Advanced search

Message boards : News : New Nvidia application 60% faster is now out

Author Message
Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15140 - Posted: 11 Feb 2010 | 9:22:17 UTC

Along side the old application, we have uploaded the new faster application for Windows (6.01) and Linux (6.02). GDF.

Profile CNT - IQE
Send message
Joined: 21 Sep 09
Posts: 3
Credit: 1,951,950,972
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15197 - Posted: 12 Feb 2010 | 21:08:35 UTC - in response to Message 15140.

Hi, we have problem with new app.

Host: http://www.gpugrid.net/hosts_user.php?userid=38072

OS: Ubuntu 9.10 64bit
BM 6.10.17
driver: 195.22

new WUs runs for 38hrs!!! and progress is 87%

Is it normal? I think, no.

Thanks for reply.

Progress of all Wus is here: http://stats.czechnationalteam.cz/index.php?akce=teampc2

Profile tocx
Send message
Joined: 28 Nov 09
Posts: 2
Credit: 5,646,840
RAC: 0
Level
Ser
Scientific publications
watwatwat
Message 15209 - Posted: 13 Feb 2010 | 12:11:10 UTC

I have the same problem with the new app. After a half day the progress is only 14% => I aborted the
Task

OS: Debian GNU Linux (Squeeze)
BM: 6.10.17
driver: 190.53

STARBASEn
Avatar
Send message
Joined: 17 Feb 09
Posts: 91
Credit: 1,603,303,394
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 15213 - Posted: 13 Feb 2010 | 18:53:27 UTC

New application v6.02 runs about 60% slower

WU: 195-GIANNI_BIND_166_119-73-100-RND1057 Task ID: 1842880

This WU took about 25.06 hours to successfully complete. Normally GIANNI_BIND type WU's consistantly take 14.6 hours to finish.

Platform: Fedora 11, Client 6.10.17, GTS-250, nvdriver 190.53, AMD Phenom II 945 x4, 4GB ram and lots of drive space.

Observations: GPU normally runs 60-63 degrees C but with this app, temps are 39 C. The CPU cores concurrently run WCG WU's. I freed one CPU core and that appeared to speed the above GPU WU progress rate but the one in progress now (126-GIANNI_BIND_166_119-78-100-RND6037_1 Task ID: 1157945) is progressing very slowly (4.75 hors and 4.7% complete). In fact, it appears the driver has been switched from 3D to 2D mode.

I have unchecked the ACEMD ver 2.0 in my preferences until the issue is resolved.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15214 - Posted: 13 Feb 2010 | 22:34:09 UTC - in response to Message 15213.

I have removed the Linux application.

gdf

GEPO
Send message
Joined: 17 Dec 08
Posts: 1
Credit: 17,565,835
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwat
Message 15243 - Posted: 15 Feb 2010 | 20:19:39 UTC

Same problem on Windows 64. Just finished a workunit (1159133) and it took 172,905.55s (about 48h).
A previous one (1158551) took 24,889.70 s and old units running v6.71 usually take between 65000 and 75000 s (for similar credits, less foe lower credit units)

OS: Vista Ultimate 64Bit SP2
Driver 196.34

Ed1934158
Send message
Joined: 15 Mar 09
Posts: 32
Credit: 3,313,639
RAC: 0
Level
Ala
Scientific publications
watwatwatwatwatwatwat
Message 15255 - Posted: 16 Feb 2010 | 14:07:57 UTC - in response to Message 15243.

I also had to abort the unit on Linux, because I could not meet the deadline. 1%/hour on my 9800 GTX+.

Profile Misfit
Avatar
Send message
Joined: 23 May 08
Posts: 33
Credit: 610,551,356
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15267 - Posted: 17 Feb 2010 | 2:46:43 UTC

Same extended time problem with Windows 32-bit Vista.
____________
me@rescam.org

Skip Da Shu
Send message
Joined: 13 Jul 09
Posts: 63
Credit: 2,026,770,165
RAC: 10,702,429
Level
Phe
Scientific publications
watwatwatwatwatwatwat
Message 15268 - Posted: 17 Feb 2010 | 5:13:57 UTC
Last modified: 17 Feb 2010 | 5:44:38 UTC

Well I've got one Achmed (the dead terrorist) WU that I'm gonna let finish as it's 103 hours into it with 10 hours left. No premature detonation here.

