GPU doesnt load up next project

[Lohnewolf]

I am having some issues with my GPU, in the sense that it doesn't seem to ever automatically start another WU. It downloads a WU, starts trying to run it, and seemingly does this about five times before "failing". I then have to manually pause and start the folding process again. Sometimes I have to do this more than three times for one to start. Does anybody have any idea why this could be happening? The CPU seemed to automatically start the next one over night. I do apologise if this is in the wrong thread etc, I am new to the forum so will do my best to come to speed with everything as quickly as possible.

Thanks in advance,

[JimboPalmer]

We lack details like what GPU you are running, what OS, which client you are using, which core is failing, and your hardware specs.
Many of these are included in the first 100 lines of your log.

[Joe_H]

Welcome to the folding support forum.

If you could post the beginning section of your log file, that would be useful in figuring out what is happening. Directions for locating and posting log files can be found in the Welcome topic.

However a common recent problem posted both here and in the Pande Group blog is related to updated GPU drivers for nVidia video cards. If you have one and recently updated your drivers to any of the version 375 drivers, they cause folding to fail on any GPU WU that uses Core_21. The topic where this is being followed is here.

Currently the only workaround is to revert to version 373 or earlier drivers.

[Lohnewolf]

I do apologise. I suppose you learn something new everyday! Unfortunately I am not sure how to put them into quotes correctly (I'm assuming this site uses html code?) so I will post here what I am running:

Code: Select all

*********************** Log Started 2016-11-23T08:41:00Z ***********************
08:41:00:************************* Folding@home Client *************************
08:41:00:      Website: http://folding.stanford.edu/
08:41:00:    Copyright: (c) 2009-2014 Stanford University
08:41:00:       Author: Joseph Coffland <joseph@cauldrondevelopment.com>
08:41:00:         Args: 
08:41:00:       Config: C:/Users/matty/AppData/Roaming/FAHClient/config.xml
08:41:00:******************************** Build ********************************
08:41:00:      Version: 7.4.4
08:41:00:         Date: Mar 4 2014
08:41:00:         Time: 20:26:54
08:41:00:      SVN Rev: 4130
08:41:00:       Branch: fah/trunk/client
08:41:00:     Compiler: Intel(R) C++ MSVC 1500 mode 1200
08:41:00:      Options: /TP /nologo /EHa /Qdiag-disable:4297,4103,1786,279 /Ox -arch:SSE
08:41:00:               /QaxSSE2,SSE3,SSSE3,SSE4.1,SSE4.2 /Qopenmp /Qrestrict /MT /Qmkl
08:41:00:     Platform: win32 XP
08:41:00:         Bits: 32
08:41:00:         Mode: Release
08:41:00:******************************* System ********************************
08:41:00:          CPU: Intel(R) Core(TM) i7-6800K CPU @ 3.40GHz
08:41:00:       CPU ID: GenuineIntel Family 6 Model 79 Stepping 1
08:41:00:         CPUs: 12
08:41:00:       Memory: 31.84GiB
08:41:00:  Free Memory: 28.57GiB
08:41:00:      Threads: WINDOWS_THREADS
08:41:00:   OS Version: 6.2
08:41:00:  Has Battery: false
08:41:00:   On Battery: false
08:41:00:   UTC Offset: 0
08:41:00:          PID: 692
08:41:00:          CWD: C:/Users/matty/AppData/Roaming/FAHClient
08:41:00:           OS: Windows 10 Pro
08:41:00:      OS Arch: AMD64
08:41:00:         GPUs: 2
08:41:00:        GPU 0: NVIDIA:5 GP104 [GeForce GTX 1080]
08:41:00:        GPU 1: UNSUPPORTED: NV3 [PCI]
08:41:00:         CUDA: 6.1
08:41:00:  CUDA Driver: 8000
08:41:00:Win32 Service: false

As you said Joe_H, I do indeed have a NVIDIA GPU running 375.xx update, which one I am not too sure without looking (I will have a look at the thread you have linked in your post).

As far as I can tell, each time a WU tries to run, the following sort of log is printer, each time with the error code BAD_WORK_UNIT. I wonder if this is relevant to the GPU issue, or if it is something else as well as the faulty driver? Ill have a go at reverting to the older driver and see what happens.

Code: Select all

19:40:15:WU02:FS01:0x21:*********************** Log Started 2016-11-23T19:40:15Z ***********************
19:40:15:WU02:FS01:0x21:Project: 13110 (Run 31, Clone 0, Gen 602)
19:40:15:WU02:FS01:0x21:Unit: 0x000001b3ab436c65577170b558aa5c1d
19:40:15:WU02:FS01:0x21:CPU: 0x00000000000000000000000000000000
19:40:15:WU02:FS01:0x21:Machine: 1
19:40:15:WU02:FS01:0x21:Reading tar file core.xml
19:40:15:WU02:FS01:0x21:Reading tar file integrator.xml
19:40:15:WU02:FS01:0x21:Reading tar file state.xml
19:40:16:WU02:FS01:0x21:Reading tar file system.xml
19:40:17:WU02:FS01:0x21:Digital signatures verified
19:40:17:WU02:FS01:0x21:Folding@home GPU Core21 Folding@home Core
19:40:17:WU02:FS01:0x21:Version 0.0.17
19:40:42:WU02:FS01:0x21:ERROR:exception: Error downloading array interactionCount: clEnqueueReadBuffer (-5)
19:40:42:WU02:FS01:0x21:Saving result file logfile_01.txt
19:40:42:WU02:FS01:0x21:Saving result file log.txt
19:40:42:WU02:FS01:0x21:Folding@home Core Shutdown: BAD_WORK_UNIT

In the meantime, just in case the help given doesn't help my particular case, I thought I would put this information up (I hope it is enough? Don't hesitate to ask for more if I am still far too vague with the question).

Mod edit: added Code tags for the two log sections - j

[Lohnewolf]

(Apologies, I am practising with posting techniques whilst sending this message)

Code: Select all

I do apologise, I have since read up on your pages Joe_H and I see that what I was asking is in there, so I will have a play around. The strange thing I find, is that I have actually managed to run some WU's on my GPU, and now it has seemingly completely stopped running them. I do apologise for not doing things as is expected on here (I remember it was a little frustrating seeing new people post on gaming forums with incorrect syntax when I used to play on some online games). I do plan on getting up to speed as soon as possible.

[Joe_H]

Code: Select all

19:40:42:WU02:FS01:0x21:ERROR:exception: Error downloading array interactionCount: clEnqueueReadBuffer (-5)

This error reported in the log is the usual one seen with any of the version 375 drivers released so far. This version broke the use of one of the OpenCL functions used in the newer GPU folding core, Core_21. The reason you could have seen some GPU WU's processed successfully is that there are some WU's still being assigned for older GPU projects that use an older core, Core_18. That core will work with the newer driver as it does not use that particular OpenCL function, but there is currently no way to assign just those older projects based on which driver version you are running.

Persons have reported here that the latest driver that works is version 373.06. Also reported as working are ones from version 372.

As a hint, this forum gives you a quick posting window with the option to use the Full Editor at the bottom. That allows you to select the commonly used tags such as Quote, Code and the like from a menu at the top of the posting window.