.s.w(<8 x i16>, <8 x i16>) nounwind
    364 
    365 ; CHECK: llvm_mips_dotp_s_w_test:
    366 ; CHECK: ld.h
    367 ; CHECK: ld.h
    368 ; CHECK: dotp_s.w
    369 ; CHECK: st.w
    370 ; CHECK: .size llvm_mips_dotp_s_w_test
    371 ;
    372 @llvm_mips_dotp_s_d_ARG1 = global <4 x i32> <i32 0, i32 1, i32 0, i32 1>,
    373                                             align 16
    374 @llvm_mips_dotp_s_d_ARG2 = global <4 x i32> <i32 2, i32 3, i32 2, i32 3>,
    375                                             align 16
    376 @llvm_mips_dotp_s_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
    377 
    378 define void @llvm_mips_dotp_s_d_test() nounwind {
    379 entry:
    380   %0 = load <4 x i32>* @llvm_mips_dotp_s_d_ARG1
    381   %1 = load <4 x i32>* @llvm_mips_dotp_s_d_ARG2
    382   %2 = tail call <2 x i64> @llvm.mips.dotp.s.d(<4 x i32> %0, <4 x i32> %1)
    383   store <2 x i64> %2, <2 x i64>* @llvm_mips_dotp_s_d_RES
    384   ret void
    385 }
    386 
    387 declare <2 x i64> @llvm.mips.dotp.s.d(<4 x i32>, <4 x i32>) nounwind
    388 
    389 ; CHECK: llvm_mips_dotp_s_d_test:
    390 ; CHECK: ld.w
    391 ; CHECK: ld.w
    392 ; CHECK: dotp_s.d
    393 ; CHECK: st.d
    394 ; CHECK: .size llvm_mips_dotp_s_d_test
    395 ;
    396 @llvm_mips_dotp_u_h_ARG1 = global <16 x i8> <i8  0, i8  1, i8  2, i8  3,
    397                                              i8  4, i8  5, i8  6, i8  7,
    398                                              i8  8, i8  9, i8 10, i8 11,
    399                                              i8 12, i8 13, i8 14, i8 15>,
    400                                             align 16
    401 @llvm_mips_dotp_u_h_ARG2 = global <16 x i8> <i8 16, i8 17, i8 18, i8 19,
    402                                              i8 20, i8 21, i8 22, i8 23,
    403                                              i8 24, i8 25, i8 26, i8 27,
    404                                              i8 28, i8 29, i8 30, i8 31>,
    405                                             align 16
    406 @llvm_mips_dotp_u_h_RES  = global <8 x i16> <i16 0, i16 0, i16 0, i16 0,
    407                                              i16 0, i16 0, i16 0, i16 0>,
    408                                             align 16
    409 
    410 define void @llvm_mips_dotp_u_h_test() nounwind {
    411 entry:
    412   %0 = load <16 x i8>* @llvm_mips_dotp_u_h_ARG1
    413   %1 = load <16 x i8>* @llvm_mips_dotp_u_h_ARG2
    414   %2 = tail call <8 x i16> @llvm.mips.dotp.u.h(<16 x i8> %0, <16 x i8> %1)
    415   store <8 x i16> %2, <8 x i16>* @llvm_mips_dotp_u_h_RES
    416   ret void
    417 }
    418 
    419 declare <8 x i16> @llvm.mips.dotp.u.h(<16 x i8>, <16 x i8>) nounwind
    420 
    421 ; CHECK: llvm_mips_dotp_u_h_test:
    422 ; CHECK: ld.b
    423 ; CHECK: ld.b
    424 ; CHECK: dotp_u.h
    425 ; CHECK: st.h
    426 ; CHECK: .size llvm_mips_dotp_u_h_test
    427 ;
    428 @llvm_mips_dotp_u_w_ARG1 = global <8 x i16> <i16 0, i16 1, i16 2, i16 3,
    429                                              i16 4, i16 5, i16 6, i16 7>,
    430                                             align 16
    431 @llvm_mips_dotp_u_w_ARG2 = global <8 x i16> <i16  4, i16  5, i16  6, i16  7,
    432                                              i16  8, i16  9, i16 10, i16 11>,
    433                                             align 16
    434 @llvm_mips_dotp_u_w_RES  = global <4 x i32> <i32 0, i32 0, i32 0, i32 0>,
    435                                             align 16
    436 
    437 define void @llvm_mips_dotp_u_w_test() nounwind {
    438 entry:
    439   %0 = load <8 x i16>* @llvm_mips_dotp_u_w_ARG1
    440   %1 = load <8 x i16>* @llvm_mips_dotp_u_w_ARG2
    441   %2 = tail call <4 x i32> @llvm.mips.dotp.u.w(<8 x i16> %0, <8 x i16> %1)
    442   store <4 x i32> %2, <4 x i32>* @llvm_mips_dotp_u_w_RES
    443   ret void
    444 }
    445 
    446 declare <4 x i32> @llvm.mips.dotp.u.w(<8 x i16>, <8 x i16>) nounwind
    447 
    448 ; CHECK: llvm_mips_dotp_u_w_test:
    449 ; CHECK: ld.h
    450 ; CHECK: ld.h
    451 ; CHECK: dotp_u.w
    452 ; CHECK: st.w
    453 ; CHECK: .size llvm_mips_dotp_u_w_test
    454 ;
    455 @llvm_mips_dotp_u_d_ARG1 = global <4 x i32> <i32 0, i32 1, i32 0, i32 1>,
    456                                             align 16
    457 @llvm_mips_dotp_u_d_ARG2 = global <4 x i32> <i32 2, i32 3, i32 2, i32 3>,
    458                                             align 16
    459 @llvm_mips_dotp_u_d_RES  = global <2 x i64> <i64 0, i64 0>, align 16
    460 
    461 define void @llvm_mips_dotp_u_d_test() nounwind {
    462 entry:
    463   %0 = load <4 x i32>* @llvm_mips_dotp_u_d_ARG1
    464   %1 = load <4 x i32>* @llvm_mips_dotp_u_d_ARG2
    465   %2 = tail call <2 x i64> @llvm.mips.dotp.u.d(<4 x i32> %0, <4 x i32> %1)
    466   store <2 x i64> %2, <2 x i64>* @llvm_mips_dotp_u_d_RES
    467   ret void
    468 }
    469 
    470 declare <2 x i64> @llvm.mips.dotp.u.d(<4 x i32>, <4 x i32>) nounwind
    471 
    472 ; CHECK: llvm_mips_dotp_u_d_test:
    473 ; CHECK: ld.w
    474 ; CHECK: ld.w
    475 ; CHECK: dotp_u.d
    476 ; CHECK: st.d
    477 ; CHECK: .size llvm_mips_dotp_u_d_test
    478 ;
    479