BOINC DEV or A SETI problem.

Message boards : Number crunching : BOINC DEV or A SETI problem.
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Bob Bainbridge

Send message
Joined: 21 Mar 01
Posts: 63
Credit: 128,919,333
RAC: 82
United States
Message 1369522 - Posted: 19 May 2013, 3:25:59 UTC
Last modified: 19 May 2013, 3:27:21 UTC

I have one rig with an Intel 970 running 12 cpu's with a GTX 480 GPU.(currently dead, awaiting a new one). It normally downloads WU's so it has 100-150 ready to run. I have my preferences set for a 10-day supply. My second rig has an AMD 4-core cpu with a GTX 570 GPU. The downloaded WU's ready to run reached over 3000 at one point. Now I am getting errors from WU's failing to start by deadline. What would cause this?

Thanx, Bob B.
ID: 1369522 · Report as offensive
Profile Fred E.
Volunteer tester

Send message
Joined: 22 Jul 99
Posts: 768
Credit: 24,140,697
RAC: 0
United States
Message 1369652 - Posted: 19 May 2013, 11:16:45 UTC

I have one rig with an Intel 970 running 12 cpu's with a GTX 480 GPU.(currently dead, awaiting a new one). It normally downloads WU's so it has 100-150 ready to run. I have my preferences set for a 10-day supply. My second rig has an AMD 4-core cpu with a GTX 570 GPU. The downloaded WU's ready to run reached over 3000 at one point. Now I am getting errors from WU's failing to start by deadline. What would cause this?

Don't know why they are both over the project's task limits. Suggest you lower the cache setting and go on No New Tasks (NNT), especially for the one with over 2k tasks. It is still receiving work, includiong some after your post. BOINC will report completions promptly on NNT while you work down the task count. Suggest you open a thread to discuss as this is not the right place for a discussion. I know people that would love to know how this happened!
Another Fred
Support SETI@home when you search the Web with GoodSearch or shop online with GoodShop.
ID: 1369652 · Report as offensive
Claggy
Volunteer tester

Send message
Joined: 5 Jul 99
Posts: 4654
Credit: 47,537,079
RAC: 4
United Kingdom
Message 1369654 - Posted: 19 May 2013, 11:22:23 UTC - in response to Message 1369522.  
Last modified: 19 May 2013, 11:23:46 UTC

I have one rig with an Intel 970 running 12 cpu's with a GTX 480 GPU.(currently dead, awaiting a new one). It normally downloads WU's so it has 100-150 ready to run. I have my preferences set for a 10-day supply. My second rig has an AMD 4-core cpu with a GTX 570 GPU. The downloaded WU's ready to run reached over 3000 at one point. Now I am getting errors from WU's failing to start by deadline. What would cause this?

Thanx, Bob B.

I suggest you leave your settings as they are, so we can report this to the Boinc Devs/Setiathome, then we can get what your preferences are when this occured, change them and the problem may go away, and we may not be able to replicate it.

Claggy
ID: 1369654 · Report as offensive
Profile Cliff Harding
Volunteer tester
Avatar

Send message
Joined: 18 Aug 99
Posts: 1432
Credit: 110,967,840
RAC: 67
United States
Message 1369841 - Posted: 19 May 2013, 23:14:30 UTC - in response to Message 1369654.  

I have one rig with an Intel 970 running 12 cpu's with a GTX 480 GPU.(currently dead, awaiting a new one). It normally downloads WU's so it has 100-150 ready to run. I have my preferences set for a 10-day supply. My second rig has an AMD 4-core cpu with a GTX 570 GPU. The downloaded WU's ready to run reached over 3000 at one point. Now I am getting errors from WU's failing to start by deadline. What would cause this?

Thanx, Bob B.

I suggest you leave your settings as they are, so we can report this to the Boinc Devs/Setiathome, then we can get what your preferences are when this occured, change them and the problem may go away, and we may not be able to replicate it.

Claggy


I have had my preferences set to 10 & 10 way before the current limits were put in place and never had a problem maintaining the current limits. Could something have gotten corrupted on your AMD machine?


