| Author |
Message |
|---|
|
|
|
check out that WU: http://www.primegrid.com/workunit.php?wuid=122565546 |
|
|
Ken_g6 Volunteer developer
Send message
Joined: 4 Jul 06
Posts: 929
ID: 3110
Credit: 237,002,205
RAC: 39,933
                           
|
pmax not specified, using default pmax = pmin + 1e9
Please specify an input file or all of kmin, kmax, and nmax
called boinc_finish
Looks like none (or fewer than all) of the parameters were specified for this WU!
____________
|
|
|
|
|
pmax not specified, using default pmax = pmin + 1e9
Please specify an input file or all of kmin, kmax, and nmax
called boinc_finish
Looks like none (or fewer than all) of the parameters were specified for this WU!
yup! and there are a lot of them.
does anyone running the CUDA-apps have valids??
|
|
|
Scott BrownVolunteer moderator
Project administrator
Volunteer tester
Project scientist
Send message
Joined: 17 Oct 05
Posts: 2329
ID: 1178
Credit: 15,631,005,591
RAC: 10,487,038
                                           
|
does anyone running the CUDA-apps have valids??
Some...see here, for example.
____________
141941*2^4299438-1 is prime!
 |
|
|
|
|
does anyone running the CUDA-apps have valids??
Some...see here, for example.
look like it's darwin.. ;)
|
|
|
|
|
|
or murphy..
(as in murphy's law)
:p
|
|
|
|
|
|
Yeah, I've got 40+ valid ppsieve cuda WUs so far, over the last several days on my fairly lame GT120 GPU on my iMac (OSX 10.6.4). About 2:05:00 wall clock time per WU. The key was to update to cuda version 3.1.10 from the nvidia website. Before that everything "completed", but was marked invalid, after just a couple seconds runtime.
--Gary |
|
|
Ken_g6Volunteer developer
Send message
Joined: 4 Jul 06
Posts: 929
ID: 3110
Credit: 237,002,205
RAC: 39,933
|
|
I suspect it's the same problem with old Linux drivers. See, the Linux CUDA code had to be compiled against some drivers, so I downloaded the only ones I could find, 256.25. Any versions before that cause problems.
I'd be working on a fix, but actually, roadrunner_gs seems to be making good progress on fixing my code. :S
____________
|
|
|
|
|
I suspect it's the same problem with old Linux drivers. See, the Linux CUDA code had to be compiled against some drivers, so I downloaded the only ones I could find, 256.25. Any versions before that cause problems.
maybe someone should put up a hint for those apple-users which are wasting their effort..
|
|
|
Ken_g6Volunteer developer
Send message
Joined: 4 Jul 06
Posts: 929
ID: 3110
Credit: 237,002,205
RAC: 39,933
|
|
I didn't say that quite right. Any driver versions before 256.25 currently cause problems on Linux. I don't know what version Iain may have used to compile the Mac app.
It looks like the app can be compiled without the driver dependency. Iain will be back and able to compile fixes soon, I hope.
____________
|
|
|
|
|
|
there are still a lot of MACs crashing everything and flooding the server.
PLEASE update your drivers! |
|
|
|
|
|
I've built the new app, which should resolve much of these problems. Unfortunately we can't add it to PrimeGrid until Rytis is back, which I believe will be next week.
Cheers
- Iain |
|
|
|
|
there are still a lot of MACs crashing everything and flooding the server.
PLEASE update your drivers!
Agreed!, I'm on OSX 10.6.4 and updating the CUDA driver to 3.1.10 resolved the "immediately error out" problem. I'm happily crunching and validating now, although I hope Iain's next update addresses screen-lag issues too (thanks Iain!)
--Gary |
|
|
|
|
I've built the new app, which should resolve much of these problems. Unfortunately we can't add it to PrimeGrid until Rytis is back, which I believe will be next week.
Cheers
- Iain
Any chance it's gonna be also for Windows? |
|
|
Ken_g6Volunteer developer
Send message
Joined: 4 Jul 06
Posts: 929
ID: 3110
Credit: 237,002,205
RAC: 39,933
|
Any chance it's gonna be also for Windows?
I built a Windows version. Testers say it works. The rest is up to Rytis.
____________
|
|
|
|
|
|
I Guess that Windows Cuda app has been released, as BOINC is downloading dozens of 1.24 (cuda23) tasks for my GeForce 9800GT to do :-) - unfortunately they all crash out instantly - so obviously that Mac problem is still also with us??
I have the latest Nvidia (258.96) drivers and am crunching Collatz tasks with no problems.. - These tasks appear to try to run and crash even while a Collatz task is in progress. I tried suspending all other tasks (CPU and GPU) - but still no dice.. - Have disabled GPU tasks for now :-(
|
|
|
|
|
I Guess that Windows Cuda app has been released, as BOINC is downloading dozens of 1.24 (cuda23) tasks for my GeForce 9800GT to do :-) - unfortunately they all crash out instantly - so obviously that Mac problem is still also with us??