Don't suppose u wanna slip me a little credit under the table for the 100+ hours in aborted Achmeds ;-)

On the 103 hour (so far) WU, it had dropped out of clocking mode 2 (performance) down to one of the lower 2D modes... but this might just be coincidental.
____________
- da shu @ HeliOS,
"A child's exposure to technology should never be predicated on an ability to afford it."

Bedrich Hajek
Send message
Joined: 28 Mar 09
Posts: 485
Credit: 10,456,923,466
RAC: 15,173,439
Level
Trp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15280 - Posted: 18 Feb 2010 | 0:17:14 UTC - in response to Message 15140.

I have no problems on my computer, with this application, so far. I am running Windows XP 32 bit. But it is not 60% faster, my results are only about 30% to 45% faster.

Profile CNT - IQE
Send message
Joined: 21 Sep 09
Posts: 3
Credit: 1,951,950,972
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15283 - Posted: 18 Feb 2010 | 6:08:29 UTC - in response to Message 15214.

6.04 Linux looks very good

http://www.gpugrid.net/result.php?resultid=1864286

Profile [AF>Libristes] Dudumomo
Send message
Joined: 30 Jan 09
Posts: 45
Credit: 425,620,748
RAC: 0
Level
Gln
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15284 - Posted: 18 Feb 2010 | 8:12:27 UTC - in response to Message 15280.

I have no problems on my computer, with this application, so far. I am running Windows XP 32 bit. But it is not 60% faster, my results are only about 30% to 45% faster.


Yeah some for me.
Anyway it is a great improvement !

BUT, in my case (Gentoo 64b), the 6.04 perform less well as the 6.02

6.02: GPU Time = 15,672.34; CPU Time = 1,050.29
6.04: GPU Time = 16,304.57; CPU Time = 16,284.93

GPU Time is okay, just a bit longer, but CPU Time...damn...

If there is a problem of CPU priority, may be a cc_config can resolve the problem (As I did with Collatz)

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15285 - Posted: 18 Feb 2010 | 9:15:50 UTC - in response to Message 15284.

Due to the fact that the same synchronization mode of Windows does not work for all Linux flavors, we needed to use the other option. Now, for Linux only,
we have used the other synchronization mode which should use 100% if there is nothing else running but release to other applications if they are computing.

gdf

Skip Da Shu
Send message
Joined: 13 Jul 09
Posts: 63
Credit: 2,026,770,165
RAC: 10,702,429
Level
Phe
Scientific publications
watwatwatwatwatwatwat
Message 15320 - Posted: 19 Feb 2010 | 0:47:50 UTC

I've got an Achmed 6.04 running on primary Desktop machine now. It's running on a GTX-275. Just under 3 hours into it and it's at almost 25% so looking like a 12 hour WU. Much better. Thank you for the fast turn around on this.

Ubuntu v9.10, 64b, NVIDIA 190.53 driver (dual card machine with a GTS-250 in the other slot running a Full Atom-The-Ant WU).
____________
- da shu @ HeliOS,
"A child's exposure to technology should never be predicated on an ability to afford it."

ftpd
Send message
Joined: 6 Jun 08
Posts: 152
Credit: 328,250,382
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15326 - Posted: 19 Feb 2010 | 9:12:17 UTC

