1 /* Test program for bfloat16 of AVX 512 registers. 2 3 Copyright 2020-2024 Free Software Foundation, Inc. 4 5 This file is part of GDB. 6 7 This program is free software; you can redistribute it and/or modify 8 it under the terms of the GNU General Public License as published by 9 the Free Software Foundation; either version 3 of the License, or 10 (at your option) any later version. 11 12 This program is distributed in the hope that it will be useful, 13 but WITHOUT ANY WARRANTY; without even the implied warranty of 14 MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the 15 GNU General Public License for more details. 16 17 You should have received a copy of the GNU General Public License 18 along with this program. If not, see <http://www.gnu.org/licenses/>. */ 19 20 typedef struct 21 { 22 float f[4]; 23 } v4sd_t; 24 25 typedef struct 26 { 27 float f[8]; 28 } v8sd_t; 29 30 typedef struct 31 { 32 float f[16]; 33 } v16sd_t; 34 35 v4sd_t xmm_data[] = 36 { 37 { { 0.0, 0.125, 0.25, 0.375 } }, 38 { { 0.5, 0.625, 0.75, 0.875 } }, 39 { { 1.0, 1.125, 1.25, 1.375 } }, 40 { { 1.5, 1.625, 1.75, 1.875 } }, 41 { { 2.0, 2.125, 2.25, 2.375 } }, 42 { { 2.5, 2.625, 2.75, 2.875 } }, 43 { { 3.0, 3.125, 3.25, 3.375 } }, 44 { { 3.5, 3.625, 3.75, 3.875 } }, 45 }; 46 47 v8sd_t ymm_data[] = 48 { 49 { { 8.0, 8.25, 8.5, 8.75, 9.0, 9.25, 9.5, 9.75 } }, 50 { { 10.0, 10.25, 10.5, 10.75, 11.0, 11.25, 11.5, 11.75 } }, 51 { { 12.0, 12.25, 12.5, 12.75, 13.0, 13.25, 13.5, 13.75 } }, 52 { { 14.0, 14.25, 14.5, 14.75, 15.0, 15.25, 15.5, 15.75 } }, 53 { { 16.0, 16.25, 16.5, 16.75, 17.0, 17.25, 17.5, 17.75 } }, 54 { { 18.0, 18.25, 18.5, 18.75, 19.0, 19.25, 19.5, 19.75 } }, 55 { { 20.0, 20.25, 20.5, 20.75, 21.0, 21.25, 21.5, 21.75 } }, 56 { { 22.0, 22.25, 22.5, 22.75, 23.0, 23.25, 23.5, 23.75 } }, 57 }; 58 59 v16sd_t zmm_data[] = 60 { 61 { { 20.0, 20.5, 21.0, 21.5, 22.0, 22.5, 23.0, 23.5, 24.0, 24.5, 62 25.0, 25.5, 26.0, 26.5, 27.0, 27.5 } }, 63 { { 28.0, 28.5, 29.0, 29.5, 30.0, 30.5, 31.0, 31.5, 32.0, 32.5, 64 33.0, 33.5, 34.0, 34.5, 35.0, 35.5 } }, 65 { { 36.0, 36.5, 37.0, 37.5, 38.0, 38.5, 39.0, 39.5, 40.0, 40.5, 66 41.0, 41.5, 42.0, 42.5, 43.0, 43.5 } }, 67 { { 44.0, 44.5, 45.0, 45.5, 46.0, 46.5, 47.0, 47.5, 48.0, 48.5, 68 49.0, 49.5, 50.0, 50.5, 51.0, 51.5 } }, 69 { { 52.0, 52.5, 53.0, 53.5, 54.0, 54.5, 55.0, 55.5, 56.0, 56.5, 70 57.0, 57.5, 58.0, 58.5, 59.0, 59.5 } }, 71 { { 60.0, 60.5, 61.0, 61.5, 62.0, 62.5, 63.0, 63.5, 64.0, 64.5, 72 65.0, 65.5, 66.0, 66.5, 67.0, 67.5 } }, 73 { { 68.0, 68.5, 69.0, 69.5, 70.0, 70.5, 71.0, 