I have the latest Nvidia (258.96) drivers and am crunching Collatz tasks with no problems.. - These tasks appear to try to run and crash even while a Collatz task is in progress. I tried suspending all other tasks (CPU and GPU) - but still no dice.. - Have disabled GPU tasks for now :-(
I have a similar experience on my EVGA GTX 465SC, running on WinVista_X64 SP2 under BOINC 6.10.58 with the WHQL 258.96 drivers. The batch d/l'd while I wasn't looking and started crashing immediately. I was running a moderate (15%) overclock so I snoozed the GPU and reset it to default values (Shader clock 1250/Core clock 625, Memory clock 1620). The next WU actually made it to 3:30 (ET) before crashing, but the rest went after just seconds.
I just upgraded to the 465 last week, but I've been running GPUGRID (cuda31), Collatz (cuda23), DNETC (cuda31) & Seti (cuda_fermi) without problem, even at the overclock.
Sorry, but I have to call this build a write-off, at least on my system, and I've disabled d/l to the GPU. Fingers crossed tho -- I'm looking forward to some serious sieving.
-----
EDIT--I notice the Applications page doesn't say this cuda app is for Windows running on Intel EM64T. I hope the server isn't sending them to me by mistake (hint) ;)
____________
  |
|
|
|
|
|
Hi I've the same problem.
My System is Win Vista 64,4Gigs RAM,NVIDIA 130M GPU 1GB.
I get today 19 WU'S and have same Compute error.
03.08.2010 08:57:36 PrimeGrid Requesting new tasks for GPU
03.08.2010 08:57:38 PrimeGrid Scheduler request completed: got 19 new tasks
03.08.2010 08:58:08 PrimeGrid Output file pps_sr2sieve_1429233_3_0 for task pps_sr2sieve_1429233_3 absent
03.08.2010 08:58:08 PrimeGrid Starting pps_sr2sieve_1429232_3
03.08.2010 08:58:08 PrimeGrid Starting task pps_sr2sieve_1429232_3 using pps_sr2sieve version 124
03.08.2010 08:58:09 PrimeGrid Finished download of primegrid_ppsieve_1.24_windows_x86_64.exe
03.08.2010 08:58:09 PrimeGrid Finished download of psp_sr2sieve_36269064_cmd
03.08.2010 08:58:10 PrimeGrid Computation for task pps_sr2sieve_1429232_3 finished
03.08.2010 08:58:10 PrimeGrid Output file pps_sr2sieve_1429232_3_0 for task pps_sr2sieve_1429232_3 absent
03.08.2010 08:58:10 PrimeGrid Starting pps_sr2sieve_1429153_3
03.08.2010 08:58:10 PrimeGrid Starting task pps_sr2sieve_1429153_3 using pps_sr2sieve version 124
03.08.2010 08:58:14 PrimeGrid Computation for task pps_sr2sieve_1429153_3 finished
03.08.2010 08:58:14 PrimeGrid Output file pps_sr2sieve_1429153_3_0 for task pps_sr2sieve_1429153_3 absent
03.08.2010 08:58:14 PrimeGrid Starting pps_sr2sieve_1429105_3
03.08.2010 08:58:14 PrimeGrid Starting task pps_sr2sieve_1429105_3 using pps_sr2sieve version 124
03.08.2010 08:58:16 PrimeGrid Computation for task pps_sr2sieve_1429105_3 finished
03.08.2010 08:58:16 PrimeGrid Output file pps_sr2sieve_1429105_3_0 for task pps_sr2sieve_1429105_3 absent
03.08.2010 08:58:16 PrimeGrid Starting pps_sr2sieve_1429088_3
I hope you can help.
Thanks a lot. :-)) |
|
|
Ken_g6Volunteer developer
Send message
Joined: 4 Jul 06
Posts: 929
ID: 3110
Credit: 237,002,205
RAC: 39,933
|
|
FYI, there seems to be a problem with the app on all Fermi-based GPUs. We're investigating, but it is going to take some time to resolve. I could use a Linux tester in the testing thread.
____________
|
|
|
|
|
|
any update on OSX CUDA? |
|
|
|
|
|
I'm working on it at the moment. In any case, until the server-side configuration is fixed, it won't run on your GPU due to the minumum memory requirement.
Cheers
- Iain |
|
|
|
|
any update on OSX CUDA?
I've run the new app on OSX, and with the latest BOINC version and the latest CUDA version, I was absolutely fine on the latest version of OS 10.5.
____________
 |
|
|
|
|
|
Hi Michael - I looked at your WUs, and it looks like you had lots of failed WUs, then you managed to complete one on 3rd August - can you confirm that this was after you upgraded your CUDA driver? And the ones before - what driver version did you have then?
Thanks for the info
- Iain |
|
|
|
|
Hi Michael - I looked at your WUs, and it looks like you had lots of failed WUs, then you managed to complete one on 3rd August - can you confirm that this was after you upgraded your CUDA driver? And the ones before - what driver version did you have then?
