Short estimated runtimes - don't panic

Message boards : Number crunching : Short estimated runtimes - don't panic
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · 4 . . . 5 · Next

AuthorMessage
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14679
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1216612 - Posted: 10 Apr 2012, 21:32:19 UTC

People running fast GPUs - especially with optimised or other third-party applications - may have noticed that work allocated since the end of maintenance today has exceptionally short estimated runtimes.

Like the thread title says, Don't Panic!

This was planned and expected. The work itself is exactly the same as before, it's just the estimates which may be screwy for a while. Not everyone will even be affected.

If you are one of the ones seeing short runtime estimates, you'll also see that BOINC tries - but in the current state of comms, probably fails - to fetch more work than normal. This is why the limits on 'tasks in progress' were put in place - tonight's the night when we need them. But even with the limits in place, some people may end up with more work cached than they're really comfortable with. Just set 'no new tasks' if you seem to be getting too much.

As soon as the first short-estimate task reaches the head of the queue and crunches - in the normal time, of course - all the estimates will jump back to normal. You may end up with more tasks than you really need, and perhaps a bit of 'high priority' running, but it should be manageable and burn itself off without any action on your part.

And then, in another two or three weeks, we'll do it all again. Enjoy.

(and no, this is nothing whatsoever to do with the release of BOINC 7)
ID: 1216612 · Report as offensive
Profile red-ray
Avatar

Send message
Joined: 24 Jun 99
Posts: 308
Credit: 9,029,848
RAC: 0
United Kingdom
Message 1216618 - Posted: 10 Apr 2012, 21:49:07 UTC - in response to Message 1216612.  

Thank you for the warning. I assume this is why my DCF is jumping all over the place.

10-Apr-2012 21:20:40 [SETI@home] [dcf] DCF: 0.431077->5.996127, raw_ratio 5.996127, adj_ratio 13.909641

With my mix of GPU speeds I have only just got things stable so now I guess I will need to reduce the <flops> on my GPUs.
ID: 1216618 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14679
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1216621 - Posted: 10 Apr 2012, 22:00:04 UTC - in response to Message 1216618.  

Thank you for the warning. I assume this is why my DCF is jumping all over the place.

10-Apr-2012 21:20:40 [SETI@home] [dcf] DCF: 0.431077->5.996127, raw_ratio 5.996127, adj_ratio 13.909641

With my mix of GPU speeds I have only just got things stable so now I guess I will need to reduce the <flops> on my GPUs.

Actually, if your flops entries were accurately calculated in the first place - difficult-to-impossible with your mixture of cards, I know - you should have seen no change at all.

My post - more of an informational message than a warning - was intended mainly for people who have installed optimised apps, but haven't fiddled by hand with their app_info.xml file. They will see the changes most clearly, but - as I hope my post made clear - they don't need to change anything. Do not adjust your set
ID: 1216621 · Report as offensive
Profile SciManStev Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 20 Jun 99
Posts: 6658
Credit: 121,090,076
RAC: 0
United States
Message 1216624 - Posted: 10 Apr 2012, 22:13:34 UTC

To me this sounds like very good news! I am very pleased to see something happening with this issue!

Steve
Warning, addicted to SETI crunching!
Crunching as a member of GPU Users Group.
GPUUG Website
ID: 1216624 · Report as offensive
Cosmic_Ocean
Avatar

Send message
Joined: 23 Dec 00
Posts: 3027
Credit: 13,516,867
RAC: 13
United States
Message 1216627 - Posted: 10 Apr 2012, 22:20:33 UTC

My AP_v6 estimates have dropped just about in half. Last week, they dropped to just under 5 hours, now they're down to 5:42:42. It should only take just one task to complete (which should be in the next hour or so) to self-correct though. Good thing my 10-day cache was already full though.
Linux laptop:
record uptime: 1511d 20h 19m (ended due to the power brick giving-up)
ID: 1216627 · Report as offensive
AndrewM
Volunteer tester

Send message
Joined: 5 Jan 08
Posts: 369
Credit: 34,275,196
RAC: 0
Australia
Message 1216667 - Posted: 11 Apr 2012, 0:07:39 UTC
Last modified: 11 Apr 2012, 0:08:56 UTC

Amused rather than paniced. I have some MB work units due back on 25th April. They're estimated at 40 seconds.
I mean the 680's are good, but not that good. lol
AndrewM
ID: 1216667 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13855
Credit: 208,696,464
RAC: 304
Australia
Message 1216720 - Posted: 11 Apr 2012, 4:28:41 UTC - in response to Message 1216667.  

