RPCS3 Forums

It must have happened, metal slug on the desk.

Game crashes but wants to show something to us.

No crash anymore. Not sure if thats changing anything. Black screen. Some garbled textres in rsx debugger, last texture in screenshot.

Another log. Game goes quite far. But i see black screen. Maybe would be better with dx12

Game finally display picture. Until the graphic error: ·F {rsx::thread} Exception: compilation failed: 'Validation failed! - Different sampler types for same sample texture unit in vertex shader. >> This is what is logged after closing the emu. Im sure it can go further when fixed.

Log is with writecolor buffer enabled, but it works with this option off.

LLE Sar, Spurs jq, dec jpg, dec png, sail,

ppu2/spu2

Last screen is where game stopped.

(02-20-2016, 10:05 AM)ps0ne Wrote: [ -> ]Game finally display picture. Until the graphic error: ·F {rsx::thread} Exception: compilation failed: 'Validation failed! - Different sampler types for same sample texture unit in vertex shader. >> This is what is logged after closing the emu. Im sure it can go further when fixed.

Log is with writecolor buffer enabled, but it works with this option off.

LLE Sar, Spurs jq, dec jpg, dec png, sail,

ppu2/spu2

Last screen is where game stopped.

Are you using the latest master? This issue should have been fixed a while back. Also, it only affects AMD graphics. You can test on Nvidia hardware and see if it still happens.

(02-20-2016, 01:48 PM)kd-11 Wrote: [ -> ]

(02-20-2016, 10:05 AM)ps0ne Wrote: [ -> ]Game finally display picture. Until the graphic error: ·F {rsx::thread} Exception: compilation failed: 'Validation failed! - Different sampler types for same sample texture unit in vertex shader. >> This is what is logged after closing the emu. Im sure it can go further when fixed.

Log is with writecolor buffer enabled, but it works with this option off.

LLE Sar, Spurs jq, dec jpg, dec png, sail,

ppu2/spu2

Last screen is where game stopped.

Are you using the latest master? This issue should have been fixed a while back. Also, it only affects AMD graphics. You can test on Nvidia hardware and see if it still happens.

Could be memory corruption or something.

I use master, ofc. I have AMD card. I cannot test on nvidia.

This error was not present earlier. I will see what can i find more about this.

(02-20-2016, 11:00 PM)ps0ne Wrote: [ -> ]I use master, ofc. I have AMD card. I cannot test on nvidia.

This error was not present earlier. I will see what can i find more about this.

You can try find PR/commit that caused this issue by bisecting.

My findings:
All commits by commit date:
commit 1501 - very first commit that makes game visible, error with message
commit 1500 - last commit with black screeen, error with message
commit 1477 - error with message
commit 1446 - early error, no message about vertex
commit 1428 - early error, no message about vertex
commit 1415 - game continue even after errors
commit 1417 - game continue even after errors


Not sure if anything is useful, because depend of commit error appears at he beginnig, or later as on screen above. Maybe be problem with amd drivers too. Just dont care, maybe that error was from begging in this game.

(02-21-2016, 02:14 AM)ps0ne Wrote: [ -> ]My findings:
All commits by commit date:
commit 1501 - very first commit that makes game visible, error with message
commit 1500 - last commit with black screeen, error with message
commit 1477 - error with message
commit 1446 - early error, no message about vertex
commit 1428 - early error, no message about vertex
commit 1415 - game continue even after errors
commit 1417 - game continue even after errors


Not sure if anything is useful, because depend of commit error appears at he beginnig, or later as on screen above. Maybe be problem with amd drivers too. Just dont care, maybe that error was from begging in this game.

Probably malformed vertex shader, or an uninitialized sampler in the vertex shader. You can uncomment the line m_program->validate() in GLGSRender.cpp, function end() and see if it displays correctly. I'll add an option to relax the strict requirements by the AMD driver to the GUI when I can.

Line to comment out:
https://github.com/RPCS3/rpcs3/blob/mast...Lundefined