sgilissen/llamaplayer
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Common: Add intrin.h

Browse Source
This commit is contained in:
Stenzek
2023-10-01 14:12:25 +10:00
parent b1bb33a566
commit 01e505ec8b
26 changed files with 235 additions and 206 deletions
Download Patch File Download Diff File Expand all files Collapse all files

19
src/core/gpu_sw.cpp
View File

@ -8,6 +8,7 @@
#include "common/align.h"
#include "common/assert.h"
#include "common/intrin.h"
#include "common/log.h"
#include "common/make_array.h"
@ -15,16 +16,6 @@
Log_SetChannel(GPU_SW);
#if defined(CPU_ARCH_X64)
#include <emmintrin.h>
#elif defined(CPU_ARCH_ARM64)
#ifdef _MSC_VER
#include <arm64_neon.h>
#else
#include <arm_neon.h>
#endif
#endif
template<typename T>
ALWAYS_INLINE static constexpr std::tuple<T, T> MinMax(T v1, T v2)
{
@ -162,7 +153,7 @@ ALWAYS_INLINE void CopyOutRow16<GPUTexture::Format::RGBA5551, u16>(const u16* sr
{
u32 col = 0;
#if defined(CPU_ARCH_X64)
#if defined(CPU_ARCH_SSE)
const u32 aligned_width = Common::AlignDownPow2(width, 8);
for (; col < aligned_width; col += 8)
{
@ -176,7 +167,7 @@ ALWAYS_INLINE void CopyOutRow16<GPUTexture::Format::RGBA5551, u16>(const u16* sr
_mm_storeu_si128(reinterpret_cast<__m128i*>(dst_ptr), value);
dst_ptr += 8;
}
#elif defined(CPU_ARCH_ARM64)
#elif defined(CPU_ARCH_NEON)
const u32 aligned_width = Common::AlignDownPow2(width, 8);
for (; col < aligned_width; col += 8)
{
@ -201,7 +192,7 @@ ALWAYS_INLINE void CopyOutRow16<GPUTexture::Format::RGB565, u16>(const u16* src_
{
u32 col = 0;
#if defined(CPU_ARCH_X64)
#if defined(CPU_ARCH_SSE)
const u32 aligned_width = Common::AlignDownPow2(width, 8);
for (; col < aligned_width; col += 8)
{
@ -216,7 +207,7 @@ ALWAYS_INLINE void CopyOutRow16<GPUTexture::Format::RGB565, u16>(const u16* src_
_mm_storeu_si128(reinterpret_cast<__m128i*>(dst_ptr), value);
dst_ptr += 8;
}
#elif defined(CPU_ARCH_ARM64)
#elif defined(CPU_ARCH_NEON)
const u32 aligned_width = Common::AlignDownPow2(width, 8);
const uint16x8_t single_mask = vdupq_n_u16(0x1F);
for (; col < aligned_width; col += 8)