Report all problems here...........

Message boards : Number crunching : Report all problems here...........
Message board moderation

To post messages, you must log in.

Previous · 1 · 2 · 3 · 4 · 5 · 6 · 7 · 8 . . . 10 · Next

AuthorMessage
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13720
Credit: 208,696,464
RAC: 304
Australia
Message 1014023 - Posted: 10 Jul 2010, 0:07:43 UTC - in response to Message 1014014.  

Sorting the Wheat from the Chaff has always been a difficult & laborious process.

So? What's the best response to that (very real) problem?

Recruit some knowledgeable volunteer Wheat/Chaff sorters, and give them the tools to undertake the task?

Or throw everything - Wheat and Chaff alike - onto the rubbish tip?

I'd be a big job for just one person when things get hectic, but if there's more than one they'd need to co-ordinate with each other so they're not reporting the same thing, and missing others.

But it would be good if a couple of people that are in contact with the staff directly were able to give them a summary of what the discussion is about, and a copy the more usefull/informative posts, considering how quickly things get burried amongst the noise.
If it gets used it gets used, if not *shrug*. Once they've got the information, it's up to them to prioritise things.
Grant
Darwin NT
ID: 1014023 · Report as offensive
Profile Eduardo Marino Rodriguez

Send message
Joined: 8 Sep 99
Posts: 11
Credit: 192,922
RAC: 0
Venezuela
Message 1014024 - Posted: 10 Jul 2010, 0:08:20 UTC

Hello everyone. After return to work on SETI @ Home this week I noticed that the graphics suddenly ceased to function after starting work at WU 05no09af.22632.4980.11.10.168, received on 07/06/2010. I am currently running Seti @ home Enhanced 6.03. There has been no change in equipment or software in a few days ago ... there will be some problem with the program? or does it have something to do with the type of WU?
Thanks.
ID: 1014024 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13720
Credit: 208,696,464
RAC: 304
Australia
Message 1014025 - Posted: 10 Jul 2010, 0:11:33 UTC - in response to Message 1014019.  

I see people mentioning a 5 work unit per CPU limit but isn't that going to starve the mega crunchers?

Nope. They'll still get work, just not be able to build up their cache while the limit is in place.


Is this new per CPU quota get kicked up as we get closer to the outage?

That's what many have suggested. At this stage the limit stays in place over the weekend & then come Monday it's removed. But all things are subject to change.


Is this some measure to prevent orphan work units from systems that fetch 20 days worth then are never seen again or have some catastrophic hardware problem?

Has been for years, although it gets tweaked occasionally.
Grant
Darwin NT
ID: 1014025 · Report as offensive
JohnDK Crowdfunding Project Donor*Special Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 28 May 00
Posts: 1222
Credit: 451,243,443
RAC: 1,127
Denmark
Message 1014026 - Posted: 10 Jul 2010, 0:14:15 UTC - in response to Message 1014025.  
Last modified: 10 Jul 2010, 0:15:03 UTC

Is this new per CPU quota get kicked up as we get closer to the outage?

That's what many have suggested. At this stage the limit stays in place over the weekend & then come Monday it's removed. But all things are subject to change.

I just got 10 new WUs and have 150 now, guess they may have started raising the limit slowly.
ID: 1014026 · Report as offensive
Rich-E
Volunteer tester

Send message
Joined: 23 Feb 01
Posts: 41
Credit: 1,685,487
RAC: 0
United States
Message 1014035 - Posted: 10 Jul 2010, 0:43:25 UTC
Last modified: 10 Jul 2010, 0:48:41 UTC

Good evening.

Is this where I post issues with downloading CUDA job packets? My first attempt at CUDA processing since upgrading from MSI 8600GT to a EVA GTX480. The work unit hangs at 4.41% 0.53/11.90MB and appears to start to create 19 tasks. Tried aborting and retying, but to no available. I want to see if is just a bad packet, but no way to move on. I have a bunch of CPU SETI, but the whole idea of the GTX480 was for BOINC GPU processing.

File is named cufft32_30_14.dll. I'd really like to get it running and see what the baby can do.

