Advanced search

Message boards : Graphics cards (GPUs) : How to troubleshoot a new card?

Author Message
Profile caffeineyellow5
Avatar
Send message
Joined: 30 Jul 14
Posts: 225
Credit: 2,658,976,345
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwatwatwatwat
Message 41043 - Posted: 6 May 2015 | 22:38:52 UTC
Last modified: 6 May 2015 | 23:15:34 UTC

I had a 780 go bad on me and took it back and traded in for a 980. I got it home and it continued the task that the old 780 was in the middle of, which I thought was odd, but fine. It ran a few minutes. I stopped it and rebooted. After the reboot, the the OS, MSI Afterburner, and all other things I can think of can see the card that seems to be working fine, but the task on BOINC was showing an error on the app like "Task not running. No GPU Available." I aborted the task, rebooted, chacked the NVIDIA GeForce Experience application for updates, etc. I was even able to turn off the LED light on the new card. I then uninstalled BOINC and manually deleted the BOINC folder in ProgramData (Windows 8.1), rebooted, and reinstalled.

This is the machine in question:
https://www.gpugrid.net/show_host_detail.php?hostid=198084

Below is the log with a few triggers set that I thought may be helpful.

5/6/2015 6:05:11 PM | | Starting BOINC client version 7.4.42 for windows_x86_64
5/6/2015 6:05:11 PM | | log flags: file_xfer, sched_ops, task, coproc_debug, sched_op_debug, slot_debug
5/6/2015 6:05:11 PM | | log flags: statefile_debug, task_debug, work_fetch_debug
5/6/2015 6:05:11 PM | | Libraries: libcurl/7.39.0 OpenSSL/1.0.1j zlib/1.2.8
5/6/2015 6:05:11 PM | | Running as a daemon
5/6/2015 6:05:11 PM | | Data directory: C:\ProgramData\BOINC
5/6/2015 6:05:11 PM | | Running under account boinc_master
5/6/2015 6:05:11 PM | | No usable GPUs found
5/6/2015 6:05:11 PM | | [statefile] set dirty: Set mode
5/6/2015 6:05:11 PM | | [statefile] set dirty: Set mode
5/6/2015 6:05:11 PM | | [statefile] set dirty: Set mode
5/6/2015 6:05:11 PM | | Host name: Office-PC
5/6/2015 6:05:11 PM | | Processor: 8 GenuineIntel Intel(R) Core(TM) i7-4790K CPU @ 4.00GHz [Family 6 Model 60 Stepping 3]
5/6/2015 6:05:11 PM | | Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss htt tm pni ssse3 fma cx16 sse4_1 sse4_2 movebe popcnt aes f16c rdrandsyscall nx lm avx avx2 vmx tm2 pbe fsgsbase bmi1 hle smep bmi2
5/6/2015 6:05:11 PM | | OS: Microsoft Windows 8.1: Professional x64 Edition, (06.03.9600.00)
5/6/2015 6:05:11 PM | | Memory: 15.94 GB physical, 21.44 GB virtual
5/6/2015 6:05:11 PM | | Disk: 476.59 GB total, 423.29 GB free
5/6/2015 6:05:11 PM | | Local time is UTC -4 hours
5/6/2015 6:05:11 PM | | VirtualBox version: 5.0.0_BETA2
5/6/2015 6:05:11 PM | | Config: report completed tasks immediately
5/6/2015 6:05:11 PM | | Config: use all coprocessors
5/6/2015 6:05:11 PM | | [statefile] Writing state file
5/6/2015 6:05:11 PM | | [statefile] Done writing state file
5/6/2015 6:05:11 PM | GPUGRID | URL http://www.gpugrid.net/; Computer ID 198084; resource share 100
5/6/2015 6:05:11 PM | GPUGRID | General prefs: from GPUGRID (last modified 21-Mar-2015 03:56:21)
5/6/2015 6:05:11 PM | GPUGRID | Computer location: home
5/6/2015 6:05:11 PM | GPUGRID | General prefs: no separate prefs for home; using your defaults
5/6/2015 6:05:11 PM | | Reading preferences override file
5/6/2015 6:05:11 PM | | Preferences:
5/6/2015 6:05:11 PM | | max memory usage when active: 16326.86MB
5/6/2015 6:05:11 PM | | max memory usage when idle: 16326.86MB
5/6/2015 6:05:11 PM | | max disk usage: 418.29GB
5/6/2015 6:05:11 PM | | (to change preferences, visit a project web site or select Preferences in the Manager)
5/6/2015 6:05:11 PM | | [work_fetch] Request work fetch: Prefs update
5/6/2015 6:05:11 PM | | [work_fetch] Request work fetch: Startup
5/6/2015 6:05:11 PM | | [statefile] set dirty: init
5/6/2015 6:05:11 PM | | Not using a proxy
