Darwin (Feb 17 2009)


log in

Advanced search

Message boards : Technical News : Darwin (Feb 17 2009)

Author Message
Profile Matt Lebofsky
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar
Send message
Joined: 1 Mar 99
Posts: 1390
Credit: 74,079
RAC: 0
United States
Message 866577 - Posted: 17 Feb 2009, 23:42:50 UTC
Last modified: 18 Feb 2009, 17:49:31 UTC

Over the long (President's Day) weekend one of our storage servers had a headache. Not a big deal, and we got to the bottom of it today (pretty much just a RAID drive failure). We were able to get a workaround in place so we could start generating/saving workunits again, and will slowly transition back to normal over the next day or two. It has been a bit rocky the last few days because the workaround involves a different RAID with far less I/O throughput.

There's always a bright side during work transmission failures: we get to catch up on backlogged queues. So by the time we had our usual database compression/backup outage today the result table was relatively small, and therefore got packed down nice and tight. That's always helpful.

Spent most of the day with the fallout of the above, while also getting a couple systems configured for new duty - mostly administrative/CPU servers that will replace a some older clunkers.

- Matt
____________
-- BOINC/SETI@home network/web/science/development person
-- "Any idiot can have a good idea. What is hard is to do it." - Jeanne-Claude

Profile Blurf
Volunteer tester
Send message
Joined: 2 Sep 06
Posts: 7612
Credit: 7,026,456
RAC: 1,455
United States
Message 866579 - Posted: 17 Feb 2009, 23:57:38 UTC

Thanks, Matt....looking forward to NTPCKER going live
____________


Profile Dr. C.E.T.I.
Avatar
Send message
Joined: 29 Feb 00
Posts: 15993
Credit: 690,597
RAC: 0
United States
Message 866747 - Posted: 18 Feb 2009, 15:16:32 UTC


. . . Accolades to Berkeley . . . and To Each of You [Thanks for the Post Matt]


____________
BOINC Wiki . . .

Science Status Page . . .

PhonAcq
Send message
Joined: 14 Apr 01
Posts: 1624
Credit: 22,506,288
RAC: 5,154
United States
Message 866756 - Posted: 18 Feb 2009, 16:10:34 UTC - in response to Message 866577.


There's always a bright side during work tranmission failures: we get to catch up on backlogged queues. So by the time we had our usual database compression/backup outage today the result table was relatively small, and therefore got packed down nice and tight. That's always helpful.
- Matt


Would it make any sense to turn off accepting new work a few hours before you shutdown the system on Tuesdays to get this benefit? Maybe you already do that?

Profile KWSN THE Holy Hand Grenade!
Volunteer tester
Avatar
Send message
Joined: 20 Dec 05
Posts: 1987
Credit: 10,972,074
RAC: 15,573
United States
Message 866763 - Posted: 18 Feb 2009, 16:48:19 UTC
Last modified: 18 Feb 2009, 16:54:10 UTC

Query - what percentage of MB units are designated as "CUDA" units, and what percentage are available to plain old cpu apps?

I'm often (even when the servers are working properly...) getting the "no work available for your cpu" flag when requesting new work. Please remember that a lot of us are either working with non-CUDA capable clients, (pre 6.4...) or have ATI graphics boards (or older Nvidia boards...) which are non-CUDA capable. I've occasionally had cpu cores (I have 10 working...) go "cold" on SETI (all computers have more than one project attached, of course, so they aren't actually "cold"!...) due to the above flag.
____________
.

Profile Matt Lebofsky
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar
Send message
Joined: 1 Mar 99
Posts: 1390
Credit: 74,079
RAC: 0
United States
Message 866786 - Posted: 18 Feb 2009, 17:50:44 UTC - in response to Message 866756.

Would it make any sense to turn off accepting new work a few hours before you shutdown the system on Tuesdays to get this benefit? Maybe you already do that?