Amused rather than paniced. I have some MB work units due back on 25th April. They're estimated at 40 seconds.
I mean the 680's are good, but not that good. lol

Wait for the applications to be optimised for the new cards. Running one at a time, 40 seconds wouldn't surprise me for a shorty.

As for the reduced estimated times; instead of them being way, way, way, way, way out they're just way, way out.
A lot closer than they were, but still a long way from the mark.
Grant
Darwin NT
ID: 1216720 · Report as offensive
Profile soft^spirit
Avatar

Send message
Joined: 18 May 99
Posts: 6497
Credit: 34,134,168
RAC: 0
United States
Message 1216722 - Posted: 11 Apr 2012, 4:34:45 UTC

Could we get the "Don't Panic" in bold friendly letters please?
Janice
ID: 1216722 · Report as offensive
Profile Slavac
Volunteer tester
Avatar

Send message
Joined: 27 Apr 11
Posts: 1932
Credit: 17,952,639
RAC: 0
United States
Message 1216737 - Posted: 11 Apr 2012, 5:01:39 UTC - in response to Message 1216722.  

Running Boinc 6.10.xx and the latest app from Jason.

Interestingly I had a ton of really short runtime units. After running only one of these, estimates for all tasks bounced back to where they should be.


Executive Director GPU Users Group Inc. -
brad@gpuug.org
ID: 1216737 · Report as offensive
Horacio

Send message
Joined: 14 Jan 00
Posts: 536
Credit: 75,967,266
RAC: 0
Argentina
Message 1216738 - Posted: 11 Apr 2012, 5:08:41 UTC - in response to Message 1216621.  

Actually, if your flops entries were accurately calculated in the first place - difficult-to-impossible with your mixture of cards, I know - you should have seen no change at all.

My post - more of an informational message than a warning - was intended mainly for people who have installed optimised apps, but haven't fiddled by hand with their app_info.xml file. They will see the changes most clearly, but - as I hope my post made clear - they don't need to change anything. Do not adjust your set


Just curiosity, if flops are used then there should no be changes in the estimations? Does it means that they are playing with the APR?
(I'm using flops, and I've not seen any change, but neither I've seen any change in my, ussually, very high APR values..)

ID: 1216738 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14679
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1216773 - Posted: 11 Apr 2012, 7:15:29 UTC - in response to Message 1216737.  

Running Boinc 6.10.xx and the latest app from Jason.

Interestingly I had a ton of really short runtime units. After running only one of these, estimates for all tasks bounced back to where they should be.

Er, I think that's what I said would happen. Nice when things go according to plan.
ID: 1216773 · Report as offensive
Profile Slavac
Volunteer tester
Avatar

Send message
Joined: 27 Apr 11
Posts: 1932
Credit: 17,952,639
RAC: 0
United States
Message 1216774 - Posted: 11 Apr 2012, 7:18:47 UTC - in response to Message 1216773.  

Running Boinc 6.10.xx and the latest app from Jason.

Interestingly I had a ton of really short runtime units. After running only one of these, estimates for all tasks bounced back to where they should be.

Er, I think that's what I said would happen. Nice when things go according to plan.


Well, sorta. Now the figures are all over the board. It'll sort itself out in a bit.


Executive Director GPU Users Group Inc. -
brad@gpuug.org
ID: 1216774 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14679
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1216775 - Posted: 11 Apr 2012, 7:20:14 UTC - in response to Message 1216738.  

Actually, if your flops entries were accurately calculated in the first place - difficult-to-impossible with your mixture of cards, I know - you should have seen no change at all.

My post - more of an informational message than a warning - was intended mainly for people who have installed optimised apps, but haven't fiddled by hand with their app_info.xml file. They will see the changes most clearly, but - as I hope my post made clear - they don't need to change anything. Do not adjust your set

Just curiosity, if flops are used then there should no be changes in the estimations? Does it means that they are playing with the APR?
(I'm using flops, and I've not seen any change, but neither I've seen any change in my, ussually, very high APR values..)

No, not 'playing with' - just a step towards using the whole APR properly.