18-2-2010 15:25:44 GPUGRID Starting I88-OTTO_HERGflip656-2-100-RND4338_5
18-2-2010 15:25:45 GPUGRID Starting task I88-OTTO_HERGflip656-2-100-RND4338_5 using acemd2 version 603
18-2-2010 15:26:00 GPUGRID Computation for task I88-OTTO_HERGflip656-2-100-RND4338_5 finished
18-2-2010 15:26:00 GPUGRID Output file I88-OTTO_HERGflip656-2-100-RND4338_5_1 for task I88-OTTO_HERGflip656-2-100-RND4338_5 absent
18-2-2010 15:26:00 GPUGRID Output file I88-OTTO_HERGflip656-2-100-RND4338_5_2 for task I88-OTTO_HERGflip656-2-100-RND4338_5 absent
18-2-2010 15:26:00 GPUGRID Output file I88-OTTO_HERGflip656-2-100-RND4338_5_3 for task I88-OTTO_HERGflip656-2-100-RND4338_5 absent
18-2-2010 15:26:00 GPUGRID Starting 77-GIANNI_BIND_11-66-100-RND9856_1
18-2-2010 15:26:01 GPUGRID Starting task 77-GIANNI_BIND_11-66-100-RND9856_1 using acemd2 version 603
18-2-2010 15:26:01 Milkyway@home Computation for task de_s222_opt_2_v04_979000_1266483063_0 finished
18-2-2010 15:26:01 GPUGRID Starting 477-IBUCH_esrever_pYEEI_0301-27-40-RND8008_1
18-2-2010 15:26:02 GPUGRID Starting task 477-IBUCH_esrever_pYEEI_0301-27-40-RND8008_1 using acemd version 671
18-2-2010 15:35:27 GPUGRID Computation for task 477-IBUCH_esrever_pYEEI_0301-27-40-RND8008_1 finished
18-2-2010 15:35:27 GPUGRID Output file 477-IBUCH_esrever_pYEEI_0301-27-40-RND8008_1_1 for task 477-IBUCH_esrever_pYEEI_0301-27-40-RND8008_1 absent
18-2-2010 15:35:27 GPUGRID Output file 477-IBUCH_esrever_pYEEI_0301-27-40-RND8008_1_2 for task 477-IBUCH_esrever_pYEEI_0301-27-40-RND8008_1 absent
18-2-2010 15:35:27 GPUGRID Output file 477-IBUCH_esrever_pYEEI_0301-27-40-RND8008_1_3 for task 477-IBUCH_esrever_pYEEI_0301-27-40-RND8008_1 absent
18-2-2010 15:35:27 GPUGRID Starting 247-GIANNI_BIND_166_119-74-100-RND9378_1
18-2-2010 15:35:28 GPUGRID Starting task 247-GIANNI_BIND_166_119-74-100-RND9378_1 using acemd2 version 603
18-2-2010 15:35:28 Cosmology@Home Resuming task wu_021610_073538_0_1_0 using camb version 216
18-2-2010 15:38:12 World Community Grid Computation for task nd628_00090_0 finished
18-2-2010 15:38:12 Cosmology@Home Resuming task wu_021610_073537_2_1_0 using camb version 216
18-2-2010 15:38:14 Cosmology@Home Resuming task wu_021610_073537_1_0_0 using camb version 216
18-2-2010 15:39:01 climateprediction.net Resuming task hadsm3fub_jzrx_006456887_0 using hadsm3 version 607
18-2-2010 15:48:57 Cosmology@Home Resuming task wu_021610_073536_1_1_0 using camb version 216
18-2-2010 15:50:30 climateprediction.net Resuming task hadsm3fub_jzrw_006456886_9 using hadsm3 version 607
18-2-2010 15:55:44 Cosmology@Home Starting wu_021610_073534_2_1_0
18-2-2010 15:55:44 Cosmology@Home Starting task wu_021610_073534_2_1_0 using camb version 216
18-2-2010 15:58:04 GPUGRID Computation for task 247-GIANNI_BIND_166_119-74-100-RND9378_1 finished
18-2-2010 15:58:04 GPUGRID Output file 247-GIANNI_BIND_166_119-74-100-RND9378_1_1 for task 247-GIANNI_BIND_166_119-74-100-RND9378_1 absent
18-2-2010 15:58:04 GPUGRID Output file 247-GIANNI_BIND_166_119-74-100-RND9378_1_2 for task 247-GIANNI_BIND_166_119-74-100-RND9378_1 absent
18-2-2010 15:58:04 GPUGRID Output file 247-GIANNI_BIND_166_119-74-100-RND9378_1_3 for task 247-GIANNI_BIND_166_119-74-100-RND9378_1 absent

