Advanced search

Message boards : Number crunching : CPU only app v1.04

Author Message
Bedrich Hajek
Send message
Joined: 28 Mar 09
Posts: 411
Credit: 6,063,920,459
RAC: 5,390
Level
Tyr
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 35489 - Posted: 3 Mar 2014 | 23:20:59 UTC

I successfully completed one of these units.

http://www.gpugrid.net/workunit.php?wuid=5228104

3 Mar 2014 | 22:14:10 UTC 3 Mar 2014 | 23:07:04 UTC Completed and validated 1,477.65 1,359.77 9.02 CPU only app v1.04

Two others are still running on my other machine.

Why are we bothering with CPU work units? They are so much slower than GPU WU.



Profile MJH
Project administrator
Project developer
Project scientist
Send message
Joined: 12 Nov 07
Posts: 696
Credit: 27,266,655
RAC: 0
Level
Val
Scientific publications
watwat
Message 35491 - Posted: 4 Mar 2014 | 0:01:27 UTC - in response to Message 35489.

Why are we bothering with CPU work units? They are so much slower than GPU WU.


These are pathfinder WUs for a new project that we are currently scoping. Please keep crunching them, it's giving us useful information. More details on the project soon!

Matt

Alez
Send message
Joined: 17 Nov 12
Posts: 10
Credit: 185,958,753
RAC: 0
Level
Ile
Scientific publications
watwatwatwatwatwatwatwatwatwatwat
Message 35493 - Posted: 4 Mar 2014 | 3:06:58 UTC

v1.07 finally completed on my android without failing. Woohoo.

John C MacAlister
Send message
Joined: 17 Feb 13
Posts: 181
Credit: 144,871,276
RAC: 0
Level
Cys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwat
Message 35504 - Posted: 4 Mar 2014 | 22:25:07 UTC - in response to Message 35493.

Folding@home at Stanford just rolled out a really nifty Chrome app which just chugs along using CPU power. So easy to use!



v1.07 finally completed on my android without failing. Woohoo.

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 35505 - Posted: 4 Mar 2014 | 22:43:15 UTC - in response to Message 35504.
Last modified: 4 Mar 2014 | 23:20:18 UTC

Trailblazing again .........

A few basic observations (Linux):
Task progress remains at 0% and then jumps to 100%.
After tasks reach 100% they proceed to about 10000000000%
After a task reaches 100% if it's suspended it resumes from 0%.
____________
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

captainjack
Send message
Joined: 9 May 13
Posts: 171
Credit: 1,281,813,256
RAC: 123,814
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 35506 - Posted: 4 Mar 2014 | 23:47:21 UTC

Just started a CPU only app 1.04 on a Windows 7 box with BOINC 7.2.37. After the job ran for about a minute, the progress column now says "1.#IO%" and the Remaining column says " ---" (instead of showing hh:mm:ss).

Also running a CPU only app 1.01 on a Linux Ubuntu 13.10 box. After 6 minutes, it shows that progress is 18302.850%. Remaining says " ---". Job completed after ~10 minutes and shows that it is "Completed and Validated".

Let me know if you need more info.
captainjack

Profile ritterm
Avatar
Send message
Joined: 31 Jul 09
Posts: 88
Credit: 244,413,897
RAC: 0
Level
Leu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 35507 - Posted: 5 Mar 2014 | 1:58:49 UTC
Last modified: 5 Mar 2014 | 2:30:13 UTC

I'm not sure what the expected run times are, but the tasks I've completed have run between about 5000-6500 seconds.

Observations for my hosts (Win7-64, 3 Intel, 1 AMD):

1) Many tasks still running long after progress has reached 100%. WU properties show nonsensical values for the fraction done.
2) When suspended then resumed, those tasks finished shortly after starting to run again*
3) I've seen progress values in a wide range between 0%-100%. Often, WU properties fraction done does not match progress value.

*[edit]Except when they don't finish shortly after restarting.[/edit]
____________

Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 35509 - Posted: 5 Mar 2014 | 3:07:57 UTC

In addition to observing the same behaviors others have already posted ...
I just attached a Win7 x64 client that does not have an NVidia card and while the server status says there are no CPU tasks available the event log message doesn't match ... says no "long" tasks available?

3/4/2014 10:04:45 PM | GPUGRID | update requested by user
3/4/2014 10:04:48 PM | GPUGRID | Sending scheduler request: Requested by user.
3/4/2014 10:04:48 PM | GPUGRID | Requesting new tasks for CPU
3/4/2014 10:04:50 PM | GPUGRID | Scheduler request completed: got 0 new tasks
3/4/2014 10:04:50 PM | GPUGRID | No tasks sent
3/4/2014 10:04:50 PM | GPUGRID | No tasks are available for Long runs (8-12 hours on fastest card)
____________
Thanks - Steve

Dylan
Send message
Joined: 16 Jul 12
Posts: 98
Credit: 386,043,752
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwat
Message 35510 - Posted: 5 Mar 2014 | 4:21:59 UTC - in response to Message 35509.

Look in your preferences. You have it set somewhere to only do Long runs and not do the cpu tasks.

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1947
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 35511 - Posted: 5 Mar 2014 | 7:51:00 UTC - in response to Message 35510.

Thanks for the help guys.

Probably the progress info is not working.

How many CPU jobs do you get per CPU core?