5/6/2015 6:05:11 PM | | [work_fetch] ------- start work fetch state -------
5/6/2015 6:05:11 PM | | [work_fetch] target work buffer: 180.00 + 21600.00 sec
5/6/2015 6:05:11 PM | | [work_fetch] --- project states ---
5/6/2015 6:05:11 PM | GPUGRID | [work_fetch] REC 0.000 prio -0.000 can request work
5/6/2015 6:05:11 PM | | [work_fetch] --- state for CPU ---
5/6/2015 6:05:11 PM | | [work_fetch] shortfall 174240.00 nidle 8.00 saturated 0.00 busy 0.00
5/6/2015 6:05:11 PM | GPUGRID | [work_fetch] share 1.000
5/6/2015 6:05:11 PM | | [work_fetch] ------- end work fetch state -------
5/6/2015 6:05:11 PM | GPUGRID | [work_fetch] set_request() for CPU: ninst 8 nused_total 0.00 nidle_now 8.00 fetch share 1.00 req_inst 8.00 req_secs 174240.00
5/6/2015 6:05:11 PM | GPUGRID | [sched_op] Starting scheduler request
5/6/2015 6:05:11 PM | GPUGRID | [work_fetch] request: CPU (174240.00 sec, 8.00 inst)
5/6/2015 6:05:11 PM | GPUGRID | Sending scheduler request: To fetch work.
5/6/2015 6:05:11 PM | GPUGRID | Requesting new tasks for CPU
5/6/2015 6:05:11 PM | GPUGRID | [sched_op] CPU work request: 174240.00 seconds; 8.00 devices
5/6/2015 6:05:11 PM | | [statefile] Writing state file
5/6/2015 6:05:11 PM | | [statefile] Done writing state file
5/6/2015 6:05:12 PM | | [statefile] set dirty: set_language
5/6/2015 6:05:12 PM | | [statefile] Writing state file
5/6/2015 6:05:12 PM | | [statefile] Done writing state file
5/6/2015 6:05:13 PM | GPUGRID | Scheduler request completed: got 0 new tasks
5/6/2015 6:05:13 PM | GPUGRID | [sched_op] Server version 613
5/6/2015 6:05:13 PM | GPUGRID | No tasks sent
5/6/2015 6:05:13 PM | GPUGRID | No tasks are available for Long runs (8-12 hours on fastest card)
5/6/2015 6:05:13 PM | GPUGRID | Project requested delay of 31 seconds
5/6/2015 6:05:13 PM | GPUGRID | [slot] linked projects/www.gpugrid.net/logogpugrid.png to projects/www.gpugrid.net/stat_icon
5/6/2015 6:05:13 PM | GPUGRID | [slot] linked projects/www.gpugrid.net/project_1.png to projects/www.gpugrid.net/slideshow_ga_00
5/6/2015 6:05:13 PM | GPUGRID | [slot] linked projects/www.gpugrid.net/project_1.png to projects/www.gpugrid.net/slideshow_cellmd_00
5/6/2015 6:05:13 PM | GPUGRID | [slot] linked projects/www.gpugrid.net/project_2.png to projects/www.gpugrid.net/slideshow_ga_01
5/6/2015 6:05:13 PM | GPUGRID | [slot] linked projects/www.gpugrid.net/project_2.png to projects/www.gpugrid.net/slideshow_cellmd_01
5/6/2015 6:05:13 PM | GPUGRID | [slot] linked projects/www.gpugrid.net/project_3.png to projects/www.gpugrid.net/slideshow_ga_02
5/6/2015 6:05:13 PM | GPUGRID | [slot] linked projects/www.gpugrid.net/project_3.png to projects/www.gpugrid.net/slideshow_cellmd_02
5/6/2015 6:05:13 PM | GPUGRID | [work_fetch] backing off CPU 815 sec
5/6/2015 6:05:13 PM | GPUGRID | [sched_op] Deferring communication for 00:00:31
5/6/2015 6:05:13 PM | GPUGRID | [sched_op] Reason: requested by project
5/6/2015 6:05:13 PM | | [statefile] set dirty: RPC complete
5/6/2015 6:05:13 PM | | [work_fetch] Request work fetch: RPC complete
5/6/2015 6:05:13 PM | | [statefile] Writing state file
5/6/2015 6:05:13 PM | | [statefile] Done writing state file
5/6/2015 6:05:18 PM | | [work_fetch] ------- start work fetch state -------
5/6/2015 6:05:18 PM | | [work_fetch] target work buffer: 180.00 + 21600.00 sec
5/6/2015 6:05:18 PM | | [work_fetch] --- project states ---
5/6/2015 6:05:18 PM | GPUGRID | [work_fetch] REC 0.000 prio 0.000 can't request work: scheduler RPC backoff (25.97 sec)
5/6/2015 6:05:18 PM | | [work_fetch] --- state for CPU ---
5/6/2015 6:05:18 PM | | [work_fetch] shortfall 174240.00 nidle 8.00 saturated 0.00 busy 0.00
5/6/2015 6:05:18 PM | GPUGRID | [work_fetch] share 0.000 project is backed off (resource backoff: 810.37, inc 600.00)
5/6/2015 6:05:18 PM | | [work_fetch] ------- end work fetch state -------
5/6/2015 6:05:18 PM | | [work_fetch] No project chosen for work fetch


