64-bit client random crashes - caused by shader compiler caching?

generaldisaster

I have been getting random (and I do mean random) crashes from the 64-bit client for quite some time. and now that I have a number of logs built up from each time STO crashed, I have been looking for differences and commonalities. the one thing I have found so far, is makeshaderbins log file. (..\startrek online\startrek online\live\logs) everytime the game game crashes it stops at the same point in what I am assuming is compiling the shaders. I will post the last few lines of the log below. it looks like something gets cutoff (by the crash). it looks like STO always runs into trouble when it reaches 308/325. I don't see anything marked error or fail in the log. I have verified, and verified, and verified my sto files till the cows came home, and not once has anything been re-downloaded, or fixed. the crashes keep occurring. naturally when we start talking about shaders and so on, one common suggestion is to reinstall the video drivers, which I have done. I am using the latest released drivers for my video card (dated : 2016.12.4 v. 342.01)


181221 02:11:31 309 Client[0 [NOENT]]: Compiling shader template Materials/Templates/Character/Emblem_Master_Character_Template_01.Material (296/325)
181221 02:11:31 310 Client[0 [NOENT]]: Compiling shader template Materials/Templates/Character/Diff_Ref_Fresnel_Template.Material (297/325)
181221 02:11:31 311 Client[0 [NOENT]]: Compiling shader template Materials/Templates/Character/Diff_Ref_Fresnel_Alphatocov_01.Material (298/325)
181221 02:11:31 312 Client[0 [NOENT]]: Compiling shader template Materials/Templates/Character/Consoletest_Ship_Master_Template_Nx_Gloss_Vertex_Color_Multiply_01.Material (299/325)
181221 02:11:31 313 Client[0 [NOENT]]: Compiling shader template Materials/Templates/Character/Consoletest_Character_Master_Fresnel_Template_Nx_01_Fresreflect.Material (300/325)
181221 02:11:31 314 Client[0 [NOENT]]: Compiling shader template Materials/Templates/Character/Character_Skin_Facebits_Template_Nx_2nd_Uv.Material (301/325)
181221 02:11:31 315 Client[0 [NOENT]]: Compiling shader template Materials/Templates/Character/Character_Skin_Facebits_Template_Nx_01.Material (302/325)
181221 02:11:31 316 Client[0 [NOENT]]: Compiling shader template Materials/Templates/Character/Character_Skin_Earblend_Template_01.Material (303/325)
181221 02:11:31 317 Client[0 [NOENT]]: Compiling shader template Materials/Templates/Character/Character_Skin_Earblend_01.Material (304/325)
181221 02:11:31 318 Client[0 [NOENT]]: Compiling shader template Materials/Templates/Character/Character_Skin_Body_Template_02.Material (305/325)
181221 02:11:31 319 Client[0 [NOENT]]: Compiling shader template Materials/Templates/Character/Character_Skin_Body_Template_01.Material (306/325)
181221 02:11:31 320 Client[0 [NOENT]]: Compiling shader template Materials/Templates/Character/Character_Pattern_Template_01.Material (307/325)
181221 02:11:31 321 Client[0 [NOENT]]: Compiling shader template Materials/Templates/Character/Character_M

generaldisaster

I may have found a way around the issue. though I don't know why it seems to work, or what the option does or how it would effect STO. I went into the NVidia control panel, and went to manage 3D settings and turned shader cache to off.

generaldisaster

Since disabling this option, STO has been completely stable (x64 client only). of course STO is running a bit slower than it was previously, but it is no longer crashing. Load times between maps have gone up too. during the elachi alert event, by the time loading of the map was finished, everyone had already started the event, but me. and the loading of Q's wonderland is horrible.

generaldisaster

HA HA HA....... Looks like I really jinxed it.

STO is back to crashing when I goes to compile the same shader that was causing issues once before.

generaldisaster

well, now I can not even get past the toon selection screen even when I turn shader caching back on. I select my toon, and it looks like I will get in, but a few seconds into the load, the game crashes, and brings up the cryptic error reporter.

frtoaster

Some searching reveals that Nvidia keeps a shader cache in C:\Windows\Temp or %TEMP%. It seems that STO keeps its own shader cache in Live\cache. You could try deleting either the Nvidia cache or the STO cache. I have no idea if that will fix your issue or make it worse, so you do so at your own risk.

https://forums.geforce.com/default/topic/805794/is-there-a-way-to-clear-refresh-the-nvidia-shader-cache-/
https://forums.gearboxsoftware.com/t/nvidia-cards-and-the-shader-cache-in-windows/1573020
https://www.reddit.com/r/StarWarsBattlefront/comments/8gbwue/psa_delete_nvidia_cache_to_remove_random_stutters/
https://forums.guru3d.com/threads/nvidis-shader-cache-still-again.418111/

generaldisaster

thx, will try that.

when I enabled the -console at startup for both the launcher and the game, this is the last thing displayed when the game crashes, it may or may not be related in anyway to the shaders.

ERROR: devassert: handle->debug_in_file_access==1;Shouldn't be in the file critical section if we're
going to try and delete a file
Line: c:\src\libs\UtilitiesLib\file\hoglib.c(723)

don't know what the error means, but I want to find out....

sanokskyrat

Did you find out what it meant? I keep getting the "ERROR: devassert: handle->debug_in_file_access==1;Shouldn't be in the file critical section if we're going to try and delete a file" message as well.