I installed the einstein project after hearing of the pulsar discovery. Since then my system slows to an unacceptable level. I've been running BOINC for several other projects for a year with no problems.
I built and am running Ubuntu 10.10 on a AMD64 Phenom II Quad 4 processor w/8 gig RAM. BOINC says my GPU isn't compatable so it's not being used. I'm overpowered for a home desktop so I run BOINC full time on two processors. I can't even tell its running.
I got a couple projects from einstein and my system slowed almost to a halt. The mouse skips across the screen, when repositioning windows they pause then catch up. I noticed Compiz, a display manager, struggling with increased CPU usage.
I pause the Arecibo project and everything returns to normal. I also have a Global Correlations project and it doesn't cause problems. Just the Arecibo. BOINC Should be confined to the two CPUs. I still have the other projects running. Even the einstein global correlations project. No problem until i resume the Arecibo project. Then everything slows to a crawl. There's somthing unique to this project my system doesn't like.
Does anyone know of an issue with this project? Maybe some incompatibility with Ubuntu 10.10, Compiz, maybe it's using my GPU behind BOINCs back?
My only option absent a solution is see if I can limit the einstein projects to just the gravity or uninstall the einstein projects all together.
Thanks for any information!
-[d]-
Copyright © 2024 Einstein@Home. All rights reserved.
Problems w/Arecibo
)
Hi Dean,
could you post the messages (during BOINC startup) from the message tab here.
Welcome,
Michael
Team Linux Users Everywhere
RE: I got a couple projects
)
Not behind BOINC's back, but the symptoms you describe, point strongly to the GPU.
What happens if you disable GPU usage in your Einstein@Home preferences?
By the way, why doesn't your account show any attached computers? Did you create a new account just for posting here?
Gruß,
Gundolf
Computer sind nicht alles im Leben. (Kleiner Scherz)
Thank you for responding. I
)
Thank you for responding. I agree w/Gundolf Jahn that it points to GPU usage. BOINC has always indicated my GPU wasn't up to snuff and that it would not be used. I double checked the preferences and the "Use GPU..." was not marked.
To get the message file I reactivated the project and the symptoms reappeared. Then I rebooted to get a fresh start file. On reboot, the symptoms went away. I don't know why. I'm going to let it run for awhile to see if the lag returns. If not I'll toggle the "Use GPU..." box to see if the symptoms reappear.
If they do, I'll assume that for whatever reason upon installing this project BOINC decided to start using my GPU, even though it wasn't marked to be used, and as far as I know BOINC had it disabled and it was being used only for this project an no others... You know, one of those computer things that only a reboot fixes.
Thanks for your reply and if the lag returns I'll play with the GPU box and update this post. If there's anything in my startup I should be aware of let me know. With that said follows is my message file from startup. Thanks!!!