Sometimes we do this, but usually if the queues are large enough to be deemed a "problem" then it usually take days to drain during "low usage" times like this weekend (sometime days even during "no usage" times - the assimilator can only work so fast).

- Matt
____________
-- BOINC/SETI@home network/web/science/development person
-- "Any idiot can have a good idea. What is hard is to do it." - Jeanne-Claude

Profile Matt Lebofsky
Volunteer moderator
Project administrator
Project developer
Project scientist
Avatar
Send message
Joined: 1 Mar 99
Posts: 1390
Credit: 74,079
RAC: 0
United States
Message 866787 - Posted: 18 Feb 2009, 17:53:04 UTC - in response to Message 866763.

Query - what percentage of MB units are designated as "CUDA" units, and what percentage are available to plain old cpu apps?


I think that all workunits are deemed available for either CUDA or regular CPU, but any given instance of the CUDA client can only claim up to N workunits per day. I'm not sure what N is offhand...

- Matt
____________
-- BOINC/SETI@home network/web/science/development person
-- "Any idiot can have a good idea. What is hard is to do it." - Jeanne-Claude

Profile Rick A. Sponholz
Volunteer tester
Avatar
Send message
Joined: 13 Jun 99
Posts: 43
Credit: 20,486,123
RAC: 6,544
United States
Message 866794 - Posted: 18 Feb 2009, 18:27:03 UTC - in response to Message 866787.

I have NEVER experienced a 6.08 CUDA work unit execute in the CPU, even when I ran out of other SETI work. I would be interested to hear if anyone has had a 6.08 CUDA work unit run in their CPU.
Rick
____________

Profile KWSN THE Holy Hand Grenade!
Volunteer tester
Avatar
Send message
Joined: 20 Dec 05
Posts: 1987
Credit: 10,972,074
RAC: 15,573
United States
Message 866797 - Posted: 18 Feb 2009, 18:48:56 UTC - in response to Message 866794.
Last modified: 18 Feb 2009, 18:53:11 UTC

I have NEVER experienced a 6.08 CUDA work unit execute in the CPU, even when I ran out of other SETI work. I would be interested to hear if anyone has had a 6.08 CUDA work unit run in their CPU.
Rick


The reason I asked is that I have NO CUDA capable GPU's to date, and I'm getting the occasional "no work available for your CPU" flag, (as mentioned in my first post...)
____________
.

Fred W
Volunteer tester
Send message
Joined: 13 Jun 99
Posts: 2524
Credit: 11,954,210
RAC: 0
United Kingdom
Message 866813 - Posted: 18 Feb 2009, 19:38:09 UTC - in response to Message 866763.

Query - what percentage of MB units are designated as "CUDA" units, and what percentage are available to plain old cpu apps?

I'm often (even when the servers are working properly...) getting the "no work available for your cpu" flag when requesting new work. Please remember that a lot of us are either working with non-CUDA capable clients, (pre 6.4...) or have ATI graphics boards (or older Nvidia boards...) which are non-CUDA capable. I've occasionally had cpu cores (I have 10 working...) go "cold" on SETI (all computers have more than one project attached, of course, so they aren't actually "cold"!...) due to the above flag.

As I understand there is no such thing as a "CUDA" work unit from a server perspective. In the server they are all MB WU's (i.e. 6.03's) and it is given the "6.08" designation only when it is delivered to a CUDA client. That's why most CUDA WU's currently validate again CPU-crunched tasks.

F.
____________

Profile Alan Smith
Send message
Joined: 19 Sep 00
Posts: 106
Credit: 26,816
RAC: 0
United States
Message 866818 - Posted: 18 Feb 2009, 19:53:33 UTC - in response to Message 866797.

>That's why most CUDA WU's currently validate again CPU-crunched tasks.<

Just wondering... is this why my CUDA capable wingman's credit of 59.71 got knocked down to my non-CUDA credit of 43.71?

