Mercurial > vec
diff utils/gendouble.c @ 45:7955bed1d169 default tip
*: add preliminary floating point support
no x86 intrinsics just yet, but I did add altivec since it's
(arguably) the simplest :)
author | Paper <paper@tflc.us> |
---|---|
date | Wed, 30 Apr 2025 18:36:38 -0400 |
parents | b0a3f0248ecc |
children |
line wrap: on
line diff
--- a/utils/gendouble.c Tue Apr 29 16:54:13 2025 -0400 +++ /dev/null Thu Jan 01 00:00:00 1970 +0000 @@ -1,244 +0,0 @@ -/** - * vec - a tiny SIMD vector library in C99 - * - * Copyright (c) 2024-2025 Paper - * - * Permission is hereby granted, free of charge, to any person obtaining a copy - * of this software and associated documentation files (the "Software"), to deal - * in the Software without restriction, including without limitation the rights - * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell - * copies of the Software, and to permit persons to whom the Software is - * furnished to do so, subject to the following conditions: - * - * The above copyright notice and this permission notice shall be included in all - * copies or substantial portions of the Software. - * - * THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR - * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, - * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE - * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER - * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, - * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE - * SOFTWARE. -**/ - -/* Use this file to generate include/vec/impl/double.h !! - * - * `gcc -o gendouble gendouble.c` */ - -#include <stdio.h> -#include <stdlib.h> -#include <string.h> -#include <ctype.h> - -#define ARRAY_SIZE(x) (sizeof(x)/sizeof((x)[0])) - -/* XXX: would it be faster to unroll literally everything instead of defining everything, - * and then unpacking it all? */ -static const char *header = - "/**\n" - " * vec - a tiny SIMD vector library in C99\n" - " * \n" - " * Copyright (c) 2024-2025 Paper\n" - " * \n" - " * Permission is hereby granted, free of charge, to any person obtaining a copy\n" - " * of this software and associated documentation files (the \"Software\"), to deal\n" - " * in the Software without restriction, including without limitation the rights\n" - " * to use, copy, modify, merge, publish, distribute, sublicense, and/or sell\n" - " * copies of the Software, and to permit persons to whom the Software is\n" - " * furnished to do so, subject to the following conditions:\n" - " * \n" - " * The above copyright notice and this permission notice shall be included in all\n" - " * copies or substantial portions of the Software.\n" - " * \n" - " * THE SOFTWARE IS PROVIDED \"AS IS\", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR\n" - " * IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,\n" - " * FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE\n" - " * AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER\n" - " * LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,\n" - " * OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE\n" - " * SOFTWARE.\n" - "**/\n" - "\n" - "/* This file is automatically generated! Do not edit it directly!