Gain: 1% at 4x on SotC (it partially compensates recent additions)
When the color is constant and equal to 128, the MODULATE mode is
equivalent to the DECAL mode. It saves 5 instructions on the FS.
Accurate options do a better jobs. Technically it can still
be useful for old gpu/driver that doesn't support the GL4.5 extension.
On Windows, you can still rely on Dx
On linux, free driver support it (except Intel)
It fixes the bad light on Silent Hill with the SW renderer.
Full story
if Q is NaN, m_vt.m_eq.q becomes wrongly true
/Q will wrongly be optimized in the "Vertex Shader" of the SW
Note: Add an assert for the STQ handler
Code path is quite hot so no need to add extra check for nothing
GS uses integer value and does integer operation too.
This commit trunc the sampled texture, the interpoled fragment color
and the product of the 2.
It impacts negatively the perf of about 3/4% (GPU) but it fixes rendering on
suikoden and potentially some others games too.
Code was completey bitrotten
Code was a partial test (and yet 500 lines already)
Shader is more and more complex and multithreading support greatly
reduce the cost of shader switch
Nvidia allows to get the ASM of the shader of the compiled shader. It is useful
to check the performance.
It also allow me to compile most of shader code path for QA
Dump is enabled in linux replayer + debug_glsl_shader = 2
This means that we are now no longer touching files that haven't technically been written to. Some games use timestamp information to automatically highlight the save that was last written to, so this should fix a small but annoying bug where it would highlight the wrong one.
Do note that while there is a much simpler check that looks like this:
// Remove (== don't flush) all memory card pages that haven't actually changed.
for ( auto oldIt = m_oldDataCache.begin(); oldIt != m_oldDataCache.end(); ++oldIt ) {
auto newIt = m_cache.find( oldIt->first );
assert( newIt != m_cache.end() ); // if this isn't true something broke somewhere, the two maps should always contain the same pages
if ( memcmp( &oldIt->second.raw[0], &newIt->second.raw[0], PageSize ) == 0 ) {
m_cache.erase( newIt );
}
}
m_oldDataCache.clear();
It can fail in edge cases that don't actually seem too unlikely. Imagine a save being deleted, and then a new save from the same game but in a different slot being created quickly afterwards. It seems quite possible that the new save's file data then occupies the exact same pages as the old save's, and since it's from the same game it might be close enough to where a page sized section (juse 0x200 bytes!) matches the data from the old save that previously resided in that location -- which would cause this code to throw away and not flush this data! It's a shame too, since this variant would be a few ms faster as well, but I feel it's better to be safe than sorry here.
This mainly means that the superblock is now no longer written every single time the memory card is closed, but only when it's changed (which should be exactly once, when the memory card is formatted). It also means that you can format a memory card and then have the emulator crash later without having to reformat the card next time.
We're not actually deleting files though, we just rename them to prepend _pcsx2_deleted_, in case something breaks or whatever, so the user can in an emergency just restore the save by removing that part of the filename.
Gregory Hainaut