I upgraded to BOINC 7.0.28 a while ago and noticed that I don't seem to run GPU work units any longer like I did when I was running 6.0.12. At first it seemed that they all had compute errors, but then CUDA tasks stopped coming in.

Currently there is only one CUDE task in the work queue. The lone CUDA task is from moo_wrapper where I found something about creating an app_info.xml file for running dual ATI cards. I have dual GeForce GTX 550 Ti cards that used to run under 6.0.12. This note is at http://moowrap.net/forum_thread.php?id=272

The first lines from my Event Log are:
================================================================================
8/30/2012 9:05:19 PM | | No config file found - using defaults
8/30/2012 9:05:19 PM | | Starting BOINC client version 7.0.28 for windows_x86_64
8/30/2012 9:05:19 PM | | log flags: file_xfer, sched_ops, task
8/30/2012 9:05:19 PM | | Libraries: libcurl/7.25.0 OpenSSL/1.0.1 zlib/1.2.6
8/30/2012 9:05:19 PM | | Data directory: C:\ProgramData\BOINC
8/30/2012 9:05:19 PM | | Running under account Dude
8/30/2012 9:05:19 PM | | Processor: 4 AuthenticAMD AMD FX(tm)-4100 Quad-Core Processor [Family 21 Model 1 Stepping 2]
8/30/2012 9:05:19 PM | | Processor: 2.00 MB cache
8/30/2012 9:05:19 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 cx16 sse4_1 sse4_2 syscall nx lm svm sse4a osvw ibs xop skinit wdt lwp fma4 page1gb rdtscp
8/30/2012 9:05:19 PM | | OS: Microsoft Windows 7: Home Premium x64 Edition, Service Pack 1, (06.01.7601.00)
8/30/2012 9:05:19 PM | | Memory: 16.00 GB physical, 31.99 GB virtual
8/30/2012 9:05:19 PM | | Disk: 494.73 GB total, 377.00 GB free
8/30/2012 9:05:19 PM | | Local time is UTC -4 hours
8/30/2012 9:05:19 PM | | VirtualBox version: 4.1.8
8/30/2012 9:05:19 PM | | NVIDIA GPU 0: GeForce GTX 550 Ti (driver version 301.42, CUDA version 4.20, compute capability 2.1, 1024MB, 887MB available, 697 GFLOPS peak)
8/30/2012 9:05:19 PM | | NVIDIA GPU 1: GeForce GTX 550 Ti (driver version 301.42, CUDA version 4.20, compute capability 2.1, 1024MB, 887MB available, 697 GFLOPS peak)
8/30/2012 9:05:19 PM | | OpenCL: NVIDIA GPU 0: GeForce GTX 550 Ti (driver version 301.42, device version OpenCL 1.1 CUDA, 1024MB, 887MB available)
8/30/2012 9:05:19 PM | | OpenCL: NVIDIA GPU 1: GeForce GTX 550 Ti (driver version 301.42, device version OpenCL 1.1 CUDA, 1024MB, 887MB available)
8/30/2012 9:05:19 PM | Constellation | URL http://aerospaceresearch.net/constellation/; Computer ID 7006; resource share 500
8/30/2012 9:05:19 PM | rosetta@home | URL http://boinc.bakerlab.org/rosetta/; Computer ID 1503442; resource share 150
8/30/2012 9:05:19 PM | DistrRTgen | URL http://boinc.freerainbowtables.com/distrrtgen/; Computer ID 31325; resource share 500
8/30/2012 9:05:19 PM | Poem@Home | URL http://boinc.fzk.de/poem/; Computer ID 110049; resource share 100
8/30/2012 9:05:19 PM | Leiden Classical | URL http://boinc.gorlaeus.net/; Computer ID 92770; resource share 100
8/30/2012 9:05:19 PM | Collatz Conjecture | URL http://boinc.thesonntags.com/collatz/; Computer ID 90139; resource share 200
8/30/2012 9:05:19 PM | The Lattice Project | URL http://boinc.umiacs.umd.edu/; Computer ID 73419; resource share 100
8/30/2012 9:05:19 PM | boincsimap | URL http://boincsimap.org/boincsimap/; Computer ID 219448; resource share 100
8/30/2012 9:05:19 PM | BURP | URL http://burp.renderfarming.net/; Computer ID 52919; resource share 500
8/30/2012 9:05:19 PM | climateprediction.net | URL http://climateprediction.net/; Computer ID 1185692; resource share 200
8/30/2012 9:05:19 PM | Docking | URL http://docking.cis.udel.edu/; Computer ID 107264; resource share 100
8/30/2012 9:05:19 PM | Einstein@Home | URL http://einstein.phys.uwm.edu/; Computer ID 4309125; resource share 200