Thanks for the info
- Iain
Yes, that's exactly what happened, I sensed that there was a disturbance in the force so I upgraded from CUDA 2.3 to CUDA 3.1.
____________
|
|
|
|
|
|
OK, I am beginning to understand - I thought I had build the app so it would run on 2.3 drivers, but apparently not. I am waiting from feedback from nvidia re: if it is possible to build a 2.3-compatible app with the 3.1 SDK, and if so I'll post a new app for testing, otherwise everyone will need to update their drivers...
- Iain |
|
|
|
|
|
Well, I have win7 and when i suspended gpugrid, PPS sieve fired up and validated twice. (cuda 2.3) Maybe i can do SoB in two da7ys. yeah right. Has the gurus fixed this problem? or update to cuda 3.1?
thanx, |
|
|
Ken_g6Volunteer developer
Send message
Joined: 4 Jul 06
Posts: 929
ID: 3110
Credit: 237,002,205
RAC: 39,933
|
|
Win7 is fine. It was never affected by this particular issue. Only Macs were. They're stuck using only 3.1 or later drivers; the rest of us can use 2.3 or later.
Otherwise, right now, pretty much everything is working. :)
And, no, you can't do LLR on CUDA yet. Most sieves don't work there (well) either.
____________
|
|
|
Scott BrownVolunteer moderator
Project administrator
Volunteer tester
Project scientist
Send message
Joined: 17 Oct 05
Posts: 2329
ID: 1178
Credit: 15,631,005,591
RAC: 10,487,038
|
...the rest of us can use 2.3 or later.
Not exactly. Some of the earlier 2.3 drivers (at least 191.xx and earlier) have problems. At least 197.xx and beyond work. I have not tested the 195.xx or 196.xx series yet, but given the driver updates that occurred in the 195.xx series, I suspect that this is the windows driver cutpoint for some cards with the Windows PPS Sieve CUDA application.
____________
141941*2^4299438-1 is prime!
|
|
|
|
|
Otherwise, right now, pretty much everything is working. :)
Any idea when EM64T Windows will be available? |
|
|
|
|
Otherwise, right now, pretty much everything is working. :)
Any idea when EM64T Windows will be available?
no need for this - the x86-version does fine on 64bit win.
|
|
|
|
|
Otherwise, right now, pretty much everything is working. :)
Any idea when EM64T Windows will be available?
no need for this - the x86-version does fine on 64bit win.
The server won't send work to Windows x64 Computerid 134966. With "Use NVIDIA GPU"=Yes and "Send work from any subproject..."=Yes, I get
165 PrimeGrid 8/27/2010 09:36:55 Sending scheduler request: To fetch work.
166 PrimeGrid 8/27/2010 09:36:55 Requesting new tasks for GPU
167 PrimeGrid 8/27/2010 09:36:58 Scheduler request completed: got 0 new tasks
168 PrimeGrid 8/27/2010 09:36:58 Message from project server: No work sent
169 PrimeGrid 8/27/2010 09:36:58 Message from project server: No work available for the applications you have selected. Please check your project preferences on the web site.
I've tried resetting the client and detaching/reattaching, with no change.
I assumed it was because no Windows EM64T version showed up on the Applications page.
EDIT--Forgot to mention: the main page says ~20K PPS Sieve available.
____________
|
|
|
 Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13804
ID: 53948
Credit: 345,369,032
RAC: 2,648
                         
|
|
Just a guess here:
BOINC version 6.11.6
and on the BOINC download page:
(MAY BE UNSTABLE - USE ONLY FOR TESTING)
Have you tried rolling back to 6.10.xx? There could be something funky in 6.11.
____________
My lucky number is 75898524288+1 |
|
|
|
|
Just a guess here:
BOINC version 6.11.6
and on the BOINC download page:
(MAY BE UNSTABLE - USE ONLY FOR TESTING)
Have you tried rolling back to 6.10.xx? There could be something funky in 6.11.
Good guess, but I only u/d to the beta a couple of days ago. I've been unable to download Primegrid's GPU work to that machine since 1.27 came out, and I'd been running 6.10.58. I did get a bunch of the previous cuda release, all of which crashed instantly. (That pesky problem with Fermi, apparently.) GPUGRID, Collatz, DNETC (when it's up) and Seti all like that machine/BOINC version fine. But thanks for chiming in. I'm open to any/all suggestions to enable d/l of PPS Sieve cuda.
____________
|
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13804
ID: 53948
Credit: 345,369,032
RAC: 2,648
|
|
ok, guess #2.
If you have this computer assigned to a location (home, school, or work), try setting the default preferences to allow GPU processing, as well as the specific location's preferences.
It's a long shot, but there was some weird behavior with this earlier in the year. It may or may not still be an issue as there have been multiple updates to the server, but it can't hurt to try.
Mike
____________
My lucky number is 75898524288+1 |
|
|
|
|
ok, guess #2.
