nishi/milsko
1
Fork 0
mirror of https://gitea.nishi.boats/pyrite-dev/milsko synced 2026-01-10 19:33:28 +00:00
Code Issues Packages Projects Releases Wiki Activity

merge pr #7 (take 2)

Browse Source 
git-svn-id: http://svn2.nishi.boats/svn/milsko/trunk@540 b9cfdab3-6d41-4d17-bbe4-086880011989
This commit is contained in:
IoIxD
2025-11-01 03:37:40 +00:00
parent 141ac076e3
commit 3719f17250
10 changed files with 1053 additions and 0 deletions
Download Patch File Download Diff File Expand all files Collapse all files

103
src/math/mmx.c Normal file
View File

@@ -0,0 +1,103 @@
#include <Mw/LowLevelMath.h>
#include "math.h"
#include <assert.h>
#include <mmintrin.h>
#include <stdio.h>
#include <x86intrin.h>
#define DO_MMX_INTRINSIC(intrin, _ty, _rty, _tyn) \
__m64 m = intrin(*(__m64*)&a->un._ty, *(__m64*)&b->un._ty); \
out->un._rty = *(struct _tyn*)&m;
static void mmx_add_u8(MwLLVec* a, MwLLVec* b, MwLLVec* out) {
DO_MMX_INTRINSIC(_m_paddusb, u8, u8, _MwLLVecDataU8x8);
};
static void mmx_sub_u8(MwLLVec* a, MwLLVec* b, MwLLVec* out) {
DO_MMX_INTRINSIC(_m_psubusb, u8, u8, _MwLLVecDataU8x8);
};
static void mmx_equal_u8(MwLLVec* a, MwLLVec* b, MwLLVec* out) {
DO_MMX_INTRINSIC(_m_pcmpeqb, u8, u8, _MwLLVecDataU8x8);
};
static void mmx_greaterThen_u8(MwLLVec* a, MwLLVec* b, MwLLVec* out) {
DO_MMX_INTRINSIC(_m_pcmpgtb, u8, u8, _MwLLVecDataU8x8);
};
static void mmx_add_u16(MwLLVec* a, MwLLVec* b, MwLLVec* out) {
DO_MMX_INTRINSIC(_m_paddusw, u16, u16, _MwLLVecDataU16x4);
}
static void mmx_sub_u16(MwLLVec* a, MwLLVec* b, MwLLVec* out) {
DO_MMX_INTRINSIC(_m_psubusw, u16, u16, _MwLLVecDataU16x4);
}
static void mmx_shiftRight_u16(MwLLVec* a, MwLLVec* b, MwLLVec* out) {
DO_MMX_INTRINSIC(_m_psrlw, u16, u16, _MwLLVecDataU16x4);
};
static void mmx_shiftLeft_u16(MwLLVec* a, MwLLVec* b, MwLLVec* out) {
DO_MMX_INTRINSIC(_m_psllw, u16, u16, _MwLLVecDataU16x4);
}
static void mmx_equal_u16(MwLLVec* a, MwLLVec* b, MwLLVec* out) {
DO_MMX_INTRINSIC(_m_pcmpeqw, u16, u16, _MwLLVecDataU16x4);
}
static void mmx_greaterThen_u16(MwLLVec* a, MwLLVec* b, MwLLVec* out) {
DO_MMX_INTRINSIC(_m_pcmpgtw, u16, u16, _MwLLVecDataU16x4);
}
static void mmx_add_u32(MwLLVec* a, MwLLVec* b, MwLLVec* out) {
DO_MMX_INTRINSIC(_m_paddd, u32, u32, _MwLLVecDataU32x2);
}
