| 1 |
/* { dg-do compile } */ |
| 2 |
/* { dg-options "-mvsx -O2" } */ |
| 3 |
/* { dg-additional-options "-mdejagnu-cpu=power8" { target { ! has_arch_pwr8 } } } */ |
| 4 |
/* { dg-require-effective-target powerpc_vsx } */ |
| 5 |
/* { dg-require-effective-target int128 } */ |
| 6 |
|
| 7 |
#include <altivec.h> |
| 8 |
|
| 9 |
/* Test POWER8 vector built-ins added for version 1.1 of ELFv2 ABI. */ |
| 10 |
|
| 11 |
vector signed char vsca, vscb, vscc; |
| 12 |
vector unsigned char vuca, vucb, vucc; |
| 13 |
vector signed short vssa, vssb; |
| 14 |
vector bool char vbca, vbcb; |
| 15 |
vector unsigned short vusa, vusb; |
| 16 |
vector bool short vbsa, vbsb; |
| 17 |
vector signed int vsia, vsib, vsic; |
| 18 |
vector unsigned int vuia, vuib, vuic; |
| 19 |
vector bool int vbia, vbib; |
| 20 |
vector signed long long vsla, vslb; |
| 21 |
vector unsigned long long vula, vulb, vulc; |
| 22 |
vector bool long long vbla, vblb, vblc; |
| 23 |
vector signed __int128 vsxa, vsxb, vsxc; |
| 24 |
vector unsigned __int128 vuxa, vuxb, vuxc; |
| 25 |
vector double vda, vdb; |
| 26 |
|
| 27 |
void foo (vector signed char *vscr, |
| 28 |
vector unsigned char *vucr, |
| 29 |
vector bool char *vbcr, |
| 30 |
vector signed short *vssr, |
| 31 |
vector unsigned short *vusr, |
| 32 |
vector bool short *vbsr, |
| 33 |
vector signed int *vsir, |
| 34 |
vector unsigned int *vuir, |
| 35 |
vector bool int *vbir, |
| 36 |
vector unsigned long long *vulr, |
| 37 |
vector bool long long *vblr, |
| 38 |
vector signed __int128 *vsxr, |
| 39 |
vector unsigned __int128 *vuxr, |
| 40 |
vector double *vdr) |
| 41 |
{ |
| 42 |
*vsir++ = vec_addc (vsia, vsib); |
| 43 |
*vuir++ = vec_addc (vuia, vuib); |
| 44 |
*vsxr++ = vec_addc (vsxa, vsxb); |
| 45 |
*vuxr++ = vec_addc (vuxa, vuxb); |
| 46 |
*vsir++ = vec_adde (vsia, vsib, vsic); |
| 47 |
*vuir++ = vec_adde (vuia, vuib, vuic); |
| 48 |
*vsxr++ = vec_adde (vsxa, vsxb, vsxc); |
| 49 |
*vuxr++ = vec_adde (vuxa, vuxb, vuxc); |
| 50 |
*vsir++ = vec_addec (vsia, vsib, vsic); |
| 51 |
*vuir++ = vec_addec (vuia, vuib, vuic); |
| 52 |
*vsxr++ = vec_addec (vsxa, vsxb, vsxc); |
| 53 |
*vuxr++ = vec_addec (vuxa, vuxb, vuxc); |
| 54 |
*vscr++ = vec_adds (vsca, vscb); |
| 55 |
*vucr++ = vec_adds (vuca, vucb); |
| 56 |
*vsir++ = vec_adds (vsia, vsib); |
| 57 |
*vuir++ = vec_adds (vuia, vuib); |
| 58 |
*vssr++ = vec_adds (vssa, vssb); |
| 59 |
*vusr++ = vec_adds (vusa, vusb); |
| 60 |
*vscr++ = vec_adds (vbca, vscb); |
| 61 |
*vscr++ = vec_adds (vsca, vbcb); |
| 62 |
*vucr++ = vec_adds (vbca, vucb); |
| 63 |
*vucr++ = vec_adds (vuca, vbcb); |
| 64 |
*vsir++ = vec_adds (vbia, vsib); |
| 65 |
*vsir++ = vec_adds (vsia, vbib); |
| 66 |
*vuir++ = vec_adds (vbia, vuib); |
| 67 |
*vuir++ = vec_adds (vuia, vbib); |
| 68 |
*vssr++ = vec_adds (vbsa, vssb); |
| 69 |
*vssr++ = vec_adds (vssa, vbsb); |
| 70 |
*vusr++ = vec_adds (vbsa, vusb); |
| 71 |
*vusr++ = vec_adds (vusa, vbsb); |
| 72 |
*vucr++ = vec_bperm (vuca, vucb); |
| 73 |
*vulr++ = vec_bperm (vuxa, vucb); |
| 74 |
*vbcr++ = vec_eqv (vbca, vbcb); |
| 75 |
*vbir++ = vec_eqv (vbia, vbib); |
| 76 |
*vblr++ = vec_eqv (vbla, vblb); |
| 77 |
*vbsr++ = vec_eqv (vbsa, vbsb); |
| 78 |
*vucr++ = vec_gb (vuca); |
| 79 |
*vbcr++ = vec_nand (vbca, vbcb); |
| 80 |
*vbir++ = vec_nand (vbia, vbib); |
| 81 |
*vblr++ = vec_nand (vbla, vblb); |
| 82 |
*vbsr++ = vec_nand (vbsa, vbsb); |
| 83 |
*vbcr++ = vec_orc (vbca, vbcb); |
| 84 |
*vbir++ = vec_orc (vbia, vbib); |
| 85 |
*vblr++ = vec_orc (vbla, vblb); |
| 86 |
*vbsr++ = vec_orc (vbsa, vbsb); |
| 87 |
