1 #include <stdio.h> 2 3 typedef unsigned char v16i8 __attribute__((ext_vector_type(16))); 4 typedef short v8i16 __attribute__((ext_vector_type(16))); 5 typedef int v4i32 __attribute__((ext_vector_type(4))); 6 typedef float v4f32 __attribute__((ext_vector_type(4))); 7 typedef long long v2i64 __attribute__((ext_vector_type(2))); 8 typedef double v2f64 __attribute__((ext_vector_type(2))); 9 10 void print_v16i8(const char *str, const v16i8 v) { 11 union { 12 unsigned char elts[16]; 13 v16i8 vec; 14 } tv; 15 tv.vec = v; 16 printf("%s = { %hhu, %hhu, %hhu, %hhu, %hhu, %hhu, %hhu, " 17 "%hhu, %hhu, %hhu, %hhu, %hhu, %hhu, %hhu, " 18 "%hhu, %hhu }\n", 19 str, tv.elts[0], tv.elts[1], tv.elts[2], tv.elts[3], tv.elts[4], tv.elts[5], 20 tv.elts[6], tv.elts[7], tv.elts[8], tv.elts[9], tv.elts[10], tv.elts[11], 21 tv.elts[12], tv.elts[13], tv.elts[14], tv.elts[15]); 22 } 23 24 void print_v16i8_hex(const char *str, const v16i8 v) { 25 union { 26 unsigned char elts[16]; 27 v16i8 vec; 28 } tv; 29 tv.vec = v; 30 printf("%s = { 0x%02hhx, 0x%02hhx, 0x%02hhx, 0x%02hhx, 0x%02hhx, 0x%02hhx, 0x%02hhx, " 31 "0x%02hhx, 0x%02hhx, 0x%02hhx, 0x%02hhx, 0x%02hhx, 0x%02hhx, 0x%02hhx, " 32 "0x%02hhx, 0x%02hhx }\n", 33 str, tv.elts[0], tv.elts[1], tv.elts[2], tv.elts[3], tv.elts[4], tv.elts[5], 34 tv.elts[6], tv.elts[7], tv.elts[8], tv.elts[9], tv.elts[10], tv.elts[11], 35 tv.elts[12], tv.elts[13], tv.elts[14], tv.elts[15]); 36 } 37 38 void print_v8i16_hex(const char *str, v8i16 v) { 39 union { 40 short elts[8]; 41 v8i16 vec; 42 } tv; 43 tv.vec = v; 44 printf("%s = { 0x%04hx, 0x%04hx, 0x%04hx, 0x%04hx, 0x%04hx, " 45 "0x%04hx, 0x%04hx, 0x%04hx }\n", 46 str, tv.elts[0], tv.elts[1], tv.elts[2], tv.elts[3], tv.elts[4], 47 tv.elts[5], tv.elts[6], tv.elts[7]); 48 } 49 50 void print_v4i32(const char *str, v4i32 v) { 51 printf("%s = { %d, %d, %d, %d }\n", str, v.x, v.y, v.z, v.w); 52 } 53 54 void print_v4f32(const char *str, v4f32 v) { 55 printf("%s = { %f, %f, %f, %f }\n", str, v.x, v.y, v.z, v.w); 56 } 57 58 void print_v2i64(const char *str, v2i64 v) { 59 printf("%s = { %lld, %lld }\n", str, v.x, v.y); 60 } 61 62 void print_v2f64(const char *str, v2f64 v) { 63 printf("%s = { %g, %g }\n", str, v.x, v.y); 64 } 65 66 /*----------------------------------------------------------------------*/ 67 68 v16i8 v16i8_mpy(v16i8 v1, v16i8 v2) { 69 return v1 * v2; 70 } 71 72 v16i8 v16i8_add(v16i8 v1, v16i8 v2) { 73 return v1 + v2; 74 } 75 76 v4i32 v4i32_shuffle_1(v4i32 a) { 77 v4i32 c2 = a.yzwx; 78 return c2; 79 } 80 81 v4i32 v4i32_shuffle_2(v4i32 a) { 82 v4i32 c2 = a.zwxy; 83 return c2; 84 } 85 86 v4i32 v4i32_shuffle_3(v4i32 a) { 87 v4i32 