What has your RAC done......since.....


log in

Advanced search

Message boards : Number crunching : What has your RAC done......since.....

1 · 2 · 3 · 4 . . . 14 · Next
Author Message
Profile MikeProject donor
Volunteer tester
Avatar
Send message
Joined: 17 Feb 01
Posts: 24558
Credit: 33,899,174
RAC: 24,146
Germany
Message 1374142 - Posted: 31 May 2013, 9:39:11 UTC

climbing.

____________

Grant (SSSF)
Send message
Joined: 19 Aug 99
Posts: 5868
Credit: 60,629,432
RAC: 47,533
Australia
Message 1374145 - Posted: 31 May 2013, 9:42:55 UTC - in response to Message 1374137.
Last modified: 31 May 2013, 9:44:45 UTC

Mine is crashing. Badly.

Yep.
Like a stone.

I expect it'll get worse before it gets better while the application optimisation does it's thing, combined with only running 1 WU on a GPU instead of 2 or 3.
Add in the random credit generator & i expect it'll take 8+ weeks before RACs become (relatively) stable again.


EDit- and if there is a problem with the v7 opencl_ati apps validating (or just the apps on certain driver versions), that will add to the plummeting as well.
____________
Grant
Darwin NT.

MikeN
Send message
Joined: 24 Jan 11
Posts: 301
Credit: 32,399,235
RAC: 18,621
United Kingdom
Message 1374150 - Posted: 31 May 2013, 9:53:51 UTC - in response to Message 1374145.

Mine is also in free fall due to:

1. Only being able to run 1WU at a time on my GTX460
2. Ongoing boinc v7.0.64 upgrade problems resulting in work running for the full length of time then declaring a computational error (see separate thread).
3. Not having an optimized CPU app.

Oh well, we all seem to have spent a year moaning about delays in the release of seti v7, I guess we should be careful what we wish for!
____________

TBar
Volunteer tester
Send message
Joined: 22 May 99
Posts: 1386
Credit: 51,089,526
RAC: 105,346
United States
Message 1374153 - Posted: 31 May 2013, 9:56:38 UTC
Last modified: 31 May 2013, 10:01:13 UTC

It's not Rocket Science. Look at the run-times/credits. This is using Lunatics Apps in v6 & v7 with the same card in the same Host. Basically, Lunatics_x41g_win32_cuda32.exe verses Lunatics_x41zc_win32_cuda23.exe. I did run 5 tasks with Lunatics_x41zc_win32_cuda32.exe, they are the ones using more CPU time...
SETI@home v7 tasks
SETI@home Enhanced tasks

3022191677 1256431768 30 May 2013, 20:44:57 UTC 31 May 2013, 6:55:15 UTC Completed and validated 2,179.92 58.50 32.93 SETI@home v7 Anonymous platform (NVIDIA GPU) 3018248696 1254694435 28 May 2013, 9:00:12 UTC 28 May 2013, 18:16:43 UTC Completed and validated 248.61 26.17 27.21 SETI@home Enhanced Anonymous platform (NVIDIA GPU) 3022101550 1256400976 30 May 2013, 19:21:29 UTC 31 May 2013, 7:33:56 UTC Completed and validated 2,566.83 103.73 89.00 SETI@home v7 Anonymous platform (NVIDIA GPU) 3019212192 1255133679 28 May 2013, 22:31:24 UTC 30 May 2013, 16:54:11 UTC Completed and validated 1,362.72 53.14 110.57 SETI@home Enhanced Anonymous platform (NVIDIA GPU)

That will kill anyone's RAC. It reminds me of the Highly Blanked APs, much longer runtimes for the same credit.
:-(

juan BFBProject donor
Volunteer tester
Avatar
Send message
Joined: 16 Mar 07
Posts: 5414
Credit: 306,715,231
RAC: 331,377
Brazil
Message 1374158 - Posted: 31 May 2013, 10:12:11 UTC - in response to Message 1374155.

The kitties have been doing a lot of v7........
Just wondered if the RAC had been doing what.

Mine is dropping. Might just be to do with advanced amounts in cache.
\

Not analyzed it all yet.

Do not read more into it than the data I have been able to collect, I was just asking for input.

+ 1 THE rac is dropping, but that could be because our hosts process very fast the V7 WU compared with the wingmen, or due the correlation extra job the RAC produced could be worst, we will know in few days.