If you have this computer assigned to a location (home, school, or work), try setting the default preferences to allow GPU processing, as well as the specific location's preferences.
It's a long shot, but there was some weird behavior with this earlier in the year. It may or may not still be an issue as there have been multiple updates to the server, but it can't hurt to try.
Mike
Another good guess, that I also had this morning (California time). Great minds, etc. The box in question is actually location Work, so I set the default location to NVIDIA and Send work from any subproject... = Yes as well. I also reassigned the box to location default and reset it. All same No work available for the applications you have selected. About the only thing I haven't done (that I can think of) is to change the MachineName and bring it in under a completely new profile. And I'm not ready to cross that line yet for what I suspect is a server-side issue.
BTW I'm back with BOINC "recommended version" 6.10.58 again, for reasons completely unrelated ;)
____________
|
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13804
ID: 53948
Credit: 345,369,032
RAC: 2,648
|
|
I'm out of ideas. The only thing left that I can think of is that you're running a Fermi card, but other people are running those successfully.
____________
My lucky number is 75898524288+1 |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13804
ID: 53948
Credit: 345,369,032
RAC: 2,648
|
The server won't send work to Windows x64 Computerid 134966. With "Use NVIDIA GPU"=Yes and "Send work from any subproject..."=Yes, I get
165 PrimeGrid 8/27/2010 09:36:55 Sending scheduler request: To fetch work.
166 PrimeGrid 8/27/2010 09:36:55 Requesting new tasks for GPU
167 PrimeGrid 8/27/2010 09:36:58 Scheduler request completed: got 0 new tasks
168 PrimeGrid 8/27/2010 09:36:58 Message from project server: No work sent
169 PrimeGrid 8/27/2010 09:36:58 Message from project server: No work available for the applications you have selected. Please check your project preferences on the web site.
I've tried resetting the client and detaching/reattaching, with no change.
Following server maintenance at 0800 UTC (4 AM local) I started getting the following:
8/28/2010 6:09:01 AM PrimeGrid Sending scheduler request: To report completed tasks.
8/28/2010 6:09:01 AM PrimeGrid Reporting 1 completed tasks, requesting new tasks for GPU
8/28/2010 6:09:03 AM PrimeGrid Scheduler request completed: got 0 new tasks
8/28/2010 6:09:03 AM PrimeGrid Message from server: No work sent
8/28/2010 6:09:03 AM PrimeGrid Message from server: _("No work available for the applications you have selected. Please check your project preferences on the web site.")
Look familiar?
Before the server maintenance, I was being sent GPU tasks; following the maintenance, nothing. Obvisously, nothing changed on my end since I was asleep at the time, and my preferences on PrimeGrid are likewise unchanged.
It looks like your problem is contagious!
Mike
____________
My lucky number is 75898524288+1 |
|
|
|
|
PrimeGrid 8/27/2010 09:36:58 Message from project server: No work available for the applications you have selected. Please check your project preferences on the web site.
8/28/2010 6:09:03 AM PrimeGrid Message from server: _("No work available for the applications you have selected. Please check your project preferences on the web site.")
It's the same thing here: No GPU WUs since the maintenance. Looking at the translation issues (the _("") macro in the second message) I guess that somebody installed a new server version...
Update: I changed the browser language to "en-us", switched off the CPU usage in the preferences and got new GPU tasks...
Update 2: After I switched the CPU usage back on in the preferences the problem reappeared...
Update 3: The problem is independent of the preferred language chosen in the browser...
____________
|
|
|
|
|
PrimeGrid 8/27/2010 09:36:58 Message from project server: No work available for the applications you have selected. Please check your project preferences on the web site.
8/28/2010 6:09:03 AM PrimeGrid Message from server: _("No work available for the applications you have selected. Please check your project preferences on the web site.")
It's the same thing here: No GPU WUs since the maintenance. Looking at the translation issues (the _("") macro in the second message) I guess that somebody installed a new server version...
Update: I changed the browser language to "en-us", switched off the CPU usage in the preferences and got new GPU tasks...
Update 2: After I switched the CPU usage back on in the preferences the problem reappeared...
Update 3: The problem is independent of the preferred language chosen in the browser...
I tried Ralf's "CPU=No" workaround, and it had no effect on GPU downloads. But I'm pretty sure by now that my machine's profile is somehow broken in the server. At least Ralf's find can give Rytis a clue to where the recent changes went wrong for the community as a whole.
____________
|
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13804
ID: 53948
Credit: 345,369,032
RAC: 2,648
|
I tried Ralf's "CPU=No" workaround, and it had no effect on GPU downloads. But I'm pretty sure by now that my machine's profile is somehow broken in the server. At least Ralf's find can give Rytis a clue to where the recent changes went wrong for the community as a whole.
I tried setting CPU to NO and that restored the GPU task downloads. I flipped the CPU switch on and off, and sure enough, that is what's effecting the flow of WUs.