Fri 29 Oct 2010 05:47:21 PM EDT Starting BOINC client version 6.10.58 for x86_64-pc-linux-gnu
Fri 29 Oct 2010 05:47:21 PM EDT Config: GUI RPC allowed from:
Fri 29 Oct 2010 05:47:21 PM EDT log flags: file_xfer, sched_ops, task
Fri 29 Oct 2010 05:47:21 PM EDT Libraries: libcurl/7.21.0 OpenSSL/0.9.8o zlib/1.2.3.4 libidn/1.18
Fri 29 Oct 2010 05:47:21 PM EDT Data directory: /var/lib/boinc-client
Fri 29 Oct 2010 05:47:21 PM EDT Processor: 4 AuthenticAMD AMD Phenom(tm) II X4 940 Processor [Family 16 Model 4 Stepping 2]
Fri 29 Oct 2010 05:47:21 PM EDT Processor: 512.00 KB cache
Fri 29 Oct 2010 05:47:21 PM EDT Processor features: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush mmx fxsr sse sse2 ht syscall nx mmxext fxsr_opt pdpe1gb rdtscp lm 3dnowext 3dnow constant_tsc rep_good nonstop_tsc extd_apicid pni monitor cx16 popcnt lahf
Fri 29 Oct 2010 05:47:21 PM EDT OS: Linux: 2.6.35-22-generic
Fri 29 Oct 2010 05:47:21 PM EDT Memory: 7.81 GB physical, 22.89 GB virtual
Fri 29 Oct 2010 05:47:21 PM EDT Disk: 894.36 GB total, 487.95 GB free
Fri 29 Oct 2010 05:47:21 PM EDT Local time is UTC -4 hours
Fri 29 Oct 2010 05:47:21 PM EDT NVIDIA GPU 0: GeForce 8400 GS (driver version unknown, CUDA version 3020, compute capability 1.1, 512MB, 22 GFLOPS peak)
Fri 29 Oct 2010 05:47:22 PM EDT Milkyway@home URL http://milkyway.cs.rpi.edu/milkyway/; Computer ID 192867; resource share 100
Fri 29 Oct 2010 05:47:22 PM EDT SETI@home URL http://setiathome.berkeley.edu/; Computer ID 5456855; resource share 100
Fri 29 Oct 2010 05:47:22 PM EDT Einstein@Home URL http://einstein.phys.uwm.edu/; Computer ID 3620159; resource share 100
Fri 29 Oct 2010 05:47:22 PM EDT rosetta@home URL http://boinc.bakerlab.org/rosetta/; Computer ID 1310607; resource share 100
Fri 29 Oct 2010 05:47:22 PM EDT climateprediction.net URL http://climateprediction.net/; Computer ID 1086434; resource share 100
Fri 29 Oct 2010 05:47:22 PM EDT No general preferences found - using BOINC defaults
Fri 29 Oct 2010 05:47:22 PM EDT Reading preferences override file
Fri 29 Oct 2010 05:47:22 PM EDT Preferences:
Fri 29 Oct 2010 05:47:22 PM EDT max memory usage when active: 4000.96MB
Fri 29 Oct 2010 05:47:22 PM EDT max memory usage when idle: 7201.73MB
Fri 29 Oct 2010 05:47:22 PM EDT max disk usage: 10.00GB
Fri 29 Oct 2010 05:47:22 PM EDT max CPUs used: 2
Fri 29 Oct 2010 05:47:22 PM EDT don't use GPU while active
Fri 29 Oct 2010 05:47:22 PM EDT suspend work if non-BOINC CPU load exceeds 25 %
Fri 29 Oct 2010 05:47:22 PM EDT (to change preferences, visit the web site of an attached project, or select Preferences in the Manager)
Fri 29 Oct 2010 05:47:23 PM EDT Not using a proxy
Fri 29 Oct 2010 05:47:23 PM EDT Milkyway@home Restarting task de_nbody_2comp_1_5263_1288349339_0 using milkyway_nbody version 21
Fri 29 Oct 2010 05:47:27 PM EDT Einstein@Home Restarting task h1_1145.00_S5R4__341_S5GC1a_1 using einstein_S5GC1 version 105
RE: RE: I got a couple
)
I agree and on reboot the lag diapered. I'll see what happens and if it comes back focus and play with the GPU settings. As far as I know GPU wasn't supposed to be used and the "Use GPU..." box has always been unchecked. Oh well...
My account:
I screwed up and didn't realize I already had an account and created a new one. I've been running BOINC for awhile but have never visited the forums or project pages before now. My projects are running, I suspect, under fox (dot) dean (at) gmail.com.
Thanks for your response!
RE: Hi Dean, could you
)
I spoke too soon. The lag has reappeared. Sometime between the reboot and finishing my last posts to this forum.
The GPU box in BOINC manager is unchecked. But Gundolf Jahn mentioned settings at Einstein@Home preferences? I followed that link and sure enough Use GPU is set to yes on that page.
I've updated the two "Use GPU..." setting there to off. I'll see if that helps.
Thanks guys!