On other hand, running 2 WU at a time on the GPUs is very easy but 3 is complicated if the task are Vlars.

____________

Lionel
Send message
Joined: 25 Mar 00
Posts: 576
Credit: 236,172,242
RAC: 231,295
Australia
Message 1374169 - Posted: 31 May 2013, 10:33:44 UTC - in response to Message 1374164.


RAC is going down like the Titanic mate.

If this continues, I'll have to enter the arms race again.

:)


____________

juan BFBProject donor
Volunteer tester
Avatar
Send message
Joined: 16 Mar 07
Posts: 5414
Credit: 306,715,231
RAC: 331,377
Brazil
Message 1374170 - Posted: 31 May 2013, 10:34:03 UTC
Last modified: 31 May 2013, 10:43:30 UTC

Hope the kittymen have a good sleeping night my friend.

Something else calls my atention, it seems like the auto-optimizer task on the server is not working fine, i have few examples, after more than a day of processing V7 only, a 590+2x580 host only receive cuda50 jobs (expected cuda42), a 690+670 host only receive cuda42 (expected cuda50), a 2x690 still receving few cuda32 (expected cuda50 only) that could be another source of why the RAC is dropping, they are crunching with the wrong version of cuda. Hope somebody fix that soon.

BTW running now 2WU at a time against the old 3 produces less RAC and don´t forget, crunching Vlars on the Nvidia GPUs are a 4/5x times slower than normal WU, just don´t know what is happening with the RAC in this case.

I see on a Jason post, now they are more focused to find & fix the problems than speed, so the RAC will take some time to stabilize or rise??? until then we all see our RAC dive, dive, dive...
____________

Grant (SSSF)
Send message
Joined: 19 Aug 99
Posts: 5868
Credit: 60,629,432
RAC: 47,533
Australia
Message 1374176 - Posted: 31 May 2013, 10:45:33 UTC - in response to Message 1374170.

BTW running now 2WU at a time against the old 3 produces less RAC

Probably part of the problem with it's optimisation attempts.
Try one at a time for several days & see if it changes it's mind about what is or isn't best.

____________
Grant
Darwin NT.

Profile Khangollo
Avatar
Send message
Joined: 1 Aug 00
Posts: 245
Credit: 36,410,524
RAC: 0
Slovenia
Message 1374183 - Posted: 31 May 2013, 11:00:30 UTC

MB v7 tasks are being assigned shockingly low credit scores (like 30%-50% of what v6 was giving on average).
____________

Grant (SSSF)
Send message
Joined: 19 Aug 99
Posts: 5868
Credit: 60,629,432
RAC: 47,533
Australia
Message 1374187 - Posted: 31 May 2013, 11:04:29 UTC - in response to Message 1374183.

MB v7 tasks are being assigned shockingly low credit scores (like 30%-50% of what v6 was giving on average).

3 cheers for the random credit generator.
____________
Grant
Darwin NT.

juan BFBProject donor
Volunteer tester
Avatar
Send message
Joined: 16 Mar 07
Posts: 5414
Credit: 306,715,231
RAC: 331,377
Brazil
Message 1374235 - Posted: 31 May 2013, 13:07:35 UTC - in response to Message 1374176.

BTW running now 2WU at a time against the old 3 produces less RAC

Probably part of the problem with it's optimisation attempts.
Try one at a time for several days & see if it changes it's mind about what is or isn't best.

One at a time is slower (due his low GPU usage), at least on my GPUs 580/590/670/690 two at a time is the optimal value it allow you to keep crunching while running other normal windows programs (not games or heavy graphic of course), 3 could works faster but produces a noticeable video lag specialy if the WU are Vlars since they are not crunching only hosts that lag that video lag makes the users some "nervous" so i avoid to use. But take a note, my CPUs are Intel (most I5) so they are relatively slow.
____________

Profile [seti.international] Dirk SadowskiProject donor
Volunteer tester
Avatar
Send message
Joined: 6 Apr 07
Posts: 7102
Credit: 60,944,436
RAC: 17,279
Germany
Message 1374245 - Posted: 31 May 2013, 13:23:34 UTC

From what I see...

Currently with SAHv7 the Cr./day will be less than with SAH Enhanced.
I used x41zc already with SAH Enhanced and switched directly to SAHv7.