Since there appears to be a 20 task limit, that means I can grab a 5 hour supply of work.
Unfortunately, this doesn't bring us any closer to finding out why your machine still isn't getting anything.
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
Can you test if you get work now ?
Lennart |
|
|
|
|
|
I am having the exact same problem with this computer:
http://www.primegrid.com/show_host_detail.php?hostid=111396
Cheers!
Alan
____________
  |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13804
ID: 53948
Credit: 345,369,032
RAC: 2,648
|
|
EDIT: Hmmmm -- I'm not sure to whom your question was directed, but in case it was me, here's my answer:
Can you test if you get work now ?
Lennart
I was out for a few hours, but considering this is the last entries in my log...
8/28/2010 4:17:55 PM PrimeGrid Sending scheduler request: To fetch work.
8/28/2010 4:17:55 PM PrimeGrid Requesting new tasks for GPU
8/28/2010 4:17:56 PM PrimeGrid Scheduler request completed: got 0 new tasks
8/28/2010 4:17:56 PM PrimeGrid Message from server: No work sent
8/28/2010 4:17:56 PM PrimeGrid Message from server: _("No work available for the applications you have selected. Please check your project preferences on the web site.")
... I'd say yes, the problem still exists.
Log times are UTC -4:00.
____________
My lucky number is 75898524288+1 |
|
|
|
|
Can you test if you get work now ?
Lennart
If you meant me, yes I can test and no, no change in status ;(
____________
|
|
|
LumiukkoVolunteer tester
Send message
Joined: 7 Jul 08
Posts: 165
ID: 25183
Credit: 785,041,078
RAC: 41,744
               
|
|
Same problem also here. It was working yesterday. Last Cuda workunit that I received was sent at "28 Aug 2010 7:15:48 UTC" from server. After that I can only get GPU-work if I change "Use CPU" to "No" in Primegrid preferences.
--
Lumiukko |
|
|
MenipeVolunteer tester
Send message
Joined: 2 Jan 08
Posts: 235
ID: 17041
Credit: 112,615,746
RAC: 2,534
         
|
|
Same here. I was receiving CUDA work units yesterday, but none since this morning.
____________
 |
|
|
LumiukkoVolunteer tester
Send message
Joined: 7 Jul 08
Posts: 165
ID: 25183
Credit: 785,041,078
RAC: 41,744
|
Same problem also here. It was working yesterday. Last Cuda workunit that I received was sent at "28 Aug 2010 7:15:48 UTC" from server. After that I can only get GPU-work if I change "Use CPU" to "No" in Primegrid preferences.
--
Lumiukko
Now it is working, GPU work is received again.
--
Lumiukko |
|
|
|
|
Same problem also here. It was working yesterday. Last Cuda workunit that I received was sent at "28 Aug 2010 7:15:48 UTC" from server. After that I can only get GPU-work if I change "Use CPU" to "No" in Primegrid preferences.
--
Lumiukko
Now it is working, GPU work is received again.
--
Lumiukko
It appears that System Administration backed out the server changes that caused other folks not to be able to get GPU work. (Good news ;) ) I'm still unable to get GPU tasks to my Computerid 134966 (Bad news for me :'( )
Since the community-wide d/l difficulty started, I've removed my Home/School/Work locations & placed all hosts (GPU & CPU-only) on the default location with the Use CPU, Use NVIDIA GPU, Send work from any subproject... options all =Yes. I still get:
1702 PrimeGrid 8/29/2010 02:58:55 Sending scheduler request: To fetch work.
1703 PrimeGrid 8/29/2010 02:58:55 Requesting new tasks for GPU
1704 PrimeGrid 8/29/2010 02:58:59 Scheduler request completed: got 0 new tasks
1705 PrimeGrid 8/29/2010 02:58:59 Message from server: No work sent
1706 PrimeGrid 8/29/2010 02:58:59 Message from server: _("No work available for the applications you have selected. Please check your project preferences on the web site.")
I'm thinking there's some flag set on that machine's profile that inhibits sending GPU work (like Invalid GPU type or something of the sort).
Rytis already had me verify that I had the 64-bit version of BOINC installed, not the 32-bit version by mistake. (That wasn't the problem.) I'm interested to know if anyone else is in the same boat as me, now that the larger problem has been solved.
____________
|
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13804
ID: 53948
Credit: 345,369,032
RAC: 2,648
|
Same problem also here. It was working yesterday. Last Cuda workunit that I received was sent at "28 Aug 2010 7:15:48 UTC" from server. After that I can only get GPU-work if I change "Use CPU" to "No" in Primegrid preferences.
--
Lumiukko
Now it is working, GPU work is received again.
--
Lumiukko
Not for me. I stopped getting GPU tasks yesterday and have NOT started receiving them again.
EDIT: It looks like yesterday was the last cool day for a while; starting today and continuing throughout the week it's going to be pretty hot again. So, no more GPU for me crunching till the end of the week, at least. Hopefully this problem will be fixed by then.