From message log - Temporarily failed download of cufft32_30_14.dll: HTTP error.

Any ideas or some other board I should post to?

Thanks and have a good day.

Rich
ID: 1014035 · Report as offensive
Profile Ed Shearin Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor

Send message
Joined: 20 May 99
Posts: 7
Credit: 28,980,031
RAC: 25
United States
Message 1014041 - Posted: 10 Jul 2010, 1:00:31 UTC

The task limit is somewhat annoying, but I wouldn't be close to running out of work if it were actually 20 (or is the new limit now 5 as a post below suggests) and the website was accurate about what I have returned. One machine (ES-1) keeps hitting the limit in asking for new work, but has only 12 work units. The website shows it has 28 outstanding, however, so something is wildly off. Is it taking a while for work units returned in today's large burst to get updated? The only other computer I checked is being denied with 15 work units, but at least the website shows only that many outstanding for it. Is anyone else seeing problems like this?
Ed
ensmail01@verizon.net
ID: 1014041 · Report as offensive
Profile Zeus Fab3r
Avatar

Send message
Joined: 17 Jan 01
Posts: 649
Credit: 275,335,635
RAC: 597
Serbia
Message 1014042 - Posted: 10 Jul 2010, 1:02:13 UTC - in response to Message 1014035.  
Last modified: 10 Jul 2010, 1:13:26 UTC

Good evening.

Is this where I post issues with downloading CUDA job packets? My first attempt at CUDA processing since upgrading from MSI 8600GT to a EVA GTX480. The work unit hangs at 4.41% 0.53/11.90MB and appears to start to create 19 tasks. Tried aborting and retying, but to no available. I want to see if is just a bad packet, but no way to move on. I have a bunch of CPU SETI, but the whole idea of the GTX480 was for BOINC GPU processing.

File is named cufft32_30_14.dll. I'd really like to get it running and see what the baby can do.

From message log - Temporarily failed download of cufft32_30_14.dll: HTTP error.

Any ideas or some other board I should post to?

Thanks and have a good day.

Rich


Good evening and welcome to the boards !

Since I don't have fermi card (yet), I can only recommend that you read Richard Haselgrove's sticky thread just few clicks away.
Well.. here is it anyway Running SETI@home on an nVidia Fermi GPU

[edit]I know it isn't small thread :) but I still strongly suggest that you read it thoroughly, cuz the guy who started it surely knows what he's doing.[/edit]

Who the hell is General Failure and why is he reading my harddisk?¿
ID: 1014042 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13720
Credit: 208,696,464
RAC: 304
Australia
Message 1014045 - Posted: 10 Jul 2010, 1:08:25 UTC - in response to Message 1014041.  

One machine (ES-1) keeps hitting the limit in asking for new work, but has only 12 work units. The only other computer I checked is being denied with 15 work units

The limit is 5 per CPU unit & 40 per GPU unit.
AFAIK the 5 per CPU unit means a Dual Core system can have up to 10, a Quad Core up to 20.
Grant
Darwin NT
ID: 1014045 · Report as offensive
Profile soft^spirit
Avatar

Send message
Joined: 18 May 99
Posts: 6497
Credit: 34,134,168
RAC: 0
United States
Message 1014046 - Posted: 10 Jul 2010, 1:09:51 UTC - in response to Message 1014015.  

...
[edit]Oh.. and the time calculations on my CPU units are really still WAY off.[/edit]

As your hosts are running stock applications, they're very interesting as examples of the kind of estimate adjustments the servers make. Most active forum members here are running anonymous platform, for which that feature is disabled for now.

The "Number of tasks completed" needs to be at least 10 before BOINC will try to adjust estimates, and for a CPU getting to that point will clearly take longer than most GPUs. That period at the beginning for each host when CPU estimates are unadjusted while GPU tasks have the adjustment is certain to be awkward. If the estimates for your AMD CPU tasks with 7 completed are even worse than those for the Intel CPU tasks with 27 completed, perhaps that's a positive sign. OTOH, the servers are maintaining an overall host average as well as the app version averages, the relative speed of CPU versus GPU tasks on a host may also affect how long it takes for estimates to stabilize as much as they can.