Profile Raistmer
Volunteer developer
Volunteer tester
Avatar
Send message
Joined: 16 Jun 01
Posts: 3587
Credit: 48,712,670
RAC: 27,328
Russia
Message 866830 - Posted: 18 Feb 2009, 20:13:20 UTC - in response to Message 866818.

>That's why most CUDA WU's currently validate again CPU-crunched tasks.<

Just wondering... is this why my CUDA capable wingman's credit of 59.71 got knocked down to my non-CUDA credit of 43.71?

Hehe, maybe it encourages you to move to CUDA too ? ;) ;)

Profile Alan Smith
Send message
Joined: 19 Sep 00
Posts: 106
Credit: 26,816
RAC: 0
United States
Message 866834 - Posted: 18 Feb 2009, 20:18:23 UTC - in response to Message 866830.

>Hehe, maybe it encourages you to move to CUDA too ?<

Sure, I'll move to CUDA today if you buy the hardware. Only been 2 years since I've had a job. That's why I play poker. ;-)

Profile KWSN THE Holy Hand Grenade!
Volunteer tester
Avatar
Send message
Joined: 20 Dec 05
Posts: 1987
Credit: 10,972,074
RAC: 15,573
United States
Message 866877 - Posted: 18 Feb 2009, 22:44:02 UTC - in response to Message 866818.

>That's why most CUDA WU's currently validate again CPU-crunched tasks.<

Just wondering... is this why my CUDA capable wingman's credit of 59.71 got knocked down to my non-CUDA credit of 43.71?


Credit is always granted at the lower level to prevent overclaiming credit fraud (don't laugh - it's happened...)
____________
.

John McLeod VII
Volunteer developer
Volunteer tester
Avatar
Send message
Joined: 15 Jul 99
Posts: 24785
Credit: 524,053
RAC: 86
United States
Message 866923 - Posted: 18 Feb 2009, 23:59:25 UTC - in response to Message 866877.

>That's why most CUDA WU's currently validate again CPU-crunched tasks.<

Just wondering... is this why my CUDA capable wingman's credit of 59.71 got knocked down to my non-CUDA credit of 43.71?


Credit is always granted at the lower level to prevent overclaiming credit fraud (don't laugh - it's happened...)

Unless there are 3 or more tasks in the validation, in which case the high and low are both dropped and the rest are averaged.
____________


BOINC WIKI

Profile RandyC
Avatar
Send message
Joined: 20 Oct 99
Posts: 714
Credit: 1,704,345
RAC: 0
United States
Message 867091 - Posted: 19 Feb 2009, 16:54:45 UTC - in response to Message 866794.

I have NEVER experienced a 6.08 CUDA work unit execute in the CPU, even when I ran out of other SETI work. I would be interested to hear if anyone has had a 6.08 CUDA work unit run in their CPU.
Rick


If you check out thread AK V8 + CUDA MB team work mod you can find out how to do it.

Profile perryjay
Volunteer tester
Avatar
Send message
Joined: 20 Aug 02
Posts: 3377
Credit: 16,208,651
RAC: 8,159
United States
Message 867100 - Posted: 19 Feb 2009, 17:21:23 UTC - in response to Message 867091.
Last modified: 19 Feb 2009, 17:23:15 UTC

You mean like this one? http://setiathome.berkeley.edu/result.php?resultid=1166518822

With the opt-App Randy C pointed out I have one running on my GPU and two running on the CPUs. They all show up as 6.08s
____________


PROUD MEMBER OF Team Starfire World BOINC

Profile Mark Becker
Send message
Joined: 19 Feb 02
Posts: 5
Credit: 66,544
RAC: 0
United States
Message 867228 - Posted: 20 Feb 2009, 0:36:11 UTC

No work units on the UltraSparc-10 since last weekend. It keeps saying "requesting work".. but none arrives.

Any available?

Mark

Message boards : Technical News : Darwin (Feb 17 2009)

Copyright © 2014 University of California