____________
My lucky number is 75898524288+1 |
|
|
RytisVolunteer moderator
Project administrator
Send message
Joined: 22 Jun 05
Posts: 2653
ID: 1
Credit: 86,414,420
RAC: 7
                  
|
|
We have some server debug output (thanks to jjwhalen), and I have emailed David to help pinpoint a problem. Since he's probably on the way to Europe for BOINC workshop already, it might take a few days to get a fix. If he doesn't fix it quick, I'll nag him in person during the workshop :)
____________
|
|
|
|
|
|
Heres my two cents,
yesterday everything worked fine. 5 cores cpu and (.64 core + gpu)
this morning, no gpu running- reset to NO cpu in preferences and d/l and everything ok, but this afternoon only gpu running no 5 cores, huh??
Went back to pref to reset YES to cpu and d/l again and 5 cores fired up.
(gpu was already running) Do some WU's only code for gpu and some ONLY for cpu it seems???
Well i've spent my two cents. bye |
|
|
|
|
|
I have been running GPU fine untill this morning. Now no matter what I do I keep getting the mesage no work available. I have even tried setting Proth Prime Sieve as my only project. My CPUs are running long jobs with no change.
____________
May your addiction to Boinc be greater than mine!
End Transmission!
 |
|
|
axnVolunteer developer
Send message
Joined: 29 Dec 07
Posts: 285
ID: 16874
Credit: 28,027,106
RAC: 0
      
|
|
Is this issue fixed? I am not getting any GPU WUs with the regular setting. Need to do the "CPU off" workaround.
And the computer keeps getting duplicated in my account. Had to do merge many times. What is going on here? |
|
|
|
|
Is this issue fixed? I am not getting any GPU WUs with the regular setting. Need to do the "CPU off" workaround.
And the computer keeps getting duplicated in my account. Had to do merge many times. What is going on here?
Please see MessageID 26084 from John in thread Proth GPU work. We're waiting for a server fix from David Anderson at BOINC Central :)
B/w |
|
|
|
|
We have some server debug output (thanks to jjwhalen), and I have emailed David to help pinpoint a problem. Since he's probably on the way to Europe for BOINC workshop already, it might take a few days to get a fix. If he doesn't fix it quick, I'll nag him in person during the workshop :)
Any update on this?
____________
|
|
|
JohnHonorary cruncher
Send message
Joined: 21 Feb 06
Posts: 2875
ID: 2449
Credit: 2,681,934
RAC: 0
            
|
We have some server debug output (thanks to jjwhalen), and I have emailed David to help pinpoint a problem. Since he's probably on the way to Europe for BOINC workshop already, it might take a few days to get a fix. If he doesn't fix it quick, I'll nag him in person during the workshop :)
Any update on this?
Interesting you should ask. :) A patch was applied earlier today but it has not been confirmed that it is working. Please feel free to check to see if it is and report back.
____________
|
|
|
|
|
We have some server debug output (thanks to jjwhalen), and I have emailed David to help pinpoint a problem. Since he's probably on the way to Europe for BOINC workshop already, it might take a few days to get a fix. If he doesn't fix it quick, I'll nag him in person during the workshop :)
Any update on this?
Interesting you should ask. :) A patch was applied earlier today but it has not been confirmed that it is working. Please feel free to check to see if it is and report back.
The reason I asked is because I noticed the "filter tasks by subproject" functions had returned, and I knew that was being held up pending the fix for this issue ;)
Other GPU crunchers should be asked to check whether they can now d/l PPS GPU work without the "Use CPU" option being reset/unchecked. That was the situation for the "general population."
As for myself, there is no change in symptom -- I'm still unable to get GPU tasks, period. Rytis knew from our debug testing that my quad CPU/GTX 465 computerID 134966 is in an apparently unique condition where all requests for GPU work are denied. That appears unchanged by this patch. Rytis said that I wouldn't have to restart my client to see the fix, I would just suddenly begin to see GPU downloads. That didn't happen. I restarted the client, but still no change. I also reset the project on that client -- I got new CPU work, but the GPU request was denied with
263 PrimeGrid 9/13/2010 02:50:26 Requesting new tasks for GPU
264 PrimeGrid 9/13/2010 02:50:33 Scheduler request completed: got 0 new tasks
265 PrimeGrid 9/13/2010 02:50:33 Message from server: No work sent
266 PrimeGrid 9/13/2010 02:50:33 Message from server: _("No work available for the applications you have selected. Please check your project preferences on the web site.")
It's as though Send work from any subproject if selected projects have no work isn't checked, even though it is.
So what do we try next, folks? |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13804
ID: 53948
Credit: 345,369,032
RAC: 2,648
|
We have some server debug output (thanks to jjwhalen), and I have emailed David to help pinpoint a problem. Since he's probably on the way to Europe for BOINC workshop already, it might take a few days to get a fix. If he doesn't fix it quick, I'll nag him in person during the workshop :)
Any update on this?
