Hi All;

I have been doing computing for GPUGrid for a couple years. Recently I wanted to up my contribution and purchased 3 Nvidia GTX 770 units. One is in a Windows 8 64 workstation and runs pretty well, but I do see that the assigned GPU in the boinc manager will toggle from showing the gpu number to not. It complets a long task an about 12 hours.

However, I built a brand new Win 7 Pro 64 workstation to use, and to crunch with. It has 2 GTX 770 units. The GPUGrid tasks constantly stop and start, and very little if any job progress is made. I have tired every driver version that has come out in the last 6 months to no avail. If I switch to Primegrid they run just fine. Is there any info that I could send in to help troubleshoot what the problem is?

Herb

Info to send in... the first 20 or so lines of the BOINC Event Log is a good start and sometimes all one needs. From the point where BOINC client has just started down to where it starts contacting projects is usually sufficient. If that portion has already scrolled out of view then you need to restart BOINC. Please do not post the entire Event Log here.

Have you noticed any error messages in Event Log? Granted they may be burried and hard tyo spot but if you fine a line that starts with "GPUgrid:, highlight it and then click "Show only this project" it helps.

Another easy way to locate error messages is to open the stdoutdae.txt file in the BOINC data directory in a text editor and search for words like "error", "not found", "absent" and "missing" since those are terms that often appear in error messages.

Just a wild guess here... are the pair on the Win7 machine in an SLI setup? I seem to recall someone saying using SLI can cause problems but unless it's easy to turn SLI on and off or uninstall/reinstall I wouldn't advise doing so until you hear from one of the Windows users (Linux here) because SLI might not be a problem at all.
____________
BOINC <<--- credit whores, pedants, alien hunters

Thanks so much for the offer to help. Nothing at all in the win event logs. Here are the log snippets for you. Not running in SLI

12/20/2013 1:12:01 PM | | cc_config.xml not found - using defaults
12/20/2013 1:12:01 PM | | Starting BOINC client version 7.2.33 for windows_x86_64
12/20/2013 1:12:01 PM | | log flags: file_xfer, sched_ops, task
12/20/2013 1:12:01 PM | | Libraries: libcurl/7.25.0 OpenSSL/1.0.1 zlib/1.2.6
12/20/2013 1:12:01 PM | | Data directory: C:\BOINC\Data
12/20/2013 1:12:01 PM | | Running under account Herb
12/20/2013 1:12:01 PM | | CUDA: NVIDIA GPU 0: GeForce GTX 770 (driver version 327.23, CUDA version 5.5, compute capability 3.0, 2048MB, 1936MB available, 3332 GFLOPS peak)
12/20/2013 1:12:01 PM | | CUDA: NVIDIA GPU 1: GeForce GTX 770 (driver version 327.23, CUDA version 5.5, compute capability 3.0, 2048MB, 1936MB available, 3332 GFLOPS peak)
12/20/2013 1:12:01 PM | | OpenCL: NVIDIA GPU 0: GeForce GTX 770 (driver version 327.23, device version OpenCL 1.1 CUDA, 2048MB, 1936MB available, 3332 GFLOPS peak)
12/20/2013 1:12:01 PM | | OpenCL: NVIDIA GPU 1: GeForce GTX 770 (driver version 327.23, device version OpenCL 1.1 CUDA, 2048MB, 1936MB available, 3332 GFLOPS peak)
12/20/2013 1:12:01 PM | | OpenCL CPU: AMD FX(tm)-8350 Eight-Core Processor (OpenCL driver vendor: Advanced Micro Devices, Inc., driver version 2.0, device version OpenCL 1.1 AMD-APP (873.1))
12/20/2013 1:12:01 PM | | Host name: Herb-PC
12/20/2013 1:12:01 PM | | Processor: 8 AuthenticAMD AMD FX(tm)-8350 Eight-Core Processor [Family 21 Model 2 Stepping 0]
12/20/2013 1:12:01 PM | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 htt pni ssse3 fma cx16 sse4_1 sse4_2 popcnt aes syscall nx lm svm sse4a osvw ibs xop skinit wdt lwp fma4 tce tbm topx page1gb rdtscp
12/20/2013 1:12:01 PM | | OS: Microsoft Windows 7: Professional x64 Edition, Service Pack 1, (06.01.7601.00)
12/20/2013 1:12:01 PM | | Memory: 8.00 GB physical, 16.00 GB virtual
12/20/2013 1:12:01 PM | | Disk: 223.47 GB total, 176.30 GB free
12/20/2013 1:12:01 PM | | Local time is UTC -6 hours
12/20/2013 1:12:01 PM | GPUGRID | URL http://www.gpugrid.net/; Computer ID 164651; resource share 100
12/20/2013 1:12:01 PM | malariacontrol.net | URL http://www.malariacontrol.net/; Computer ID 733134; resource share 25
12/20/2013 1:12:01 PM | PrimeGrid | URL http://www.primegrid.com/; Computer ID 419253; resource share 10
12/20/2013 1:12:01 PM | | General prefs: from http://registro.ibercivis.es/ (last modified 10-Dec-2013 13:24:04)
12/20/2013 1:12:01 PM | | Host location: none
12/20/2013 1:12:01 PM | | General prefs: using your defaults
12/20/2013 1:12:01 PM | | Reading preferences override file
12/20/2013 1:12:01 PM | | Preferences:
12/20/2013 1:12:01 PM | | max memory usage when active: 4095.59MB
12/20/2013 1:12:01 PM | | max memory usage when idle: 7372.06MB
12/20/2013 1:12:01 PM | | max disk usage: 10.00GB
12/20/2013 1:12:01 PM | | max CPUs used: 4
12/20/2013 1:12:01 PM | | (to change preferences, visit a project web site or select Preferences in the Manager)
12/20/2013 1:12:01 PM | | Not using a proxy

And

in the other log, the word "absent" is absent :). the only reference to not found relates to the fact that there is no cc_config.sml files. no errors related to gpu grid.

Like I said, you see the tasks running, and the (device 0) or (device 1) appears and disappears every second or so, and no job progress is made.

Thanks,

Herb

Actually part of the mystery is solved. When looking at the logs, I see that it is getting the preferences from a new project I joined. 12/20/2013 1:12:01 PM | | General prefs: from http://registro.ibercivis.es/ (last modified 10-Dec-2013 13:24:04)

It defaulted to stopping computation when the cpu was at 25% even tho that is not the setting on this machine in the computing options. I set that to 95% and then the tasks run for a bit before stopping and progress is being made.

However, I think that this problem is a new one with the new client. Anyone else seeing this?

Herb

Dear Herb,

I am also running 2*GTX770 - GPUGRID from time to time.
In order to avoid any kind of problems, I have attached
a short app_config.xml to the project so that each WU uses 1 full CPU.

At least for the Long Run's, as I never found the
ad-hoc appli name for the short one.

I use this config not only together with AMD CPU,
which suffer "a lot" when BOINC uses > CPU as the
physical "reality" :
have a Phenom II 955 + GTX660Ti and if CPU usage > 4 then slow.

=> Using this config also on the i7-4770K

In case you want me to send you the config. file, don't hesitate to send me a PM.

Hope this helps

Kind regards,

Philippe