71.5, 72.0, 72.5, 74 73.0, 73.5, 74.0, 74.5, 75.0, 75.5 } }, 75 { { 76.0, 76.5, 77.0, 77.5, 78.0, 78.5, 79.0, 79.5, 80.0, 80.5, 76 81.0, 81.5, 82.0, 82.5, 83.0, 83.5 } }, 77 }; 78 79 void 80 move_data_to_xmm_reg (void) 81 { 82 asm ("vmovups 0(%0), %%xmm0 \n\t" 83 "vmovups 16(%0), %%xmm1 \n\t" 84 "vmovups 32(%0), %%xmm2 \n\t" 85 "vmovups 48(%0), %%xmm3 \n\t" 86 "vmovups 64(%0), %%xmm4 \n\t" 87 "vmovups 80(%0), %%xmm5 \n\t" 88 "vmovups 96(%0), %%xmm6 \n\t" 89 "vmovups 112(%0), %%xmm7 \n\t" 90 : /* no output operands */ 91 : "r" (xmm_data)); 92 } 93 94 void 95 move_data_to_ymm_reg (void) 96 { 97 asm ("vmovups 0(%0), %%ymm0 \n\t" 98 "vmovups 32(%0), %%ymm1 \n\t" 99 "vmovups 64(%0), %%ymm2 \n\t" 100 "vmovups 96(%0), %%ymm3 \n\t" 101 "vmovups 128(%0), %%ymm4 \n\t" 102 "vmovups 160(%0), %%ymm5 \n\t" 103 "vmovups 192(%0), %%ymm6 \n\t" 104 "vmovups 224(%0), %%ymm7 \n\t" 105 : /* no output operands */ 106 : "r" (ymm_data)); 107 } 108 109 void 110 move_data_to_zmm_reg (void) 111 { 112 asm ("vmovups 0(%0), %%zmm0 \n\t" 113 "vmovups 64(%0), %%zmm1 \n\t" 114 "vmovups 128(%0), %%zmm2 \n\t" 115 "vmovups 192(%0), %%zmm3 \n\t" 116 "vmovups 256(%0), %%zmm4 \n\t" 117 "vmovups 320(%0), %%zmm5 \n\t" 118 "vmovups 384(%0), %%zmm6 \n\t" 119 "vmovups 448(%0), %%zmm7 \n\t" 120 : /* no output operands */ 121 : "r" (zmm_data)); 122 } 123 124 void 125 convert_xmm_from_float_to_bfloat16 (void) 126 { 127 asm("vcvtne2ps2bf16 %xmm0, %xmm1, %xmm0"); 128 asm("vcvtne2ps2bf16 %xmm6, %xmm7, %xmm6"); 129 } 130 131 void 132 convert_ymm_from_float_to_bfloat16 (void) 133 { 134 asm("vcvtne2ps2bf16 %ymm0, %ymm1, %ymm0"); 135 asm("vcvtne2ps2bf16 %ymm6, %ymm7, %ymm6"); 136 } 137 138 void 139 convert_zmm_from_float_to_bfloat16 (void) 140 { 141 asm("vcvtne2ps2bf16 %zmm0, %zmm1, %zmm0"); 142 asm("vcvtne2ps2bf16 %zmm6, %zmm7, %zmm6"); 143 } 144 145 int 146 main (int argc, char **argv) 147 { 148 /* Move initial values from array to registers and read from XMM regs. */ 149 move_data_to_xmm_reg (); 150 convert_xmm_from_float_to_bfloat16 (); 151 asm ("nop"); /* first breakpoint here */ 152 153 /* Move initial values from array to registers and read from YMM regs. */ 154 move_data_to_ymm_reg (); 155 convert_ymm_from_float_to_bfloat16 (); 156 asm ("nop"); /* second breakpoint here */ 157 158 /* Move initial values from array to registers and read from ZMM regs. */ 159 move_data_to_zmm_reg (); 160 convert_zmm_from_float_to_bfloat16 (); 161 asm ("nop"); /* third breakpoint here */ 162 163 return 0; 164 } 165
Indexes created Mon Jun 29 00:24:49 UTC 2026