\n" - " * Edit the code that generates it in utils/gengeneric.c --paper */\n" - "\n" - "#ifndef VEC_IMPL_DOUBLE_H_\n" - "#define VEC_IMPL_DOUBLE_H_\n" - "\n" - "#define VEC_DOUBLE_SPLAT(sign, bits, size, halfsize) \\\n" - " VEC_FUNC_IMPL v##sign##int##bits##x##size v##sign##int##bits##x##size##_splat(vec_##sign##int##bits x) \\\n" - " { \\\n" - " v##sign##int##bits##x##size vec; \\\n" - " \\\n" - " vec.generic[0] = v##sign##int##bits##x##halfsize##_splat(x); \\\n" - " vec.generic[1] = v##sign##int##bits##x##halfsize##_splat(x); \\\n" - " \\\n" - " return vec; \\\n" - " }\n" - "\n" - "#define VEC_DOUBLE_LOAD_EX(name, sign, bits, size, halfsize) \\\n" - " VEC_FUNC_IMPL v##sign##int##bits##x##size v##sign##int##bits##x##size##_##name(const vec_##sign##int##bits x[size]) \\\n" - " { \\\n" - " v##sign##int##bits##x##size vec; \\\n" - " \\\n" - " vec.generic[0] = v##sign##int##bits##x##halfsize##_##name(x); \\\n" - " vec.generic[1] = v##sign##int##bits##x##halfsize##_##name(x + halfsize); \\\n" - " \\\n" - " return vec; \\\n" - " }\n" - "\n" - "#define VEC_DOUBLE_LOAD(sign, bits, size, halfsize) VEC_DOUBLE_LOAD_EX(load, sign, bits, size, halfsize)\n" - "#define VEC_DOUBLE_LOAD_ALIGNED(sign, bits, size, halfsize) VEC_DOUBLE_LOAD_EX(load_aligned, sign, bits, size, halfsize)\n" - "\n" - "#define VEC_DOUBLE_STORE_EX(name, sign, bits, size, halfsize) \\\n" - " VEC_FUNC_IMPL void v##sign##int##bits##x##size##_##name(v##sign##int##bits##x##size vec, vec_##sign##int##bits x[size]) \\\n" - " { \\\n" - " v##sign##int##bits##x##halfsize##_##name(vec.generic[0], x); \\\n" - " v##sign##int##bits##x##halfsize##_##name(vec.generic[1], x + halfsize); \\\n" - " }\n" - "\n" - "#define VEC_DOUBLE_STORE(sign, bits, size, halfsize) VEC_DOUBLE_STORE_EX(store, sign, bits, size, halfsize)\n" - "#define VEC_DOUBLE_STORE_ALIGNED(sign, bits, size, halfsize) VEC_DOUBLE_STORE_EX(store_aligned, sign, bits, size, halfsize)\n" - "\n" - "#define VEC_DOUBLE_OP(name, sign, bits, size, halfsize, secondsign) \\\n" - " VEC_FUNC_IMPL v##sign##int##bits##x##size v##sign##int##bits##x##size##_##name(v##sign##int##bits##x##size vec1, v##secondsign##int##bits##x##size vec2) \\\n" - " { \\\n" - " vec1.generic[0] = v##sign##int##bits##x##halfsize##_##name(vec1.generic[0], vec2.generic[0]); \\\n" - " vec1.generic[1] = v##sign##int##bits##x##halfsize##_##name(vec1.generic[1], vec2.generic[1]); \\\n" - " \\\n" - " return vec1; \\\n" - " }\n" - "\n" - "#define VEC_DOUBLE_ADD(sign, bits, size, halfsize) VEC_DOUBLE_OP(add, sign, bits, size, halfsize, sign)\n" - "#define VEC_DOUBLE_SUB(sign, bits, size, halfsize) VEC_DOUBLE_OP(sub, sign, bits, size, halfsize, sign)\n" - "#define VEC_DOUBLE_MUL(sign, bits, size, halfsize) VEC_DOUBLE_OP(mul, sign, bits, size, halfsize, sign)\n" - "#define VEC_DOUBLE_DIV(sign, bits, size, halfsize) VEC_DOUBLE_OP(div, sign, bits, size, halfsize, sign)\n" - "#define VEC_DOUBLE_MOD(sign, bits, size, halfsize) VEC_DOUBLE_OP(mod, sign, bits, size, halfsize, sign)\n" - "#define VEC_DOUBLE_AVG(sign, bits, size, halfsize) VEC_DOUBLE_OP(avg, sign, bits, size, halfsize, sign)\n" - "#define VEC_DOUBLE_LSHIFT(sign, bits, size, halfsize) VEC_DOUBLE_OP(lshift, sign, bits, size, halfsize, u)\n" - "#define VEC_DOUBLE_RSHIFT(sign, bits, size, halfsize) VEC_DOUBLE_OP(rshift, sign, bits, size, halfsize, u)\n" - "#define VEC_DOUBLE_LRSHIFT(sign, bits, size, halfsize) VEC_DOUBLE_OP(lrshift, sign, bits, size, halfsize, u)\n" - "#define VEC_DOUBLE_AND(sign, bits, size, halfsize) VEC_DOUBLE_OP(and, sign, bits, size, halfsize, sign)\n" - "#define VEC_DOUBLE_OR(sign, bits, size, halfsize) VEC_DOUBLE_OP(or, sign, bits, size, halfsize, sign)\n" - "#define VEC_DOUBLE_XOR(sign, bits, size, halfsize) VEC_DOUBLE_OP(xor, sign, bits, size, halfsize, sign)\n" - "#define VEC_DOUBLE_MIN(sign, bits, size, halfsize) VEC_DOUBLE_OP(min, sign, bits, size, halfsize, sign)\n" - "#define