8/30/2012 9:05:19 PM | Test4Theory@Home | URL http://lhcathome2.cern.ch/test4theory/; Computer ID 14461; resource share 500
8/30/2012 9:05:19 PM | LHC@home 1.0 | URL http://lhcathomeclassic.cern.ch/sixtrack/; Computer ID 9948039; resource share 100
8/30/2012 9:05:19 PM | Milkyway@Home | URL http://milkyway.cs.rpi.edu/milkyway/; Computer ID 355680; resource share 200
8/30/2012 9:05:19 PM | MindModeling@Beta | URL http://MindModeling.org/beta/; Computer ID 19162; resource share 100
8/30/2012 9:05:19 PM | Moo! Wrapper | URL http://moowrap.net/; Computer ID 7910; resource share 200
8/30/2012 9:05:19 PM | orbit@home | URL http://orbit.psi.edu/oah/; Computer ID 82049; resource share 100
8/30/2012 9:05:19 PM | Pirates@Home | URL http://pirates.spy-hill.net/; Computer ID 71799; resource share 100
8/30/2012 9:05:19 PM | QMC@HOME | URL http://qah.uni-muenster.de/; Computer ID 203296; resource share 100
8/30/2012 9:05:19 PM | SETI@home | URL http://setiathome.berkeley.edu/; Computer ID 6293163; resource share 1000
8/30/2012 9:05:19 PM | SETI@home Beta Test | URL http://setiweb.ssl.berkeley.edu/beta/; Computer ID 55373; resource share 100
8/30/2012 9:05:19 PM | Spinhenge@home | URL http://spin.fh-bielefeld.de/; Computer ID 220826; resource share 100
8/30/2012 9:05:19 PM | Cosmology@Home | URL http://www.cosmologyathome.org/; Computer ID 151072; resource share 100
8/30/2012 9:05:19 PM | Enigma@Home | URL http://www.enigmaathome.net/; Computer ID 71412; resource share 100
8/30/2012 9:05:19 PM | FreeHAL@home | URL http://www.freehal.net/freehal_at_home/; Computer ID 72013; resource share 100
8/30/2012 9:05:19 PM | GPUGRID | URL http://www.gpugrid.net/; Computer ID 115985; resource share 200
8/30/2012 9:05:19 PM | PrimeGrid | URL http://www.primegrid.com/; Computer ID 240499; resource share 100
8/30/2012 9:05:19 PM | yoyo@home | URL http://www.rechenkraft.net/yoyo/; Computer ID 63254; resource share 100
8/30/2012 9:05:19 PM | uFluids | URL http://www.ufluids.net/; Computer ID 135801; resource share 100
8/30/2012 9:05:19 PM | World Community Grid | URL http://www.worldcommunitygrid.org/; Computer ID 1830670; resource share 200
8/30/2012 9:05:19 PM | Moo! Wrapper | General prefs: from Moo! Wrapper (last modified 09-Aug-2012 19:33:33)
8/30/2012 9:05:19 PM | Moo! Wrapper | Host location: none
8/30/2012 9:05:19 PM | Moo! Wrapper | General prefs: using your defaults
8/30/2012 9:05:19 PM | | Reading preferences override file
8/30/2012 9:05:19 PM | | Preferences:
8/30/2012 9:05:19 PM | | max memory usage when active: 12285.93MB
8/30/2012 9:05:19 PM | | max memory usage when idle: 16381.24MB
8/30/2012 9:05:19 PM | | max disk usage: 50.00GB
8/30/2012 9:05:19 PM | | don't compute while active
8/30/2012 9:05:19 PM | | don't use GPU while active
8/30/2012 9:05:19 PM | | (to change preferences, visit the web site of an attached project, or select Preferences in the Manager)
8/30/2012 9:05:19 PM | | Not using a proxy
8/30/2012 9:05:21 PM | | Suspending computation - computer is in use
8/30/2012 9:05:21 PM | | Suspending network activity - computer is in use
8/30/2012 9:24:09 PM | | Resuming network activity
8/30/2012 9:26:56 PM | | Suspending network activity - computer is in use
8/30/2012 9:35:31 PM | | Resuming network activity
================================================================================

I was wondering if there is a global configuration to get these tasks to work for all projects or do I need to create app specific configuration files?

---

David - KC2WUF

I upgraded to BOINC 7.0.28 .. I was running 6.0.12

6.0.12 doesn't exist. You must've run some 6.10.x or 6.12.x version.
But in any case, what are your settings for minimum buffer and maximum additional buffer?

