RAC sinking here at Seti, Not at Einstein

Message boards : Number crunching : RAC sinking here at Seti, Not at Einstein
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile zoom3+1=4
Volunteer tester
Avatar

Send message
Joined: 30 Nov 03
Posts: 65709
Credit: 55,293,173
RAC: 49
United States
Message 1170003 - Posted: 11 Nov 2011, 7:22:34 UTC
Last modified: 11 Nov 2011, 7:24:25 UTC

Here @ S@H I run x38g, @ Einstein I run what the server sends Me, Here My RAC is sinking link an Asteroid Dino killer heading towards impact, On Einstein the RAC there is rising.

Einstein 6 gpu friendly, maybe 12 too(8 hours or 24/7).
Lunatics 4 gpu friendly, 6-12(24/7 sure)~

I wish some expert could fix this, But I may not be able to post soon enough, I'm at just over 9000 now, a day back around 9149, tomorrow under 9000...

I know the fault lies in here somewhere and It's not a driver problem or the OS I use. Jason, Joe...
The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's
ID: 1170003 · 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 1170005 - Posted: 11 Nov 2011, 7:27:42 UTC - in response to Message 1170003.  

I thought this was pretty much traced to the poor GPU utilization you were getting on Seti crunching. Unless you were able to fix that......
"Freedom is just Chaos, with better lighting." Alan Dean Foster

ID: 1170005 · 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 1170006 - Posted: 11 Nov 2011, 7:34:03 UTC - in response to Message 1170005.  
Last modified: 11 Nov 2011, 7:37:47 UTC

I thought this was pretty much traced to the poor GPU utilization you were getting on Seti crunching. Unless you were able to fix that......

I have 16GB of ram, I may as well leave then.

Same hardware, same ram, same OS and drivers, Einstein has no problem with what I have, Yet here I do, cause I have 3-GTX295 cards in one PC, Oh and I have plenty of power, 1500w...
The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's
ID: 1170006 · 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 1170007 - Posted: 11 Nov 2011, 7:40:08 UTC
Last modified: 11 Nov 2011, 7:44:01 UTC

I'm sure not going to split up 8-GTX295 cards into 4 PCs just for Seti...

I sure don't have a problem getting wu's at either project right now.
The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's
ID: 1170007 · 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 1170008 - Posted: 11 Nov 2011, 7:50:05 UTC

As to gpu loading, crock, as in load of, I have the same loads @ Einstein as at Seti, besides I have GT200 series gpus, not 400/500 series unobtainium...

So how do I load a GT200???

Answer You can't, or 1 wu will fail for every wu that doesn't fail.
The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's
ID: 1170008 · 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 1170011 - Posted: 11 Nov 2011, 8:00:58 UTC
Last modified: 11 Nov 2011, 8:05:47 UTC

Here's two pics to illustrate My point, something at Seti or in x38g is at fault, I'm not a programmer and I won't pretend to be one.


9149-S@H...

9022-S@H...

9004-S@H...
The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's
ID: 1170011 · Report as offensive
Profile Uli
Volunteer tester
Avatar

Send message
Joined: 6 Feb 00
Posts: 10923
Credit: 5,996,015
RAC: 1
Germany
Message 1170015 - Posted: 11 Nov 2011, 8:18:34 UTC

Vic you will be able to post for a long time to come. Your RAC will not drop below one tomorrow. As to the rest, paitence is a requirement here.
Pluto will always be a planet to me.

Seti Ambassador
Not to late to order an Anni Shirt
ID: 1170015 · Report as offensive
LadyL
Volunteer tester
Avatar

Send message
Joined: 14 Sep 11
Posts: 1679
Credit: 5,230,097
RAC: 0
Message 1170027 - Posted: 11 Nov 2011, 10:10:41 UTC

Projects 'pay' differently.
If you recently added projects or changed resource shares, it will take a while for RAC to settle.
Even on longterm stable systems RAC isn't stable.
To compound the problem, whenever SETI doesn't have tasks, Einstein will readily comply. The debt system is supposed to make sure projects get their share, but in my experience it doesn't.
The new workfetch/scheduling being BOINC alpha tested atm is supposed to address a few of the long standing gripes.

Not to forget, Einstein hasn't adopted 'CreditNew'
ID: 1170027 · 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 1170098 - Posted: 11 Nov 2011, 14:31:21 UTC - in response to Message 1170027.  
Last modified: 11 Nov 2011, 14:32:39 UTC

I'm not so concerned about pay, It is nice to watch the rac go up though, as this is something to keep Me from going insane. I'd think with what I have the rac would rise. It's not even the most powerful type really, although It does have one of the most powerful pc psus made, but xeon capable It's not. But like I've said before this is mainly a prototype for the real deal and yeah It's a hobby, some people like cars a lot, Me I just like PCs more.

