possible RSX regression

[thenewcomer]

everytime i start tales of xillia and get to the character select screen, i get a very specific error that i did not get from an older rpcs3 build. looks like a simple syntax error but i have no idea how to fix it myself.

E {rsx::thread} RSX: Failed to compile shader: 0(75) : error C0000: syntax error, unexpected '.', expecting ')' at token "."

F {rsx::thread} class gl::glsl::link_exception thrown: linkage failed: 'Fragment info
-------------
0(75) : error C0000: syntax error, unexpected '.', expecting ')' at token "."
(0) : error C2003: incompatable options for link
'

link to the log files HERE

[ssshadow]

Yes that happens with OpenGL. DX12 and maybe Vulkan should work, can you try those?

[thenewcomer]

Yes that happens with OpenGL. DX12 and maybe Vulkan should work, can you try those?

tried with Vulkan since im not on win10, and didnt even get that far. again looks like a syntax error somewhere. added more log files to previous link in case you need them

E {rsx::thread} RSX: Src type 3 used, please report this to a developer.
E {rsx::thread} RSX: Src type 3 used, please report this to a developer.
E {rsx::thread} RSX: Warning, version 420 is not yet complete; most version-specific features are present, but some are missing.
ERROR: 0:67: '' : syntax error
ERROR: 1 compilation errors. No code generated.


E {rsx::thread} RSX:
E {PPU[0x70000000] Thread (main_thread) [0x00f59734]} sys_memory: sys_memory_get_page_attribute(addr=0xd000f510, attr=*0xd000f3c0)
F {rsx::thread} class std::runtime_error thrown: Failed to compile fragment shader
(in file C:\rpcs3\rpcs3\Emu\RSX\VK\VKFragmentProgram.cpp:475)

[ssshadow]

Ok so the game used to work and now has some strange errors. If you can find exactly the version that broke it I can file a bug report with a renderdoc trace and so on.

You can find all old versions here: https://ci.appveyor.com/project/rpcs3/rpcs3/history

Find some old version that works. Then download a build from in-between that old working version and today. See if it works or not, and then keep narrowing it down until you know exactly where it stopped working. As you halve the number of possibilities in each steps you will find the first bad version surprisingly fast.