VEC_DOUBLE_MAX(sign, bits, size, halfsize) VEC_DOUBLE_OP(max, sign, bits, size, halfsize, sign)\n" - "#define VEC_DOUBLE_CMPLT(sign, bits, size, halfsize) VEC_DOUBLE_OP(cmplt, sign, bits, size, halfsize, sign)\n" - "#define VEC_DOUBLE_CMPLE(sign, bits, size, halfsize) VEC_DOUBLE_OP(cmple, sign, bits, size, halfsize, sign)\n" - "#define VEC_DOUBLE_CMPEQ(sign, bits, size, halfsize) VEC_DOUBLE_OP(cmpeq, sign, bits, size, halfsize, sign)\n" - "#define VEC_DOUBLE_CMPGE(sign, bits, size, halfsize) VEC_DOUBLE_OP(cmpge, sign, bits, size, halfsize, sign)\n" - "#define VEC_DOUBLE_CMPGT(sign, bits, size, halfsize) VEC_DOUBLE_OP(cmpgt, sign, bits, size, halfsize, sign)\n" - "\n" - "#define VEC_DOUBLE_NOT(sign, bits, size, halfsize) \\\n" - " VEC_FUNC_IMPL v##sign##int##bits##x##size v##sign##int##bits##x##size##_not(v##sign##int##bits##x##size vec) \\\n" - " { \\\n" - " vec.generic[0] = v##sign##int##bits##x##halfsize##_not(vec.generic[0]); \\\n" - " vec.generic[1] = v##sign##int##bits##x##halfsize##_not(vec.generic[1]); \\\n" - " \\\n" - " return vec; \\\n" - " }\n" - "\n" - "#endif /* VEC_IMPL_DOUBLE_H_ */ \n" - "\n" - "/* ------------------------------------------------------------------------ */\n" - "/* PREPROCESSOR HELL INCOMING */\n" - ""; - -static const char *footer = - "" /* nothing */; - -/* ------------------------------------------------------------------------ */ - -static void print_generic_dbl_op(const char *op, int is_signed, int bits, int size) -{ - printf( - "#if !defined(V%sINT%dx%d_%s_DEFINED) && defined(V%sINT%dx%d_%s_DEFINED)\n" - "VEC_DOUBLE_%s(%s, %d, %d, %d)\n" - "# define V%sINT%dx%d_%s_DEFINED\n" - "#endif\n\n", - (is_signed ? "" : "U"), bits, size, op, (is_signed ? "" : "U"), bits, size / 2, op, - op, (is_signed ? "/* nothing */" : "u"), bits, size, size / 2, - (is_signed ? "" : "U"), bits, size, op); -} - -typedef void (*print_op_spec)(const char *op, int is_signed, int bits, int size); - -static inline void print_ops(int is_signed, int bits, int size, print_op_spec print_op) -{ - /* all supported operations here */ - static const char *ops[] = { - "SPLAT", - "LOAD_ALIGNED", - "LOAD", - "STORE_ALIGNED", - "STORE", - "ADD", - "SUB", - "MUL", - "DIV", - "MOD", - "AVG", - "AND", - "OR", - "XOR", - "NOT", - "CMPLT", - "CMPEQ", - "CMPGT", - "CMPLE", /* these two must be after CMPLT and CMPGT respectfully, */ - "CMPGE", /* because their definitions call those functions */ - "MIN", - "MAX", - "RSHIFT", - "LRSHIFT", - "LSHIFT", - NULL, - }; - int i; - - printf("\n\n/* v%sint%dx%d */\n\n", (is_signed ? "u" : ""), bits, size); - - for (i = 0; ops[i]; i++) - print_op(ops[i], is_signed, bits, size); -} - -int main(void) -{ - static struct { - int bits, size; - print_op_spec print_op; - } defs[] = { - /* -- 8-bit */ - {8, 4, print_generic_dbl_op}, - {8, 8, print_generic_dbl_op}, - {8, 16, print_generic_dbl_op}, - {8, 32, print_generic_dbl_op}, - {8, 64, print_generic_dbl_op}, - - /* -- 16-bit */ - {16, 4, print_generic_dbl_op}, - {16, 8, print_generic_dbl_op}, - {16, 16, print_generic_dbl_op}, - {16, 32, print_generic_dbl_op}, - - /* -- 32-bit */ - {32, 4, print_generic_dbl_op}, - {32, 8, print_generic_dbl_op}, - {32, 16, print_generic_dbl_op}, - - /* -- 64-bit */ - {64, 4, print_generic_dbl_op}, - {64, 8, print_generic_dbl_op}, - }; - int i; - - puts(header); - - for (i = 0; i < ARRAY_SIZE(defs); i++) { - print_ops(1, defs[i].bits, defs[i].size, defs[i].print_op); - print_ops(0, defs[i].bits, defs[i].size, defs[i].print_op); - } - - puts(footer); -}