Oh and Einstein is set at "0", as It's a backup project.
The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's
ID: 1170098 · Report as offensive
Profile speedbump

Send message
Joined: 19 May 01
Posts: 247
Credit: 192,906,380
RAC: 0
United States
Message 1170130 - Posted: 11 Nov 2011, 16:31:00 UTC

The biggest reason RAC is falling is the problems they have had recently with servers and the limit they reached on results, and the workarounds for all of this, which means that they just have not been able to keep all the computers fed with work units.They are in the process of figuring out how to make the software changes, and we are running a funding drive now to try to buy some equipment that they are in need of to maybe help with the hardware side of the problems. With the funding they have, I am sure it feels like they keep patching leaking holes in a very old boat, not having to funds to do a proper fix, with the proper man power.I always hate it when RAC drops, it takes so long to get it back up again.
ID: 1170130 · 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 1170218 - Posted: 11 Nov 2011, 20:51:51 UTC - in response to Message 1170215.  

Vic you might try upping your cache so that you'll have more seti work to do
before it switches to Einstein.
You could also try reverting to an earlier version of BOINC.
You appear to be running BOINC version 6.12.41 which is known throughout
the universe for setting the the download retry period unreasonably high after
each successive download failure.
Most of the Top hosts are running BOINC version 6.10.58 or 6.10.60 which
will try to download more often and thus run out of WU's less often.

I didn't know about that problem with 6.12.41, I'll go hunt up 6.10.60, can't hurt. Thanks.
The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's
ID: 1170218 · 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 1170238 - Posted: 11 Nov 2011, 21:38:49 UTC - in response to Message 1170218.  

Vic you might try upping your cache so that you'll have more seti work to do
before it switches to Einstein.
You could also try reverting to an earlier version of BOINC.
You appear to be running BOINC version 6.12.41 which is known throughout
the universe for setting the the download retry period unreasonably high after
each successive download failure.
Most of the Top hosts are running BOINC version 6.10.58 or 6.10.60 which
will try to download more often and thus run out of WU's less often.

I didn't know about that problem with 6.12.41, I'll go hunt up 6.10.60, can't hurt. Thanks.

Looks like I spoke too soon, I downgraded to 6.10.60 and I had one Einstein wu crash(over 4.5 hrs gone), so I went to to 6.10.58 and then another wu crashed(also another 2.25 gone), so I get out a new video driver that I'd previously extracted to the hdd and I did a clean install of 285.79 x64 and so far so good. No down clocking was observed at all, but everything is back to Nermal now. I also got rid of the commented out sections of My app info file, just in case.
The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's
ID: 1170238 · 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 1170261 - Posted: 11 Nov 2011, 23:17:20 UTC - in response to Message 1170249.  

" It's always something. "

...Rosanne Roseannadanna

Yeah, but no worries.
The T1 Trust, PRR T1 Class 4-4-4-4 #5550, 1 of America's First HST's
ID: 1170261 · Report as offensive
Terror Australis
Volunteer tester

Send message
Joined: 14 Feb 04
Posts: 1817
Credit: 262,693,308
RAC: 44
Australia
Message 1170448 - Posted: 12 Nov 2011, 8:30:18 UTC

RAC is dropping here too. One of the big guns has dropped by 10k over the last week.

IRC, doesn't RAC normally drop during a shorty storm as the "credit per hour" rate for shorties is less than longer AR's (and this one has been a long one)?

T.A.
ID: 1170448 · Report as offensive
Kevin Olley

Send message
Joined: 3 Aug 99
Posts: 906
Credit: 261,085,289
RAC: 572
United Kingdom
Message 1170454 - Posted: 12 Nov 2011, 9:05:49 UTC - in response to Message 1170448.  

RAC is dropping here too. One of the big guns has dropped by 10k over the last week.

IRC, doesn't RAC normally drop during a shorty storm as the "credit per hour" rate for shorties is less than longer AR's (and this one has been a long one)?

T.A.


RAC drops but pendings rise, I've tried doing the maths a few times and on this machine I think shorties earn a little more but RAC is wingman dependant so you see a drop followed by a gradual rise as pendings stabilise at higher level.



Kevin


ID: 1170454 · 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: 6651
Credit: 121,090,076
RAC: 0
United States
Message 1170517 - Posted: 12 Nov 2011, 15:56:09 UTC

When I had enough work, my pendings were rising at an astonishing rate, and my RAC stayed about the same. Now I can't keep my rig fed with work. I have been seeing either one extreme or the other for quite a while now. Under these conditions, there is no way my rig can get anywhere near it's potential. Many of us are in the same boat. My laptop, crunching on a different account, has done just fine through all of this. It has no GPU, and turns over a work unit in about 11 hours. If it gets a few work units, it is good for a week.

Steve
Warning, addicted to SETI crunching!
Crunching as a member of GPU Users Group.
GPUUG Website
ID: 1170517 · Report as offensive

Message boards : Number crunching : RAC sinking here at Seti, Not at Einstein


 
©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.