static void mmx_sub_u32(MwLLVec* a, MwLLVec* b, MwLLVec* out) {
DO_MMX_INTRINSIC(_m_psubd, u32, u32, _MwLLVecDataU32x2);
}
static void mmx_shiftRight_u32(MwLLVec* a, MwLLVec* b, MwLLVec* out) {
DO_MMX_INTRINSIC(_m_psrld, u32, u32, _MwLLVecDataU32x2);
};
static void mmx_shiftLeft_u32(MwLLVec* a, MwLLVec* b, MwLLVec* out) {
DO_MMX_INTRINSIC(_m_pslld, u32, u32, _MwLLVecDataU32x2);
}
static void mmx_equal_u32(MwLLVec* a, MwLLVec* b, MwLLVec* out) {
DO_MMX_INTRINSIC(_m_pcmpeqw, u32, u32, _MwLLVecDataU32x2);
}
static void mmx_greaterThen_u32(MwLLVec* a, MwLLVec* b, MwLLVec* out) {
DO_MMX_INTRINSIC(_m_pcmpgtw, u32, u32, _MwLLVecDataU32x2);
}
static void mmx_add_i8(MwLLVec* a, MwLLVec* b, MwLLVec* out) {
DO_MMX_INTRINSIC(_m_paddsb, i8, i8, _MwLLVecDataI8x8);
};
static void mmx_sub_i8(MwLLVec* a, MwLLVec* b, MwLLVec* out) {
DO_MMX_INTRINSIC(_m_psubsb, i8, i8, _MwLLVecDataI8x8);
};
static void mmx_add_i16(MwLLVec* a, MwLLVec* b, MwLLVec* out) {
DO_MMX_INTRINSIC(_m_paddsw, i16, i16, _MwLLVecDataI16x4);
}
static void mmx_sub_i16(MwLLVec* a, MwLLVec* b, MwLLVec* out) {
DO_MMX_INTRINSIC(_m_psubsw, i16, i16, _MwLLVecDataI16x4);
}
void mmx_apply(MwLLMathVTable** t) {
t[_MwLLVecTypeU8x8]->Add = mmx_add_u8;
t[_MwLLVecTypeU8x8]->Sub = mmx_sub_u8;
t[_MwLLVecTypeU8x8]->GreaterThen = mmx_greaterThen_u8;
t[_MwLLVecTypeU8x8]->Equal = mmx_equal_u8;
t[_MwLLVecTypeU16x4]->Add = mmx_add_u16;
t[_MwLLVecTypeU16x4]->Sub = mmx_sub_u16;
t[_MwLLVecTypeU16x4]->ShiftLeft = mmx_shiftLeft_u16;
t[_MwLLVecTypeU16x4]->ShiftRight = mmx_shiftRight_u16;
t[_MwLLVecTypeU16x4]->GreaterThen = mmx_greaterThen_u16;
t[_MwLLVecTypeU16x4]->Equal = mmx_equal_u16;
t[_MwLLVecTypeU32x2]->Add = mmx_add_u32;
t[_MwLLVecTypeU32x2]->Sub = mmx_sub_u32;
t[_MwLLVecTypeU32x2]->ShiftLeft = mmx_shiftLeft_u32;
t[_MwLLVecTypeU32x2]->ShiftRight = mmx_shiftRight_u32;
t[_MwLLVecTypeU32x2]->GreaterThen = mmx_greaterThen_u32;
t[_MwLLVecTypeU32x2]->Equal = mmx_equal_u32;
t[_MwLLVecTypeI8x8]->Add = mmx_add_i8;
t[_MwLLVecTypeI8x8]->Sub = mmx_sub_i8;
t[_MwLLVecTypeI16x4]->Add = mmx_add_i16;
t[_MwLLVecTypeI16x4]->Sub = mmx_sub_i16;
t[_MwLLVecTypeI16x4]->ShiftLeft = mmx_shiftLeft_u16;
t[_MwLLVecTypeI16x4]->ShiftRight = mmx_shiftRight_u16;
t[_MwLLVecTypeI32x2]->ShiftLeft = mmx_shiftLeft_u32;
t[_MwLLVecTypeI32x2]->ShiftRight = mmx_shiftRight_u32;
}