archive/rpcs3
1
0
Fork 0
mirror of https://github.com/RPCS3/rpcs3.git synced 2025-07-05 14:31:24 +12:00
Code Issues Projects Releases Packages Wiki Activity Actions

Initial Linux Aarch64 support

Browse source 
* Update asmjit dependency (aarch64 branch)
* Disable USE_DISCORD_RPC by default
* Dump some JIT objects in rpcs3 cache dir
* Add SIGILL handler for all platforms
* Fix resetting zeroing denormals in thread pool
* Refactor most v128:: utils into global gv_** functions
* Refactor PPU interpreter (incomplete), remove "precise"
* - Instruction specializations with multiple accuracy flags
* - Adjust calling convention for speed
* - Removed precise/fast setting, replaced with static
* - Started refactoring interpreters for building at runtime JIT
*   (I got tired of poor compiler optimizations)
* - Expose some accuracy settings (SAT, NJ, VNAN, FPCC)
* - Add exec_bytes PPU thread variable (akin to cycle count)
* PPU LLVM: fix VCTUXS+VCTSXS instruction NaN results
* SPU interpreter: remove "precise" for now (extremely non-portable)
* - As with PPU, settings changed to static/dynamic for interpreters.
* - Precise options will be implemented later
* Fix termination after fatal error dialog
This commit is contained in:
Nekotekina 2021-12-30 19:39:18 +03:00
parent d6aa834b5f
commit 580bd2b25e
89 changed files with 20360 additions and 5612 deletions
Download patch file Download diff file Expand all files Collapse all files

30
rpcs3/Emu/RSX/Program/program_state_cache2.hpp
View file

@ -2,9 +2,12 @@
#include "ProgramStateCache.h"
#include "emmintrin.h"
#include "util/asm.hpp"
#if defined(ARCH_X64)
#include "emmintrin.h"
#endif
template <typename Traits>
void program_state_cache<Traits>::fill_fragment_constants_buffer(std::span<f32> dst_buffer, const RSXFragmentProgram &fragment_program, bool sanitize) const
{
@ -19,12 +22,23 @@ void program_state_cache<Traits>::fill_fragment_constants_buffer(std::span<f32>
for (usz offset_in_fragment_program : I->second.FragmentConstantOffsetCache)
{
char* data = static_cast<char*>(fragment_program.get_data()) + offset_in_fragment_program;
#if defined(ARCH_X64)
const __m128i vector = _mm_loadu_si128(reinterpret_cast<__m128i*>(data));
const __m128i shuffled_vector = _mm_or_si128(_mm_slli_epi16(vector, 8), _mm_srli_epi16(vector, 8));
#else
for (u32 i = 0; i < 4; i++)
{
const u32 value = reinterpret_cast<u32*>(data)[i];
tmp[i] = std::bit_cast<f32, u32>(((value >> 8) & 0xff00ff) | ((value << 8) & 0xff00ff00));
}
#endif
if (!patch_table.is_empty())
{
#if defined(ARCH_X64)
_mm_store_ps(tmp, _mm_castsi128_ps(shuffled_vector));
#endif
for (int i = 0; i < 4; ++i)
{
@ -47,15 +61,29 @@ void program_state_cache<Traits>::fill_fragment_constants_buffer(std::span<f32>
}
else if (sanitize)
{
#if defined(ARCH_X64)
//Convert NaNs and Infs to 0
const auto masked = _mm_and_si128(shuffled_vector, _mm_set1_epi32(0x7fffffff));
const auto valid = _mm_cmplt_epi32(masked, _mm_set1_epi32(0x7f800000));
const auto result = _mm_and_si128(shuffled_vector, valid);
_mm_stream_si128(utils::bless<__m128i>(dst), result);
#else
for (u32 i = 0; i < 4; i++)
{
const u32 value = std::bit_cast<u32>(tmp[i]);
tmp[i] = (value & 0x7fffffff) < 0x7f800000 ? value : 0;
}
std::memcpy(dst, tmp, 16);
#endif
}
else
{
#if defined(ARCH_X64)
_mm_stream_si128(utils::bless<__m128i>(dst), shuffled_vector);
#else
std::memcpy(dst, tmp, 16);
#endif
}
dst += 4;