If these are still on the old 6.12 values, then do know that this changed in 7.0; now BOINC will use the cache settings as minimum water mark settings, where the minimum buffer is truly the minimum that BOINC will want to have work for and the maximum additional buffer is only added to it.

This means that when you ran 6.12 with e.g. 3 days and 7 days to run a 10 day buffer, that you will now have to change this to 7 and 3. The 7 days is then the minimum setting that BOINC will follow to fill the buffer. It will only start filling again when the amount of work in the cache falls to below 7 days.

Furthermore, BOINC 7.0 has completely new and separate CPU and GPU schedulers. There's also a new work-fetch module for both. These cannot be compared to BOINC 6.12, they'll work differently from previous BOINC versions. BOINC will also have to re-learn how long work takes, this depends on how much work you run for each project, and whether or not the project uses CreditNew as well.

In any case, the other project you expect CUDA work from, which project is that? Does it have CUDA work at this time? Does it still have bandwidth left-over to send out work at this time? What do its forums say about this?

---

I tried your suggestion and changed my buffers from 0.1, 0.0 to 0.9 and 0.1. All this did was downloaded 10 Moo! Wrapper cuda tasks. My lonely one expired and was never run. The 10 tasks that were downloaded later are still sitting in my queue even though I've disabled any further work from being fetched. I have 3 other tasks remaining on my quad processor and they are running, leaving a single CPU available but unused even though the cuda tasks only require 0.2 CPUs to run.

Someone on the Moo! Wrapper board suggested going back to version 7.0.25, but that is no longer available for download.

As far as other projects, you should see that I am signed up for many of the available projects and probably half of them have run cuda tasks on my machine in the past, but now seem unable to run or even download them.

Any other suggestions from anyone on what might be causing this?

---

David - KC2WUF

I just downgraded to V 7.0.25 and I am now running 2 Einstein@Home CUDA tasks and a DistRTgen CUDA task so far. SETI@home has also downloaded some task, but I haven't waited for them to run yet. Moo! Wrapper also downloaded the CUDA tasks, but they didn't run when they were the only tasks on the system after I reverted back to 7.0.25.

At least other programs seem to be downloading and running CUDA tasks, so now I might have to go to Moo! Wrapper forum to fix their project with CUDA tasks.

Thanks for your help and I'll let you know what other projects with CUDA tasks do as they download and they try to run.

---

David - KC2WUF

Although the SETI@home tasks started downloading earlier today. They were not in the queue when I just looked for them. The SETI@home site said I had more than 20 CUDA tasks waiting to run on my PC. I am currently resetting SETI@home and seeing whether tasks download and stay in the queue.

By the way, The lone DistRTgen CUDA task and 9 Einstein@Home CUDA tasks ran to completion and had valid results. 2 projects fixed so far.

---

David - KC2WUF

I am fairly certain that running V7.0.25 has resolved most, if not all, of my issues with running CUDA tasks for all of my projects except for Moo! Wrapper (project specific problem with multi-GPUs). I did notice an occasional glitch where a task doesn't seem to start on the second GPU, but I can't remember which project it was. I suppose that I could specify not to run that project on the second GPU to solve that problem when it occurs again.

---

David - KC2WUF

just digging up an old thread to say that i encountered the same issue on 7.0.28 (x64) on my notebook with a ATI 5650 graphics card.

SETI@HOME and EINSTEIN@HOME downloaded the opencl tasks but never ran it while the CPU was chugging away at the CPU tasks.


came across this thread and downgraded to 7.0.25 (x64). The opencl tasks started right after.

something is broken in rev 28.
I'm not sure if rev 31 fixes it as I didnt try it before going down to 25.


Thanks David Bean!

came across this thread and downgraded to 7.0.25 (x64). The opencl tasks started right after.

And those will be the only ones running to completion, before you won't get any further work at Einstein, because they require 7.0.27 or above for getting OpenCL work. Just a warning.

Also, Einstein uses a lot of CPU when running the OpenCL GPU tasks. It likes it best if it can use one free CPU core. Did you set up for that, set BOINC to use all but one CPU core? For if not, Einstein tasks can run either not, or very slowly. Especially when the CPU cores are very much busy with other projects.

---

Thanks for the warning. I will look out for that issue.
If it gets down to that, i'll try other versions to see where the sweet spot is.

There are free cores around.
Previously only the CPU tasks runs with the GPU idling.

Currently there is 1x (0.346 CPUs + 1 ATI GPU) Seti@home ati_opencl_100 Task running and 1x Einstein@home BRP4SSE CPU task running concurrently.

The Einstein CPU task is currently alternating with Milkyway CPU task while the GPU SETI chugs along

as you said, below 7.0.27 gets no new GPU tasks.

uninstalled 7.0.25 installed 7.0.36 and its working great