I don't buy computers, I build them!!
ID: 1369841 · Report as offensive
Profile Bob Bainbridge

Send message
Joined: 21 Mar 01
Posts: 63
Credit: 128,919,333
RAC: 82
United States
Message 1369892 - Posted: 20 May 2013, 4:07:16 UTC

I still have about 2200 tasks with many due tomorrow (5-21-13). I set it to NO NEW TASKS until it tapers off. The downloads have always been strange with the 2 GTX GPU machines. For a week at a time one would receive mainly CPU tasks and no GU and the other machine would receive all GPU tasks.

Bob B.
ID: 1369892 · Report as offensive
Profile Bob Bainbridge

Send message
Joined: 21 Mar 01
Posts: 63
Credit: 128,919,333
RAC: 82
United States
Message 1370156 - Posted: 21 May 2013, 0:21:37 UTC

I'm now down to 1500 tasks but out of CUDA work. Is there any way to download only CUDA WU's and no CPU units?

Bob B.
ID: 1370156 · Report as offensive
Profile Fred E.
Volunteer tester

Send message
Joined: 22 Jul 99
Posts: 768
Credit: 24,140,697
RAC: 0
United States
Message 1370168 - Posted: 21 May 2013, 1:01:12 UTC
Last modified: 21 May 2013, 1:02:43 UTC

I'm now down to 1500 tasks but out of CUDA work. Is there any way to download only CUDA WU's and no CPU units?

Website project preferences - turn off "use cpu". This only affects work fetch, not the tasks you have on hand. That pref will pass to BOINC and BOINC won't ask for cpu work. This will affect all your machines unless you are already using location codes and separate preferences. If you want to only do it for one, you'll need to establish a separate set of project prefs (home, school or work), and then designate that computer's location on that computer's details page.

Looks like you've added a gpu today. I like my 670, happy crunching!
Another Fred
Support SETI@home when you search the Web with GoodSearch or shop online with GoodShop.
ID: 1370168 · Report as offensive
Profile Bob Bainbridge

Send message
Joined: 21 Mar 01
Posts: 63
Credit: 128,919,333
RAC: 82
United States
Message 1370183 - Posted: 21 May 2013, 3:13:07 UTC - in response to Message 1370168.  

Thanks for the info. Yes I got the 670 to replace an EVGA GTX-480 that failed last week. Computer failed to recognize it, like it wasn't there - no display. Second EVGA GTX-480 that failed with same symptom. I replaced the first one with MSI GTX-570. I bought both 480's used on eBay so no warranty. I won't make that mistake again. 670 was about $30 more new from NewEgg but now I have a warranty.

Bob B.
ID: 1370183 · Report as offensive
Profile Bob Bainbridge

Send message
Joined: 21 Mar 01
Posts: 63
Credit: 128,919,333
RAC: 82
United States
Message 1370622 - Posted: 22 May 2013, 16:00:21 UTC - in response to Message 1370183.  

The MSI GTX-670 ran about 5 hours before crapping out. It failed just like the 480 - not recognized. I'm now paying $15 to ship it back for a replacement. I hope the replacement lasts longer. I'm glad I didn't remove the UPC for the rebate yet. NewEgg requires all UPC's to be present for an RMA.

Bob B.
ID: 1370622 · Report as offensive
Profile Fred E.
Volunteer tester

Send message
Joined: 22 Jul 99
Posts: 768
Credit: 24,140,697
RAC: 0
United States
Message 1370642 - Posted: 22 May 2013, 16:55:27 UTC

The MSI GTX-670 ran about 5 hours before crapping out. It failed just like the 480 - not recognized. I'm now paying $15 to ship it back for a replacement. I hope the replacement lasts longer. I'm glad I didn't remove the UPC for the rebate yet. NewEgg requires all UPC's to be present for an RMA.

Bob B.
That's a bummer. Hope it was just a bad card. Anyone have any ideas on things Bob could check before installing another card? Or any ideas on why some of his hosts went over limits?


Another Fred
Support SETI@home when you search the Web with GoodSearch or shop online with GoodShop.
ID: 1370642 · Report as offensive
kittyman Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Jul 00
Posts: 51468
Credit: 1,018,363,574
RAC: 1,004
United States
Message 1370947 - Posted: 23 May 2013, 15:28:29 UTC
Last modified: 23 May 2013, 15:39:01 UTC

Just a couple of thoughts.
I am guessing there is nothing wrong with the 670.
And maybe the 480 either.