Interesting you should ask. :) A patch was applied earlier today but it has not been confirmed that it is working. Please feel free to check to see if it is and report back.
It seems to be working, for me at least.
With CPU ON but "any tasks" off:
9/13/2010 10:29:05 AM PrimeGrid Sending scheduler request: To fetch work.
9/13/2010 10:29:05 AM PrimeGrid Requesting new tasks for GPU
9/13/2010 10:29:07 AM PrimeGrid Scheduler request completed: got 0 new tasks
9/13/2010 10:29:07 AM PrimeGrid Message from server: No work sent
9/13/2010 10:29:07 AM PrimeGrid Message from server: _("No work available for the applications you have selected. Please check your project preferences on the web site.")
That's as expected since "any tasks" was off.
Turning on the "any tasks" toggle and leaving CPU turned on:
9/13/2010 10:29:23 AM PrimeGrid update requested by user
9/13/2010 10:29:27 AM PrimeGrid Sending scheduler request: Requested by user.
9/13/2010 10:29:27 AM PrimeGrid Requesting new tasks for GPU
9/13/2010 10:29:28 AM PrimeGrid Scheduler request completed: got 21 new tasks
9/13/2010 10:29:28 AM PrimeGrid Message from server: No work can be sent for the applications you have selected
9/13/2010 10:29:28 AM PrimeGrid Message from server: Your preferences allow work from applications other than those selected
9/13/2010 10:29:28 AM PrimeGrid Message from server: Sending work from other applications
Of course, there's still a bunch of major design flaws in BOINC's handling of GPU/CPU scheduling (being able to select sub projects independently and having differing time compensation factors), but being able to once again download both CPU and GPU tasks is a big improvement.
Thanks!
____________
My lucky number is 75898524288+1 |
|
|
|
|
|
i had exactly the same message on my linux server some days ago - and it does not even have a GPU.. |
|
|
RytisVolunteer moderator
Project administrator
Send message
Joined: 22 Jun 05
Posts: 2653
ID: 1
Credit: 86,414,420
RAC: 7
|
It's as though Send work from any subproject if selected projects have no work isn't checked, even though it is.
This shouldn't be relevant, because you still have PPS Sieve selected in your prefs...
____________
|
|
|
|
|
It's as though Send work from any subproject if selected projects have no work isn't checked, even though it is.
This shouldn't be relevant, because you still have PPS Sieve selected in your prefs...
yup - it should work, but that machine did not get a single WU.
should i test again?
|
|
|
LookASVolunteer tester
Send message
Joined: 19 Apr 08
Posts: 38
ID: 21649
Credit: 354,890,618
RAC: 0
  
|
|
I cant get any work for GPU no matter what I do. I still get this as well:
13-Sep-10 18:30:45 PrimeGrid Sending scheduler request: To fetch work.
13-Sep-10 18:30:45 PrimeGrid Requesting new tasks for GPU
13-Sep-10 18:30:46 PrimeGrid Scheduler request completed: got 0 new tasks
13-Sep-10 18:30:46 PrimeGrid Message from server: No work sent
13-Sep-10 18:30:46 PrimeGrid Message from server: _("No work available for the applications you have selected. Please check your project preferences on the web site.")
Win7U 64, GTX470, FW258.96, BOINC 6.10.58 |
|
|
|
|
|
ok, changed host to "home" - "home" is set to run Proth Prime Search (LLR)
now i'm getting this:
Mo 13 Sep 2010 18:52:08 CEST PrimeGrid Requesting new tasks
Mo 13 Sep 2010 18:52:13 CEST PrimeGrid Scheduler request completed: got 0 new tasks
Mo 13 Sep 2010 18:52:13 CEST PrimeGrid Message from server: No work sent
Mo 13 Sep 2010 18:52:13 CEST PrimeGrid Message from server: PPS LLR _("is not available for your type of computer").
Mo 13 Sep 2010 18:52:13 CEST PrimeGrid Message from server: _("No work available for the applications you have selected. Please check your project preferences on the web site.")
Mo 13 Sep 2010 18:59:14 CEST PrimeGrid Resetting project
Mo 13 Sep 2010 18:59:17 CEST PrimeGrid Sending scheduler request: To fetch work.
Mo 13 Sep 2010 18:59:17 CEST PrimeGrid Requesting new tasks
Mo 13 Sep 2010 18:59:22 CEST PrimeGrid Scheduler request completed: got 0 new tasks
Mo 13 Sep 2010 18:59:22 CEST PrimeGrid Message from server: No work sent
Mo 13 Sep 2010 18:59:22 CEST PrimeGrid Message from server: PPS LLR _("is not available for your type of computer").
Mo 13 Sep 2010 18:59:22 CEST PrimeGrid Message from server: _("No work available for the applications you have selected. Please check your project preferences on the web site.")