The rest just keeps repeating that info over and over.

Even the Cuda version of DNetc will run on it with no problem at full pace.
How can I get BOINC to get GPUGrid work on this new card in this machine?
If you need more info, I can give it.
Thanks,
Mike
____________
1 Corinthians 9:16 "For though I preach the gospel, I have nothing to glory of: for necessity is laid upon me; yea, woe is unto me, if I preach not the gospel!"
Ephesians 6:18-20, please ;-)
http://tbc-pa.org

captainjack
Send message
Joined: 9 May 13
Posts: 157
Credit: 1,191,077,233
RAC: 531,468
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwat
Message 41044 - Posted: 7 May 2015 | 0:08:17 UTC

Mike,

Line 8 in the log gives a clue.

5/6/2015 6:05:11 PM | | No usable GPUs found

If you have a physical GPU installed in the machine, that message usually means that the GPU drivers are not being found. Since you changed out the GPU, you probably need to reinstall the driver.

That would be my first guess.

Hope that helps.

Profile caffeineyellow5
Avatar
Send message
Joined: 30 Jul 14
Posts: 225
Credit: 2,658,976,345
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwatwatwatwat
Message 41045 - Posted: 7 May 2015 | 0:17:23 UTC - in response to Message 41044.

I will uninstall:

NVIDIA 3D Vision Controller Driver 349.95 NVIDIA Corporation 4/22/2015 349.95
NVIDIA 3D Vision Driver 350.12 NVIDIA Corporation 5/6/2015 350.12
NVIDIA GeForce Experience 2.4.0.9 NVIDIA Corporation 3/18/2015 2.4.0.9
NVIDIA Graphics Driver 350.12 NVIDIA Corporation 5/6/2015 350.12
NVIDIA HD Audio Driver 1.3.33.0 NVIDIA Corporation 5/6/2015 1.3.33.0
NVIDIA PhysX System Software 9.15.0324 NVIDIA Corporation 4/22/2015 9.15.0324

And reinstall.

I'll let you know.

Profile caffeineyellow5
Avatar
Send message
Joined: 30 Jul 14
Posts: 225
Credit: 2,658,976,345
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwatwatwatwat
Message 41046 - Posted: 7 May 2015 | 1:38:10 UTC - in response to Message 41045.

No dice. Reseated the card, reinstalled the drivers, same logs.

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 2054
Credit: 15,017,649,069
RAC: 7,462,754
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41047 - Posted: 7 May 2015 | 17:12:06 UTC - in response to Message 41043.
Last modified: 7 May 2015 | 17:12:54 UTC

Your log says:

5/6/2015 6:05:11 PM | | Running as a daemon
5/6/2015 6:05:11 PM | | Data directory: C:\ProgramData\BOINC
5/6/2015 6:05:11 PM | | Running under account boinc_master
5/6/2015 6:05:11 PM | | No usable GPUs found

The first line says you've installed the BOINC manager in "protected application execution" mode, that means it runs as a system service.
System services cannot access the GPU directly (as you can see in line 4).
You should see your user name in the third line, but since BOINC runs as a daemon (system service) it needs a user account which is not associated with a real user.
Solution:
Reinstall BOINC to run under your user account, *not* in "protected application execution" mode.

Profile caffeineyellow5
Avatar
Send message
Joined: 30 Jul 14
Posts: 225
Credit: 2,658,976,345
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwatwatwatwat
Message 41049 - Posted: 7 May 2015 | 21:46:29 UTC - in response to Message 41047.