Here's why......

I have one of my better computers that used to have a 590 in it.
The 590 did indeed go belly up. Shorted out, and that was confirmed by trying it in a couple of other rigs.

I then bought a pair of 680s. Neither would show any video when booting the rig. Started to think it was a 680 thingy and gave up. Very weird, I could get a 560 to work as the primary card with a 680 in the second slot.
I eventually swapped both 680s to another rig, where they work flawlessly.

OK, round 2. I later installed a pair of 580s in the problem rig. Worked fine for a couple of weeks. Had a power glitch one day that rebooted or hung most of my rigs. The next time I rebooted the dual 580 rig, I got video from the bios and through the Windows splash screen, but video output drops dead when Windows actually loads (XP x64). I can maintain video in safe mode, and reloaded the video drivers in safe mode, but when booting normally, video still drops dead. It takes a long time to boot, but does eventually finish and crunching resumes normally, so it's still running that way.
I know it's not a driver problem, as I do a clean install when installing NV drivers.

I am suspecting a corruption of Windows, and may attempt a repair installation, but for now it is crunching full bore, and time has been short so I have not been able to confirm my suspicion. Or could possibly be some mobo/RAM demon raising it's head.

Test both the 480 and 670 in a different computer before writing them off as no good.
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 1370947 · Report as offensive
Profile Bob Bainbridge

Send message
Joined: 21 Mar 01
Posts: 63
Credit: 128,919,333
RAC: 82
United States
Message 1371178 - Posted: 24 May 2013, 3:18:55 UTC - in response to Message 1370947.  

I have 2 EVGA 480's both purchased used on eBay. They were installed in 2 different computers, one running Win7 32-bit and the other Win7-64-bit. They both fail in 3 different computers, one doesn't even run the fan. I replaced the one in the 64-bit machine with a MSI 570 card, which has been running close to a year. The MSI 670 configured OK and ran for 5 hours. I thought the computer had powered down because I lost remote console communication via Logmein and found it had a power LED but nothing on the monitor. I rebooted and got no display. I added a second old video card to get a display and booted Win7. The Device Manager showed only the old card and not the 670. I tried the 670 in third Win7 64-bit machine and got nothing. It's already on its way back to NewEgg for replacement. At $15 unreimbursed return shipping, I hope the next one works better. I have found, in the past, I had to clear the bios before the POST would recognize the card.

Bob B.
ID: 1371178 · Report as offensive
Profile Bob Bainbridge

Send message
Joined: 21 Mar 01
Posts: 63
Credit: 128,919,333
RAC: 82
United States
Message 1371182 - Posted: 24 May 2013, 3:24:11 UTC - in response to Message 1370168.  

I'm now down to 1500 tasks but out of CUDA work. Is there any way to download only CUDA WU's and no CPU units?

Website project preferences - turn off "use cpu". This only affects work fetch, not the tasks you have on hand. That pref will pass to BOINC and BOINC won't ask for cpu work. This will affect all your machines unless you are already using location codes and separate preferences. If you want to only do it for one, you'll need to establish a separate set of project prefs (home, school or work), and then designate that computer's location on that computer's details page.

Looks like you've added a gpu today. I like my 670, happy crunching!

I've been turning off CPU in the preferences to stop the downloads, turning it on only long enough to get some WU's for my other machines.
Is there any way to copy WU's to a different computer so I could transfer most of them to my machine running 12 cpu's which is currently without a GPU until I receive a replacement 670?

Bob B.
ID: 1371182 · Report as offensive
rob smith Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer moderator
Volunteer tester

Send message
Joined: 7 Mar 03
Posts: 22180
Credit: 416,307,556
RAC: 380
United Kingdom
Message 1371202 - Posted: 24 May 2013, 5:18:24 UTC
Last modified: 24 May 2013, 5:23:43 UTC

There is no way to transfer WU between machines - they are allocated to a cruncher when dispatched from the servers, and expect to see that cruncher's id in the returned results.

However if you move the ENTIRE project folder from one machine to another the "new" machine will crunch the data, but will appear to be the old machine. This may cause problems if the hardware on the two machines is different.
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?
ID: 1371202 · Report as offensive

Message boards : Number crunching : BOINC DEV or A SETI problem.


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.