-[d]-
RE: I installed the
)
Sorry for the confusing posts and the other user ID. I inadvertently created a new ID to do this post not realizing I already had one. This ID/user name should be the one this project is running under.
Everything points to GPU. I have turned off every GPU setting I can find. Both in the BOINC manager and on the einstein project webpage. However, If I resume this project the screen lag eventually comes back. On reboot everything is fine for about five minutes then the screen lag comes back. I can only fix it by suspending this project. Once the screen lag happens suspending/resuming this project fixes then causes the lag to come back.
I agree that it appears to be a GPU issue. But only with this one project. Unless there's a GPU setting I'm missing it would appear this project is ignoring those settings. I go back through the setting again. If I can resolve it I'll have to pass using this particular BOINC project :(
Any other ideas, maybe something I'm missing?
Thanks!
-[d]-
RE: I have turned off every
)
The preference I mentioned (in the Einstein@Home preferences not the Computing preferences) only prevents new tasks from being downloaded. The 4 tasks already on your host are not affected by this. You'll either have to abort them or let them run when you aren't using your computer. For that purpose, the "Suspend GPU work while computer is in use" was implemented.
For that preference to work, though, you'll have to select "Run based on preferences" (Activity menu). And remember that local preferences override the online ones.
Gruß,
Gundolf
Computer sind nicht alles im Leben. (Kleiner Scherz)
RE: RE: I have turned off
)
Thank you.
I did that (I believe) I detached the einstein project then reattached it. All my settings on the project page indicate no GPU. I even set it so einstein would not request Arecibo tasks as they're the only ones causing problems.
Webpage Preferences:
Use CPU Enforced by version 6.10+ = NO
Use NVIDIA GPU Enforced by version 6.10+ = NO
Run only the selected applications: Arecibo Binary Pulsar Search (STSP): = NO (All others set to YES)
BOINC Preferences:
BOINC Version 6.10.58
Activity TAB = Run base on preferences = YES
Use GPU based on preferences = YES
Under "Computing preferences" I set to run projects only from midnight to 6am, detached einstein and reattached. New Areciebo tasks were downloaded and started running immediately then stared to use the GPU.
"Run base on preferences = YES" and indicating no Arecibo tasks and no GPU didn't do it for me. I killed all my current einstein projects, BOINC then downloaded new tasks in including Arecibo (when my preferences are set not too) and apparently started using my GPU (when it's turned off). But apparently only for the Arecibo tasks.
I'll still concede the problem exists between keyboard and chair but I'm beginning to wonder the einstein arecibo tasks are reading my project settings. I'm not a completely brain dead user. That or the BOINC setup GUI, preferences and management are just not intuitive to me.
On my BOINC Manager, Activity TAB, I set "Use GPU Never" = YES. This appears to override it all and I'm no longer seeing the display lag. At least for now. I've never RTFM, it all seemed pretty straight forward. Until Arecibo. Although I seem to have found the solution I'm still at a loss why the Arecibo tasks and only the Arecibo tasks cause this problem for me. But at least for now the problem seems to have gone away with the new Activity setting.
Thanks for your input and all your help!
-[d]-
The problem may be because
)
The problem may be because you "detached and then reattached" Einstein. With the preference settings you've described, Einstein should not allocate new work for your GPU, but it may still resend old work lost from your computer by the detach.
Check your message log for the time period immediately after the re-attach: any work resent would be clearly marked.
The best thing to do would be to examine the task list in BOINC Manager, and explicitly select, and then abort, any which show any variant of (cuda...) at the end of the application description (you may have to stretch the column width to see that). That should get rid of the blighters for good.
RE: On my BOINC Manager,
)
This merely suspends the GPU, it doesn't really stop it being detected or eventually asking for work.
Read the CUDA & CAL FAQ as to why this is. Yes, I wrote all of that. Yes, it needs some updating, but the main gist is still quite accurate. Even the BOINC developers point it out from their CUDA pages. :-)
And it reads a little easier than a whole manual.