Please keep us posted on future developments, BOINC is trying for a 'one size fits all' method. That usually really means the fit is poor but maybe adequate for all.
                                                               Joe


Okay.. what I am seeing is the AMD estimating at around.. 2:45 per gpu task.. Which falls under not bad. these tend to run between 50 minutes and just under 3 hours. The cpu is not yet crunching SETI as it has been busy on CPDN for some time. Their units last ... 300+ hours.

On the I5 system, I would say the GPU units are pretty close to fair.... Hard to tell due to the mix of units at the moment. But 2 hours is about average.. I have everything throttled back a bit, as I need to use the computers too.

However the CPU units rating themselves at 10 hours on the GT220,(9 to 11:50 that I have seen) is very very far off. These units will take a MAX of about 3
hours to complete, each. 2 at a time.

If someone want to contact me directly to be an "example".. send a message and I will keep them posted directly. As long as it gets to those that can use the information.
Janice
ID: 1014046 · Report as offensive
Josef W. Segur
Volunteer developer
Volunteer tester

Send message
Joined: 30 Oct 99
Posts: 4504
Credit: 1,414,761
RAC: 0
United States
Message 1014054 - Posted: 10 Jul 2010, 1:18:07 UTC - in response to Message 1014041.  

...
One machine (ES-1) keeps hitting the limit in asking for new work, but has only 12 work units.
...
Ed

The limit was set at 5 per CPU this time, so your dual core hosts are allowed 10 in progress, single core hosts 5 in progress.

When your hosts finish enough work to get below the limits or the limits are increased, you'll get more work. We have no official word on when the limits may be raised, though I've seen one post suggesting there might have been some change an hour or two ago.
                                                                Joe
ID: 1014054 · Report as offensive
Profile Donald L. Johnson
Avatar

Send message
Joined: 5 Aug 02
Posts: 8240
Credit: 14,654,533
RAC: 20
United States
Message 1014056 - Posted: 10 Jul 2010, 1:21:22 UTC - in response to Message 1013967.  
Last modified: 10 Jul 2010, 1:29:57 UTC

Don't bother-DA told me personally he doesn't pay attention to the forums.


I wouldn't expect him to. But I'm sure the various project admins tell him of anything that he needs to be aware of.


Well, we know Jeff reads these boards, or at least the threads he starts, as we saw after last week's outage. AND he acts on our reasonable suggestions. Too bad he had to hide his "new limits" threads due to too much activity, so I can't quote him directly. (8{)

I suspect he is monitoring the server load and upload/download traffic levels, as he did last weekend. I hope he will be able to adjust the cpu/gpu/host limits upward as server load subsides. As Jeff said himself, this is a new protocol, and they (and we) will be adjusting things as they learn what works and what doesn't. As long as the servers remain stable....

The one thing I do NOT expect is lots of changes in a short time period. I think they realized, after the problems a couple weeks ago, that too many changes all at once will really screw up the works, so now they are going to make changes in small steps, watch what happens, then adjust some more. Which is exactly how I would do it.

So relax, stay cool (forecast here is for >100 F through the weekend), and crunch 'em if you got 'em.

Donald
Infernal Optimist / Submariner, retired
ID: 1014056 · Report as offensive
Profile zoom3+1=4
Volunteer tester
Avatar

Send message
Joined: 30 Nov 03
Posts: 65709
Credit: 55,293,173
RAC: 49
United States
Message 1014062 - Posted: 10 Jul 2010, 1:35:30 UTC - in response to Message 1014054.  

...
One machine (ES-1) keeps hitting the limit in asking for new work, but has only 12 work units.
...
Ed

The limit was set at 5 per CPU this time, so your dual core hosts are allowed 10 in progress, single core hosts 5 in progress.