For a while, we allowed APR to grow to unbelievable figures, and then stopped believing it. Now, we're starting to keep the value sane - and so we can start believing it again, as we should have been able to all along.
ID: 1216775 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14679
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1216776 - Posted: 11 Apr 2012, 7:25:22 UTC - in response to Message 1216774.  

Running Boinc 6.10.xx and the latest app from Jason.

Interestingly I had a ton of really short runtime units. After running only one of these, estimates for all tasks bounced back to where they should be.

Er, I think that's what I said would happen. Nice when things go according to plan.

Well, sorta. Now the figures are all over the board. It'll sort itself out in a bit.

Are you saying you ran that task out of sequence? Sounds like you might be one of the ones where the cache gets too large, in spite of the limits on tasks in progress. Set 'no new tasks' for a few hours while you burn off the old work in sequence, then let it settle back down again with new work only.
ID: 1216776 · Report as offensive
W-K 666 Project Donor
Volunteer tester

Send message
Joined: 18 May 99
Posts: 19407
Credit: 40,757,560
RAC: 67
United Kingdom
Message 1216779 - Posted: 11 Apr 2012, 7:35:53 UTC

Well would you believe that, after 8 months I finally get an AP task with estimate that makes sense.

Definitely a red letter day
ID: 1216779 · Report as offensive
Profile shizaru
Volunteer tester
Avatar

Send message
Joined: 14 Jun 04
Posts: 1130
Credit: 1,967,904
RAC: 0
Greece
Message 1216790 - Posted: 11 Apr 2012, 8:25:28 UTC

Probably a dumb question, but since I don't know the answer I'll ask it anyway:-)

My GPU estimates (to completion times) are still at almost 3x. I realize my ION is not a fast GPU, but it has reported around 10 tasks since the outage. Is this normal?

Here's my little cruncher
ID: 1216790 · Report as offensive
Grant (SSSF)
Volunteer tester

Send message
Joined: 19 Aug 99
Posts: 13855
Credit: 208,696,464
RAC: 304
Australia
Message 1216793 - Posted: 11 Apr 2012, 8:38:50 UTC - in response to Message 1216790.  
Last modified: 11 Apr 2012, 8:39:31 UTC

My GPU estimates (to completion times) are still at almost 3x. I realize my ION is not a fast GPU, but it has reported around 10 tasks since the outage. Is this normal?

Yep.
As you report WUs with the shorter runtime things will stabilise, sort of.
3x is a big improvement on 20-30x.

As mentioned in the first post, in a few more weeks there will be another tweak to the estimated completion times.



Which means, maybe, if we're really, really, really lucky, possibly before the end of the year, fingers crossed, we might be able to get a cache of GPU work that's good for more than 24 hours.
Maybe.
Grant
Darwin NT
ID: 1216793 · Report as offensive
W-K 666 Project Donor
Volunteer tester

Send message
Joined: 18 May 99
Posts: 19407
Credit: 40,757,560
RAC: 67
United Kingdom
Message 1216794 - Posted: 11 Apr 2012, 8:39:44 UTC - in response to Message 1216790.  

Probably a dumb question, but since I don't know the answer I'll ask it anyway:-)

My GPU estimates (to completion times) are still at almost 3x. I realize my ION is not a fast GPU, but it has reported around 10 tasks since the outage. Is this normal?

Here's my little cruncher

The new estimates are only for tasks d/loaded since the outage.
ID: 1216794 · Report as offensive
Profile shizaru
Volunteer tester
Avatar

Send message
Joined: 14 Jun 04
Posts: 1130
Credit: 1,967,904
RAC: 0
Greece
Message 1216799 - Posted: 11 Apr 2012, 9:03:00 UTC - in response to Message 1216794.  

The new estimates are only for tasks d/loaded since the outage.


I'll go find something to bang my head on! Thanx:-)
ID: 1216799 · Report as offensive
W-K 666 Project Donor
Volunteer tester

Send message
Joined: 18 May 99
Posts: 19407
Credit: 40,757,560
RAC: 67
United Kingdom
Message 1216803 - Posted: 11 Apr 2012, 9:39:05 UTC - in response to Message 1216799.  

The new estimates are only for tasks d/loaded since the outage.


I'll go find something to bang my head on! Thanx:-)

Don't use my bit of the wall, it's beginning to crumble from over use. ;-)
ID: 1216803 · Report as offensive
1 · 2 · 3 · 4 . . . 5 · Next

Message boards : Number crunching : Short estimated runtimes - don't panic


 
©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.