SVN : Pcsx2 : r5287

Pcsx2 is an opensource Playstation 2 emulator for Windows and Linux. It is probably the best playstation2 emulator out there, and can run most of the commercial games. Now it has released a 1.0 version after lots of years of great work and development.
r5287 Changelog:
GSDX: made the paletted texture handling in the cache a bit more sensible to my eyes and implemented interpolation of palette entries for pixels in paletted textures when using the 8 bit textures option.  Regressions in some games I think, such as Virtual On (which is very broken anyway), need to investigate what made them work (to some degree) before. Seems to change some performance characteristics favourably to my surprise, but I might just be bad at remembering framerates.
On previous commits:
GSDX: Err, and another file, I should be asleep.

GSDX: Forgot a file.

GSDX: Added some exceptions on shader compilation failure (with handlers in entry points because people seem to like the plugin struggling on even when nothing works any more) because I am not keen crashing Windows via my graphics driver.

GSDX: Prodded some offensive code, this isn't meant to affect emulation of any games and it probably won't.

GSDX: put all paletted texture samples through the same transformation from the UNORM 0-1 256 step mapping to a 0.5/256-255.5/256 mapping after seeing potential bugs with FMT_4HL and changed FMT_4HH because it seemed to be completely wrong. Nothing but 8 bit textures tested because I can't find a single game which uses 4 bit textures.

GSDX: Fixed my inability to remember or look up (I swear that page was hiding) coordinate systems in r5279 and assumed that it was a rounding error instead. The symptom was that palette indices above around 250 were being rounded up to the next palette entry causing visible glitches (only in D3D10 for some reason). Changed the code to keep the fractional part after multiplication by 256 around 0.5 and the steps around 1.0. Should be very safe against small errors.

GSDX: Probably fix D3D10 and maybe D3D9 (might be working anyway but I think I have it addressing texel centres now) palette lookups. I noticed that this was broken in D3D10 while fixing the Realta Nua issue in r5273.

Explanation, because this gives me a headache and this might save someone else one (or I might be wrong and they might see why): in D3D10, 0.0 points to the centre of the leftmost texel and 1.0 points one texel to the right of the rightmost texel, so to map a UNORM uniformly across a texel we need to multiply the input by (w-1)/w. In D3D9 0.0 points to the left edge of the leftmost texel and 1.0 to the right edge of the rightmost texel so after the multiplication we add 1/2w.

Actual texture sampling is probably not right for at least one of D3D9 and D3D10, but this headache is killing me.

* add a script to run cg compiler on glsl file:
+ handy to check the syntax
+ output the asm of the shader
- unfortunately don't support latest glsl construct but better than nothing
* really delete resources before context destruction
* wanted to play with opengl3 timer for profiling but not conclusive, just keeping code around for future use

Revert original fix for Realta Nua as Pseudonym found the proper fix. This should get rid of the regression on Agressive Inline and one of the Matt Hoffman games having textureless graphics.

GSdx: CRC hacks: ICO: moved the cutie extra hack from r5214 to "Aggressive mode".
Download: SVN : Pcsx2 : r5287
Source: This


Post a Comment

Copyright ©