Mo 13 Sep 2010 18:59:24 CEST PrimeGrid Started download of stat_primegrid.png
Mo 13 Sep 2010 19:00:04 CEST PrimeGrid Detaching from project
Mo 13 Sep 2010 19:00:19 CEST Fetching configuration file from http://www.primegrid.com/get_project_config.php
Mo 13 Sep 2010 19:00:37 CEST PrimeGrid Master file download succeeded
Mo 13 Sep 2010 19:00:43 CEST PrimeGrid Sending scheduler request: Project initialization.
Mo 13 Sep 2010 19:00:43 CEST PrimeGrid Requesting new tasks
Mo 13 Sep 2010 19:00:48 CEST PrimeGrid Scheduler request completed: got 0 new tasks
Mo 13 Sep 2010 19:00:48 CEST PrimeGrid Message from server: No work sent
Mo 13 Sep 2010 19:00:48 CEST PrimeGrid Message from server: PPS LLR _("is not available for your type of computer").
Mo 13 Sep 2010 19:00:48 CEST PrimeGrid Message from server: _("No work available for the applications you have selected. Please check your project preferences on the web site.")
weirdo!
you don't like that host anymore.. ;) |
|
|
Michael GoetzVolunteer moderator
Project administrator
Send message
Joined: 21 Jan 10
Posts: 13804
ID: 53948
Credit: 345,369,032
RAC: 2,648
|
|
I've got a hunch, so, if you're not able to get GPU tasks...
Please post what kind of GPU you have.
I noticed that two of the computers not getting tasks are a 465 and a 470. It seems like the 460 is the most popular Fermi, so perhaps there's something about the 465 and 470 that makes BOINC or PrimeGrid think they can't run the application?
____________
My lucky number is 75898524288+1 |
|
|
|
|
I've got a hunch, so, if you're not able to get GPU tasks...
Please post what kind of GPU you have.
I noticed that two of the computers not getting tasks are a 465 and a 470. It seems like the 460 is the most popular Fermi, so perhaps there's something about the 465 and 470 that makes BOINC or PrimeGrid think they can't run the application?
It's true that my computerID 134966 is a 465SC, and Mike & I had a backchannel discussion about this very subject a couple weeks ago. I have exactly the same hunch that he does. I'm almost tempted to prove it by swapping my old GTS 250 back into that platform. But since 5 other projects (Seti/Seti Beta, Collatz, DNETC & GPUGRID) are providing GPU work to that same host by the tonne, not so much. (Of those, Seti/Seti Beta are definitely using the new server release, while I'm pretty certain the other 3 haven't yet "upgraded" {and I use that term in its technical sense only LOL}.)
[EDIT]When the original Windows PPS Sieve Cuda application was released to Primegrid/BOINC, I was definitely getting work then. All WUs crashed immediately, but that's another story. However since v1.27 was released, nothing.[/EDIT] |
|
|
|
|
It's as though Send work from any subproject if selected projects have no work isn't checked, even though it is.
This shouldn't be relevant, because you still have PPS Sieve selected in your prefs...
NOW, yes. But when I drafted that post, no. I changed later because I want to get PPS Sieve CPU tasks for that host. Either way, no change in symptom. Same with "Use CPU" checked or unchecked. I've also tried deleting Home School & Work altogether & running all hosts off the default prefs. All same no joy in Mudville (obscure U.S. cultural reference) :(
B/w. |
|
|
|
|
ok, changed host to "home" - "home" is set to run Proth Prime Search (LLR)
.
.
.
.
Mo 13 Sep 2010 19:00:48 CEST PrimeGrid Message from server: PPS LLR _("is not available for your type of computer").
Mo 13 Sep 2010 19:00:48 CEST PrimeGrid Message from server: _("No work available for the applications you have selected. Please check your project preferences on the web site.")
weirdo!
you don't like that host anymore.. ;)
Is this about the Linux server without GPU?
Are you running Linux64?
Is this the first time you tried running LLR's on that machine?
If so, possibly you have you have to install the 32bit libraries.
____________
 |
|
|
|
|
Is this about the Linux server without GPU?
Are you running Linux64?
Is this the first time you tried running LLR's on that machine?
yes, yes and no.
it's ubuntu 10.04.
it has a new cpid now, but it has been running PG for quite a while.
If so, possibly you have you have to install the 32bit libraries.
ia32 is installed.. |
|
|
|
|
|
ok, i can get sieve_wu's on this host, but no LLR's.
i tried PPS and SGS...
looks like the current server-code does no longer want to send 32-bit work to that 64-bit host.
http://www.primegrid.com/show_host_detail.php?hostid=156218 |
|
|
|
|
|
I had this problem too awhile back.
http://www.primegrid.com/forum_thread.php?id=2561#24014
This is what solved it above.
Jack |
|
|
|
|
I had this problem too awhile back.
http://www.primegrid.com/forum_thread.php?id=2561#24014
This is what solved it above.
Jack
i know, but they are all installed. maybe uninstall and reinstall will help..
[]edit]
ok, that did the job!
somehow something must have gone defunct.
thank's for pointing me to i! |
|
|