c2 = a.wxyz; 88 return c2; 89 } 90 91 v4i32 v4i32_shuffle_4(v4i32 a) { 92 v4i32 c2 = a.xyzw; 93 return c2; 94 } 95 96 v4i32 v4i32_shuffle_5(v4i32 a) { 97 v4i32 c2 = a.xwzy; 98 return c2; 99 } 100 101 v4f32 v4f32_shuffle_1(v4f32 a) { 102 v4f32 c2 = a.yzwx; 103 return c2; 104 } 105 106 v4f32 v4f32_shuffle_2(v4f32 a) { 107 v4f32 c2 = a.zwxy; 108 return c2; 109 } 110 111 v4f32 v4f32_shuffle_3(v4f32 a) { 112 v4f32 c2 = a.wxyz; 113 return c2; 114 } 115 116 v4f32 v4f32_shuffle_4(v4f32 a) { 117 v4f32 c2 = a.xyzw; 118 return c2; 119 } 120 121 v4f32 v4f32_shuffle_5(v4f32 a) { 122 v4f32 c2 = a.xwzy; 123 return c2; 124 } 125 126 v2i64 v2i64_shuffle(v2i64 a) { 127 v2i64 c2 = a.yx; 128 return c2; 129 } 130 131 v2f64 v2f64_shuffle(v2f64 a) { 132 v2f64 c2 = a.yx; 133 return c2; 134 } 135 136 int main(void) { 137 v16i8 v00 = { 0xf4, 0xad, 0x01, 0xe9, 0x51, 0x78, 0xc1, 0x8a, 138 0x94, 0x7c, 0x49, 0x6c, 0x21, 0x32, 0xb2, 0x04 }; 139 v16i8 va0 = { 0x01, 0x02, 0x03, 0x04, 0x05, 0x06, 0x07, 0x08, 140 0x09, 0x0a, 0x0b, 0x0c, 0x0d, 0x0e, 0x0f, 0x10 }; 141 v16i8 va1 = { 0x11, 0x83, 0x4b, 0x63, 0xff, 0x90, 0x32, 0xe5, 142 0x5a, 0xaa, 0x20, 0x01, 0x0d, 0x15, 0x77, 0x05 }; 143 v8i16 v01 = { 0x1a87, 0x0a14, 0x5014, 0xfff0, 144 0xe194, 0x0184, 0x801e, 0x5940 }; 145 v4i32 v1 = { 1, 2, 3, 4 }; 146 v4f32 v2 = { 1.0, 2.0, 3.0, 4.0 }; 147 v2i64 v3 = { 691043ll, 910301513ll }; 148 v2f64 v4 = { 5.8e56, 9.103e-62 }; 149 150 puts("---- vector tests start ----"); 151 152 print_v16i8_hex("v00 ", v00); 153 print_v16i8_hex("va0 ", va0); 154 print_v16i8_hex("va1 ", va1); 155 print_v16i8_hex("va0 x va1 ", v16i8_mpy(va0, va1)); 156 print_v16i8_hex("va0 + va1 ", v16i8_add(va0, va1)); 157 print_v8i16_hex("v01 ", v01); 158 159 print_v4i32("v4i32_shuffle_1(1, 2, 3, 4)", v4i32_shuffle_1(v1)); 160 print_v4i32("v4i32_shuffle_2(1, 2, 3, 4)", v4i32_shuffle_2(v1)); 161 print_v4i32("v4i32_shuffle_3(1, 2, 3, 4)", v4i32_shuffle_3(v1)); 162 print_v4i32("v4i32_shuffle_4(1, 2, 3, 4)", v4i32_shuffle_4(v1)); 163 print_v4i32("v4i32_shuffle_5(1, 2, 3, 4)", v4i32_shuffle_5(v1)); 164 165 print_v4f32("v4f32_shuffle_1(1, 2, 3, 4)", v4f32_shuffle_1(v2)); 166 print_v4f32("v4f32_shuffle_2(1, 2, 3, 4)", v4f32_shuffle_2(v2)); 167 print_v4f32("v4f32_shuffle_3(1, 2, 3, 4)", v4f32_shuffle_3(v2)); 168 print_v4f32("v4f32_shuffle_4(1, 2, 3, 4)", v4f32_shuffle_4(v2)); 169 print_v4f32("v4f32_shuffle_5(1, 2, 3, 4)", v4f32_shuffle_5(v2)); 170 171 print_v2i64("v3 ", v3); 172 print_v2i64("v2i64_shuffle ", v2i64_shuffle(v3)); 173 print_v2f64("v4 ", v4); 174 print_v2f64("v2f64_shuffle ", v2f64_shuffle(v4)); 175 176 puts("---- vector tests end ----"); 177 178 return 0; 179 } 180