When your hosts finish enough work to get below the limits or the limits are increased, you'll get more work. We have no official word on when the limits may be raised, though I've seen one post suggesting there might have been some change an hour or two ago.
                                                                Joe

Right now I'm processing 76.8 WU's a day on My QX6700 cpu and It's going to take about 6-7 days before I get below this 20 a day WU cap... Then I should be able to get some WU's for the two GTX295 cards, Which can chew through a WU between 2 and 12 minutes a piece(outside of VLARs which GT200 type chips have problems doing, Unlike the 400 series[Fermi]), So these limits are still way too low to My thinking.
The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's
ID: 1014062 · Report as offensive
Profile perryjay
Volunteer tester
Avatar

Send message
Joined: 20 Aug 02
Posts: 3377
Credit: 20,676,751
RAC: 0
United States
Message 1014063 - Posted: 10 Jul 2010, 1:36:26 UTC - in response to Message 1014035.  

You should be able to get the cufft file here.. http://boinc2.ssl.berkeley.edu/beta/download/cufft32_30_14.dll

Try dropping it in program data/BOINC/projects/setiathome.berkeley.edu Stop BOINC and any clients you have running (SETI, Ect.) restart boinc and it should pick up on it. Check the messages tab and if still trying to download cancel the download.


PROUD MEMBER OF Team Starfire World BOINC
ID: 1014063 · Report as offensive
James Nelson
Volunteer tester
Avatar

Send message
Joined: 23 Mar 02
Posts: 381
Credit: 4,806,382
RAC: 0
United States
Message 1014066 - Posted: 10 Jul 2010, 1:37:41 UTC - in response to Message 1014056.  

my kids are brats my wife doesn't understand me work sucks my boss is an unbearable jerk.


you did say all problems ?
ID: 1014066 · Report as offensive
Lane42

Send message
Joined: 17 May 99
Posts: 59
Credit: 227,150,556
RAC: 11
United States
Message 1014102 - Posted: 10 Jul 2010, 3:09:29 UTC - in response to Message 1014066.  

That was a good one james (:
ID: 1014102 · Report as offensive
Rich-E
Volunteer tester

Send message
Joined: 23 Feb 01
Posts: 41
Credit: 1,685,487
RAC: 0
United States
Message 1014106 - Posted: 10 Jul 2010, 3:19:11 UTC - in response to Message 1014042.  

Thanks for your help. I'll give it a complete read, but it seems dead on. Thanks again to you and perryjay who offered the direct link to the file. Rich
ID: 1014106 · 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 1014107 - Posted: 10 Jul 2010, 3:20:00 UTC

Well....a second rig has run out of Cuda tasks and is not getting any.......
So that is obviously not working right.
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 1014107 · Report as offensive
Profile rebest Project Donor
Volunteer tester
Avatar

Send message
Joined: 16 Apr 00
Posts: 1296
Credit: 45,357,093
RAC: 0
United States
Message 1014111 - Posted: 10 Jul 2010, 3:24:41 UTC - in response to Message 1014107.  

Well....a second rig has run out of Cuda tasks and is not getting any.......
So that is obviously not working right.


Check your count of CPU tasks. I had a bunch of VLARs that threw me over the overall limit.

My GPU is now humming away, however I am also wanted for the mass murder of about 80 VLAR WUs.

Gotta run.




Join the PACK!
ID: 1014111 · 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 1014112 - Posted: 10 Jul 2010, 3:25:32 UTC

Gawd....term limits in office should work so well as these here.
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 1014112 · 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 1014113 - Posted: 10 Jul 2010, 3:26:38 UTC - in response to Message 1014111.  

Well....a second rig has run out of Cuda tasks and is not getting any.......
So that is obviously not working right.


Check your count of CPU tasks. I had a bunch of VLARs that threw me over the overall limit.

My GPU is now humming away, however I am also wanted for the mass murder of about 80 VLAR WUs.

Gotta run.



Yeah, I know......
The freaking point is the GPU was supposed to have a separate limit now.
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 1014113 · Report as offensive
Previous · 1 · 2 · 3 · 4 · 5 · 6 · 7 · 8 . . . 10 · Next

Message boards : Number crunching : Report all problems here...........


 
©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.