Maybe in a few days CreditNew will balance it out and the formerly Cr. are reached.

Or the admins need adjust something.


* Best regards! :-) * Philip J. Fry, team seti.international founder. * Optimize your PC for higher RAC. * SETI@home needs your help. *
____________
BR

SETI@home Needs your Help ... $10 & U get a Star!

Team seti.international

Das Deutsche Cafe. The German Cafe.

Profile Wiggo
Avatar
Send message
Joined: 24 Jan 00
Posts: 7358
Credit: 96,946,141
RAC: 66,544
Australia
Message 1374257 - Posted: 31 May 2013, 13:41:45 UTC - in response to Message 1374245.

Mine has gone down but then I've been hitting up my backup projects a fair bit.

Cheers.

rob smithProject donor
Volunteer tester
Send message
Joined: 7 Mar 03
Posts: 8535
Credit: 59,515,329
RAC: 87,397
United Kingdom
Message 1374260 - Posted: 31 May 2013, 13:44:20 UTC

Not unexpectedly mine has gone down, but I'm only running 1 task per GPU until I can get the round tuit out of hock and sort out app_info.xml on all three of my crunchers.
____________
Bob Smith
Member of Seti PIPPS (Pluto is a Planet Protest Society)
Somewhere in the (un)known Universe?

Profile SonicAgamemnon
Avatar
Send message
Joined: 8 Apr 06
Posts: 31
Credit: 11,893,945
RAC: 0
United States
Message 1374261 - Posted: 31 May 2013, 13:46:59 UTC
Last modified: 31 May 2013, 13:47:58 UTC

Down 4%, but it looks like all recently-upgraded v7 boats have lowered about the same amount-- rising/falling on the same tide . . .
____________
"History is a pack of lies about events that never happened told by people who weren't there." - Santayana

TBar
Volunteer tester
Send message
Joined: 22 May 99
Posts: 1386
Credit: 51,089,526
RAC: 105,346
United States
Message 1374479 - Posted: 31 May 2013, 18:57:01 UTC - in response to Message 1374153.

It's not Rocket Science. Look at the run-times/credits. This is using Lunatics Apps in v6 & v7 with the same card in the same Host. Basically, Lunatics_x41g_win32_cuda32.exe verses Lunatics_x41zc_win32_cuda23.exe. I did run 5 tasks with Lunatics_x41zc_win32_cuda32.exe, they are the ones using more CPU time...
SETI@home v7 tasks
SETI@home Enhanced tasks

3022191677 1256431768 30 May 2013, 20:44:57 UTC 31 May 2013, 6:55:15 UTC Completed and validated 2,179.92 58.50 32.93 SETI@home v7 Anonymous platform (NVIDIA GPU) 3018248696 1254694435 28 May 2013, 9:00:12 UTC 28 May 2013, 18:16:43 UTC Completed and validated 248.61 26.17 27.21 SETI@home Enhanced Anonymous platform (NVIDIA GPU) 3022101550 1256400976 30 May 2013, 19:21:29 UTC 31 May 2013, 7:33:56 UTC Completed and validated 2,566.83 103.73 89.00 SETI@home v7 Anonymous platform (NVIDIA GPU) 3019212192 1255133679 28 May 2013, 22:31:24 UTC 30 May 2013, 16:54:11 UTC Completed and validated 1,362.72 53.14 110.57 SETI@home Enhanced Anonymous platform (NVIDIA GPU)

That will kill anyone's RAC. It reminds me of the Highly Blanked APs, much longer runtimes for the same credit.
:-(

Let's watch my other Host as it begins the change over to v7 CUDA tasks. Computer 6979629 was running Ubuntu until I recently changed it over to Win XP running a single task with Lunatics_x41g_win32_cuda32.exe. The RAC was still climbing and it was still receiving v6 CUDA tasks until this morning. Earlier this morning I simply added the v7 Lunatics_x41zc_win32_cuda23.exe App to the app_info file and it immediately stopped receiving v6 CUDA tasks. It's just about out of v6 CUDAs now and just started v7 tasks. The only change was adding the different CUDA app, the CPU tasks are the same. The current numbers are;
Total credit ---- 415,336
Average credit -- 11,724.94

1 · 2 · 3 · 4 . . . 14 · Next

Message boards : Number crunching : What has your RAC done......since.....

Copyright © 2014 University of California