*vblr++ = vec_perm (vbla, vblb, vucc); |
| 88 |
*vusr++ = vec_pmsum_be (vuca, vucb); |
| 89 |
*vuir++ = vec_pmsum_be (vusa, vusb); |
| 90 |
*vulr++ = vec_pmsum_be (vuia, vuib); |
| 91 |
*vuxr++ = vec_pmsum_be (vula, vulb); |
| 92 |
*vuir++ = vec_shasigma_be (vuia, 0, 1); |
| 93 |
*vulr++ = vec_shasigma_be (vula, 0, 1); |
| 94 |
*vsir++ = vec_subc (vsia, vsib); |
| 95 |
*vuir++ = vec_subc (vuia, vuib); |
| 96 |
*vsxr++ = vec_subc (vsxa, vsxb); |
| 97 |
*vuxr++ = vec_subc (vuxa, vuxb); |
| 98 |
*vsir++ = vec_sube (vsia, vsib, vsic); |
| 99 |
*vuir++ = vec_sube (vuia, vuib, vuic); |
| 100 |
*vsxr++ = vec_sube (vsxa, vsxb, vsxc); |
| 101 |
*vuxr++ = vec_sube (vuxa, vuxb, vuxc); |
| 102 |
*vsir++ = vec_subec (vsia, vsib, vsic); |
| 103 |
*vuir++ = vec_subec (vuia, vuib, vuic); |
| 104 |
*vsxr++ = vec_subec (vsxa, vsxb, vsxc); |
| 105 |
*vuxr++ = vec_subec (vuxa, vuxb, vuxc); |
| 106 |
*vscr++ = vec_subs (vsca, vscb); |
| 107 |
*vucr++ = vec_subs (vuca, vucb); |
| 108 |
*vsir++ = vec_subs (vsia, vsib); |
| 109 |
*vuir++ = vec_subs (vuia, vuib); |
| 110 |
*vssr++ = vec_subs (vssa, vssb); |
| 111 |
*vusr++ = vec_subs (vusa, vusb); |
| 112 |
*vsir++ = vec_sum2s (vsia, vsib); |
| 113 |
*vsir++ = vec_sum4s (vsca, vsib); |
| 114 |
*vsir++ = vec_sum4s (vssa, vsib); |
| 115 |
*vuir++ = vec_sum4s (vuca, vuib); |
| 116 |
|
| 117 |
} |
| 118 |
|
| 119 |
/* { dg-final { scan-assembler-times "vaddcuq" 2 } } */ |
| 120 |
/* { dg-final { scan-assembler-times "vaddeuqm" 2 } } */ |
| 121 |
/* { dg-final { scan-assembler-times "vaddecuq" 2 } } */ |
| 122 |
/* { dg-final { scan-assembler-times "vaddcuw" 6 } } */ |
| 123 |
/* { dg-final { scan-assembler-times "vadduwm" 4 } } */ |
| 124 |
/* { dg-final { scan-assembler-times "vsubcuq" 2 } } */ |
| 125 |
/* { dg-final { scan-assembler-times "vsubeuqm" 2 } } */ |
| 126 |
/* { dg-final { scan-assembler-times "vsubecuq" 2 } } */ |
| 127 |
/* { dg-final { scan-assembler-times "vsubcuw" 4 } } */ |
| 128 |
/* { dg-final { scan-assembler-times "vsubuwm" 4 } } */ |
| 129 |
/* { dg-final { scan-assembler-times "vbpermq" 2 } } */ |
| 130 |
/* { dg-final { scan-assembler-times "xxleqv" 4 } } */ |
| 131 |
/* { dg-final { scan-assembler-times "vgbbd" 1 } } */ |
| 132 |
/* { dg-final { scan-assembler-times "xxlnand" 4 } } */ |
| 133 |
/* { dg-final { scan-assembler-times "xxlorc" 4 } } */ |
| 134 |
/* { dg-final { scan-assembler-times "vperm" 1 } } */ |
| 135 |
/* { dg-final { scan-assembler-times "vpmsumb" 1 } } */ |
| 136 |
/* { dg-final { scan-assembler-times "vpmsumh" 1 } } */ |
| 137 |
/* { dg-final { scan-assembler-times "vpmsumw" 1 } } */ |
| 138 |
/* { dg-final { scan-assembler-times "vpmsumd" 1 } } */ |
| 139 |
/* { dg-final { scan-assembler-times "vshasigmaw" 1 } } */ |
| 140 |
/* { dg-final { scan-assembler-times "vshasigmad" 1 } } */ |
| 141 |
/* { dg-final { scan-assembler-times "vsubsbs" 1 } } vec_subs */ |
| 142 |
/* { dg-final { scan-assembler-times "vsububs" 1 } } vec_subs */ |
| 143 |
/* { dg-final { scan-assembler-times "vsubsws" 1 } } vec_subs */ |
| 144 |
/* { dg-final { scan-assembler-times "vsubuws" 1 } } vec_subs */ |
| 145 |
/* { dg-final { scan-assembler-times "vsubshs" 1 } } vec_subs */ |
| 146 |
/* { dg-final { scan-assembler-times "vsubuhs" 1 } } vec_subs */ |
| 147 |
/* { dg-final { scan-assembler-times "vsum2sws" 1 } } vec_sum2s */ |
| 148 |
/* { dg-final { scan-assembler-times "vsum4sws" 0 } } vec_sum4s */ |
| 149 |
/* { dg-final { scan-assembler-times "vsum4shs" 1 } } vec_sum4s */ |
| 150 |
/* { dg-final { scan-assembler-times "vsum4ubs" 1 } } vec_sum4s */ |
| 151 |
|