Yup, thanks again Zoltan. That was exactly it. I don't know what made it not work the first time that made me reinstall BOINC the first time, but when I did, I used the option "Install as a service" because on this machine, I have been having to start it manually after each reboot and thought if it was installed as a Windows service, it would move easily listen to the Scheduled Task that makes it start at login. For some reason, with Windows 8.1 I have had trouble getting BOINC to start at login, but I figure out a way eventually (and then forget how I did it by the time I set another one up.)

Thanks again!
Mike
____________
1 Corinthians 9:16 "For though I preach the gospel, I have nothing to glory of: for necessity is laid upon me; yea, woe is unto me, if I preach not the gospel!"
Ephesians 6:18-20, please ;-)
http://tbc-pa.org

Profile caffeineyellow5
Avatar
Send message
Joined: 30 Jul 14
Posts: 225
Credit: 2,658,976,345
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwatwatwatwat
Message 41050 - Posted: 8 May 2015 | 1:02:32 UTC - in response to Message 41049.

Actually, that would be a big NOPE. I can reinstall it after a fresh uninstall (manually deleting directories, removing startup entries, etc) and then it works. It works, that is, until a reboot or I manually close it to check if it will come back on. Then it won't connect again. I am obviously not understanding how to install the program under my own account. I am, after all, the only user on the computer and I am logged in as such. I am also not thinking the "Install as a Service" option was my problem after all during the install. I never checked it before on any other installs.

Gonna keep trying...

Jacob Klein
Send message
Joined: 11 Oct 08
Posts: 1115
Credit: 1,831,665,089
RAC: 566,289
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41051 - Posted: 8 May 2015 | 1:41:39 UTC - in response to Message 41050.

You definitely had it installed as a service, and in that mode BOINC definitely cannot find GPUs. So, that for sure was one problem.

So, you still have a separate problem to troubleshoot. Sorry.

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 2054
Credit: 15,017,649,069
RAC: 7,462,754
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41053 - Posted: 8 May 2015 | 8:46:54 UTC - in response to Message 41050.

Can you please paste the beginning of your log file when it works and after a restart when it doesn't.
If there's the line containing "Running as a daemon" then it runs as a service, otherwise this line is simply not present.

Profile caffeineyellow5
Avatar
Send message
Joined: 30 Jul 14
Posts: 225
Credit: 2,658,976,345
RAC: 0
Level
Phe
Scientific publications
watwatwatwatwatwatwatwat
Message 41054 - Posted: 9 May 2015 | 21:29:14 UTC - in response to Message 41053.
Last modified: 9 May 2015 | 21:29:32 UTC

OK, I figured it out.

Since I did one install of BOINC as a service, it was in the registry even after uninstalling and manually deleting all the folders and files. So every time after that, it kicked in after each successful fresh install and restart of the application. I would not have thought of it except something seemed odd during each install. The install kept remembering my email address and filling it in to get GPUGrid running on BOINC, even though it made me type the password. Just that remembering of my password finally sunk in that the Windows registry must be the issue holding settings, run levels, and accounts. After uninstalling again, I went and cleaned the registry of all mentions of BOINC (including the Space Sciences Laboratory entries)(which covered the mentions of GPUGrid). After that, a fresh install and it works after restarts and reboots!

Jacob Klein
Send message
Joined: 11 Oct 08
Posts: 1115
Credit: 1,831,665,089
RAC: 566,289
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41055 - Posted: 9 May 2015 | 21:56:03 UTC

Interesting. It really should not have been that much of a hassle to uninstall, restart, and reinstall (as a service or non-service type). I'm surprised it gave you trouble.

At any rate, good sleuthing, and I'm glad it's finally working the way you want it to!!

Profile Retvari Zoltan
Avatar
Send message
Joined: 20 Jan 09
Posts: 2054
Credit: 15,017,649,069
RAC: 7,462,754
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 41056 - Posted: 10 May 2015 | 10:25:08 UTC - in response to Message 41054.

I'm glad that you've figured it out, I've had the feeling that this is the issue.
BTW you can check the services running under Windows by the Task Manager, or by typing services.msc in the search box, and running it with elevated privileges.
You can stop, restart, start or disable/enable the services with this GUI, but you can't delete them. To delete a service you should first stop it, check its short name, and open an elevated command prompt, and issue the command SC DELETE <service_name>.
Deleting it from the registry is another method, but this could be dangerous when you use an overly general search phrase to find the keys and values you should delete.

Post to thread

Message boards : Graphics cards (GPUs) : How to troubleshoot a new card?