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 35516 - Posted: 5 Mar 2014 | 9:29:50 UTC - in response to Message 35511.
Last modified: 5 Mar 2014 | 9:32:15 UTC

It's 'common' to see WU's reach 100% and run for a bit before completing. You just need to make sure the % stops at 100, and allow the estimated run times to be more accurate - didn't look but presumably the GFlops value needs to be set more accurately.

On an 8 core CPU, I expect you could get up to 8 CPU tasks. At times I was running Five CPU only 1.01 apps (as I was attached to other CPU projects, and running 2 GPU tasks). So one task per logical core/thread (on an i7 with 8 logical threads).
____________
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Snow Crash
Send message
Joined: 4 Apr 09
Posts: 450
Credit: 539,316,349
RAC: 0
Level
Lys
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 35517 - Posted: 5 Mar 2014 | 10:33:11 UTC - in response to Message 35511.

Thanks for the help guys.

Probably the progress info is not working.

How many CPU jobs do you get per CPU core?

gdf

I didn't test this specifically and while I never got more than 1 per "core" that was likely only because it was the amount of work BOINC thought I needed to meet my cache settings. If you release some more and I can test when I get home from work tonight - about 12 hours from now.

____________
Thanks - Steve

Profile MJH
Project administrator
Project developer
Project scientist
Send message
Joined: 12 Nov 07
Posts: 696
Credit: 27,266,655
RAC: 0
Level
Val
Scientific publications
watwat
Message 35521 - Posted: 5 Mar 2014 | 12:33:35 UTC - in response to Message 35506.
Last modified: 5 Mar 2014 | 12:34:14 UTC

Yeah, I broke the progress reporting somehow. It's just a cosmetic problem, though admittedly an annoying one.

Matt

ALAIN_13013
Avatar
Send message
Joined: 11 Sep 08
Posts: 18
Credit: 1,547,879,462
RAC: 0
Level
His
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 35525 - Posted: 5 Mar 2014 | 15:48:24 UTC - in response to Message 35511.

Thanks for the help guys.

Probably the progress info is not working.

How many CPU jobs do you get per CPU core?

gdf


24 WU for 24 CPU core for me
____________

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 35541 - Posted: 6 Mar 2014 | 21:01:30 UTC - in response to Message 35525.

Please, sir, I want some more...
____________
FAQ's

HOW TO:
- Opt out of Beta Tests
- Ask for Help

Werkstatt
Send message
Joined: 23 May 09
Posts: 121
Credit: 258,138,755
RAC: 0
Level
Asn
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 35567 - Posted: 9 Mar 2014 | 8:13:01 UTC

I have added two devices to gpugrid two days ago; have not got a single wu ...

Profile GDF
Volunteer moderator
Project administrator
Project developer
Project tester
Volunteer developer
Volunteer tester
Project scientist
Send message
Joined: 14 Mar 07
Posts: 1947
Credit: 629,356
RAC: 0
Level
Gly
Scientific publications
watwatwatwatwat
Message 35568 - Posted: 9 Mar 2014 | 8:23:01 UTC - in response to Message 35567.

we need to do still a couple of changes to the application for the science and for stability.

In particular one that packs for work for WU.

gdf

Variable
Send message
Joined: 20 Nov 13
Posts: 21
Credit: 364,641,105
RAC: 85
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 35588 - Posted: 10 Mar 2014 | 18:03:50 UTC - in response to Message 35491.


These are pathfinder WUs for a new project that we are currently scoping. Please keep crunching them, it's giving us useful information. More details on the project soon!

Matt


Is this a test for Android (I hope)? If so will it be in the list of available projects in the BOINC app?

Dylan
Send message
Joined: 16 Jul 12
Posts: 98
Credit: 386,043,752
RAC: 0
Level
Asp
Scientific publications
watwatwatwatwatwatwat
Message 35596 - Posted: 10 Mar 2014 | 20:13:17 UTC - in response to Message 35588.

You can already run them on an Android device.

Variable
Send message
Joined: 20 Nov 13
Posts: 21
Credit: 364,641,105
RAC: 85
Level
Asp
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 35611 - Posted: 11 Mar 2014 | 13:52:07 UTC

I see GPUgrid in the list of projects now in the BOINC app, but it says 'device not supported' and doesn't seem to be downloading work once I add the project. I have a Galaxy S4 phone.

Trotador
Send message
Joined: 25 Mar 12
Posts: 103
Credit: 2,510,352,816
RAC: 32,607
Level
Phe
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 35621 - Posted: 12 Mar 2014 | 20:41:36 UTC

I successfully run one of these units several days ago. Today, three of them have failed with the same "segment violation" error, they all in Linux in the same PC.

<core_client_version>6.10.58</core_client_version>
<![CDATA[
<message>
process exited with code 193 (0xc1, -63)
</message>
<stderr_txt>
# (BOINC) Mapping [ligand.db]->[ligand.db]
# (BOINC) Mapping [result.db]->[../../projects/www.gpugrid.net/T2-TONI_DBTEST1-0-1-RND0994_0_0]
# (BOINC) Mapping [protein.pdbqt]->[protein.pdbqt]
# (BOINC) Mapping [flexible.pdbqt]->[flexible.pdbqt]
SIGSEGV: segmentation violation
Stack trace (7 frames):
[0x4932ad]
[0x5242b0]
[0x4a4f61]
[0x4a3a1c]
[0x40a8eb]
[0x525aeb]
[0x400449]

Exiting...

</stderr_txt>
]]>

Post to thread

Message boards : Number crunching : CPU only app v1.04

//