Only errrors with new application!
____________
Ton (ftpd) Netherlands

Profile Kokomiko
Avatar
Send message
Joined: 18 Jul 08
Posts: 190
Credit: 24,093,690
RAC: 0
Level
Pro
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 15327 - Posted: 19 Feb 2010 | 10:41:02 UTC

Maybe this application is fast, but not stable. Have stopped now crunching because my GTX295 and also one of my GTX260 are crashing over and over in the last days and I prefer to use my server without have it to reboot often a day. If it was only one card that crashed, I would think, it's the card. But 2 PCs since using the new application looks more like the application is the cause.
____________

Toni
Volunteer moderator
Project administrator
Project developer
Project tester
Project scientist
Send message
Joined: 9 Dec 08
Posts: 1006
Credit: 5,068,599
RAC: 0
Level
Ser
Scientific publications
watwatwatwat
Message 15328 - Posted: 19 Feb 2010 | 12:01:48 UTC - in response to Message 15327.
Last modified: 19 Feb 2010 | 12:06:04 UTC

You can now choose to only run the old (or the new) application (Your account->gpugrid preferences), if you want to check.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1957
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 15329 - Posted: 19 Feb 2010 | 13:48:49 UTC - in response to Message 15327.

Maybe this application is fast, but not stable. Have stopped now crunching because my GTX295 and also one of my GTX260 are crashing over and over in the last days and I prefer to use my server without have it to reboot often a day. If it was only one card that crashed, I would think, it's the card. But 2 PCs since using the new application looks more like the application is the cause.



The problem is not the application but your machine. It crashes on old and new applications. The new application is as reliable as the previous one.

gdf

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15330 - Posted: 19 Feb 2010 | 14:40:38 UTC - in response to Message 15327.

Kokomiko,
3 of your systems are in good shape, but your Phenom 9850 with the GTX 260 and running Vista Ultimate SP2 is a problem system.
If you want to try to use it, start by installing the newer drivers and the newest version of Boinc.
You are using Boinc 6.6.28 on that system. You could be using the Boinc 64bit application version on that system (either 6.10.18 or the beta 6.10.32) and newer drivers:
http://boinc.berkeley.edu/download_all.php
I expect you have not updated your drivers since you had the GTX280 installed!
This might be the source of your problems. Make sure you restart twice after installing the new drivers.

wiyosaya
Send message
Joined: 22 Nov 09
Posts: 114
Credit: 589,114,683
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15334 - Posted: 19 Feb 2010 | 19:02:02 UTC

IMHO, this runs fantastic on XP 32. I see an increase of approximately 50%, and with my 8800 GTX, I can actually hope to return WU's without leaving my computer on for 6 or more hours a day.

http://www.gpugrid.net/results.php?userid=47958

The most recent work unit I received is not this application. Is this application still being sent out?

Profile robertmiles
Send message
Joined: 16 Apr 09
Posts: 503
Credit: 747,645,933
RAC: 28,428
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15444 - Posted: 25 Feb 2010 | 21:39:31 UTC

Does this web site refer to one of the applications this thread mentions, or only one with a similar name?

http://www.acellera.com/?arg=acemd


By the way, here's a list of Nvidia applications that include some similar to GPUGRID:

http://www.nvidia.com/object/tesla_computing_solutions.html


Also, should this Wikipedia article be revised to mention GPUGRID?

http://en.wikipedia.org/wiki/OpenCL

Profile skgiven
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 23 Apr 09
Posts: 3968
Credit: 1,995,359,260
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 15466 - Posted: 27 Feb 2010 | 1:18:09 UTC - in response to Message 15444.

Acellera say they are a London-based technology company.

Profile Hydropower
Avatar
Send message
Joined: 3 Apr 09
Posts: 70
Credit: 6,003,024
RAC: 0
Level
Ser
Scientific publications
watwatwatwatwatwat
Message 15850 - Posted: 20 Mar 2010 | 9:40:47 UTC - in response to Message 15466.

If you scan the website you will find this:
GPUGRID.net
August 10th, 2008
GPUGRID.net project launch, powered by Acellera’s ACEMD molecular dynamics code.

If you do some further searches, you will find more information.

Post to thread

Message boards : News : New Nvidia application 60% faster is now out

//