New application
log in

Advanced search

Message boards : News : New application

Author Message
Vadimas Starikovicius
Project administrator
Project developer
Project scientist
Send message
Joined: 3 Jan 12
Posts: 80
Credit: 255,109
RAC: 720
Message 143 - Posted: 12 Dec 2015, 2:17:33 UTC
Last modified: 25 Dec 2015, 20:45:50 UTC

Dear VGTU@Home users,

We would like to announce a new beginning of our project. We have finally managed to find and develop a true scientific application for our project. With your help, we hope to obtain interesting results!

metalius
Avatar
Send message
Joined: 8 Feb 12
Posts: 12
Credit: 3,584,670
RAC: 1,330
Message 145 - Posted: 14 Dec 2015, 20:27:56 UTC

Sveiki!
Puiku, kad projektas vėl aktyvus. Dar puikiau, kad testavimo fazė baigta, ir šį kartą jau užsiimsime rimtais dalykais. Ypač jei eksperimento rezultatai turės praktinį pritaikymą medicinoje ir galbūt kitose srityse.
Dar kartą sveiki sugrįžę ir sėkmės jums!
____________

Ananas
Send message
Joined: 9 Apr 12
Posts: 9
Credit: 185,157
RAC: 0
Message 146 - Posted: 17 Dec 2015, 17:49:58 UTC

wb :-)

Alez
Send message
Joined: 22 May 12
Posts: 3
Credit: 1,127,493
RAC: 0
Message 147 - Posted: 21 Dec 2015, 1:55:33 UTC

Great to see you back. Is this a test of a new project or do you intend for there to be new work available on a regular basis ?

Vadimas Starikovicius
Project administrator
Project developer
Project scientist
Send message
Joined: 3 Jan 12
Posts: 80
Credit: 255,109
RAC: 720
Message 148 - Posted: 21 Dec 2015, 10:08:03 UTC - in response to Message 147.
Last modified: 21 Dec 2015, 10:09:05 UTC

There will be some work with this new application on dynamic visual cryptography for 10-12 months for sure. We plan scientific publication of the results. Depending on the success (if the paper will be accepted in one or another journal), we will continue the work on this problem with our colleagues from KTU.

Using this opportunity, I want to ensure the users of project that I will not generate work and spawn workunits just to keep the project artificially alive.

As you can see from the history of the project, computations with DES application were stopped, when we had nothing new to test.

Going back to the current application, I have started with relatively small-size problem. I didn’t know how many active users are left and what will be the speed of computations. All workunits were sent in 34 hours. Now we need to wait for all results to return before starting the solution of the new problem.

Best regards

Alez
Send message
Joined: 22 May 12
Posts: 3
Credit: 1,127,493
RAC: 0
Message 149 - Posted: 22 Dec 2015, 2:25:14 UTC
Last modified: 22 Dec 2015, 2:33:51 UTC

Thanks for the information. Will remain attached and await the next batch of units, although if the work is moving through the system fast enough you may consider running more than one problem at the same time. Crunchers tend to prefer a steady flow of work or are you using the results from the previous batch to generate the problem for the next batch ?
Good luck with the research and publications.

lionelc
Send message
Joined: 28 Apr 13
Posts: 1
Credit: 101,248
RAC: 0
Message 150 - Posted: 23 Dec 2015, 9:41:25 UTC

You could limit the maximum wu in progress per host.
It will speed up the time to get the completed tasks back by spreading work on all available machines.
Seems like all work units of current batch have been captured by only a few hosts.
We are waiting them to finish since days while all our hungry machines, ready to crunch, are left unused.
Best Regards

Vadimas Starikovicius
Project administrator
Project developer
Project scientist
Send message
Joined: 3 Jan 12
Posts: 80
Credit: 255,109
RAC: 720
Message 152 - Posted: 25 Dec 2015, 20:58:16 UTC - in response to Message 150.

I have limited the number of workunits in progress to 50. Thank you for the advice. We will not solve several problems at once, so far.

Best regards

Profile Coleslaw
Avatar
Send message
Joined: 19 Feb 12
Posts: 3
Credit: 1,083,173
RAC: 0
Message 153 - Posted: 28 Dec 2015, 17:29:53 UTC - in response to Message 152.

Is that 50 per host or 50 per user? I would caution limiting per user as there are many of us with server grade systems in our homes. I have one 4P system with 48 cores alone. However, limiting to 50 per box would still be a short limit for that rig, it would certainly be much better than 50 per user...
____________

Profile Steve Hawker*
Send message
Joined: 11 Feb 13
Posts: 1
Credit: 29,635
RAC: 0
Message 154 - Posted: 30 Dec 2015, 1:19:08 UTC - in response to Message 153.

Is that 50 per host or 50 per user? I would caution limiting per user as there are many of us with server grade systems in our homes. I have one 4P system with 48 cores alone. However, limiting to 50 per box would still be a short limit for that rig, it would certainly be much better than 50 per user...


I'm getting 200 tasks on my 4 core machine so 50 per core I'd say

Vadimas Starikovicius
Project administrator
Project developer
Project scientist
Send message
Joined: 3 Jan 12
Posts: 80
Credit: 255,109
RAC: 720
Message 155 - Posted: 31 Dec 2015, 11:47:56 UTC - in response to Message 154.
Last modified: 31 Dec 2015, 16:45:14 UTC

Yes. Currently, it is 50 tasks per core: max_wus_in_progress in the project settings.

Ananas
Send message
Joined: 9 Apr 12
Posts: 9
Credit: 185,157
RAC: 0
Message 156 - Posted: 1 Jan 2016, 5:08:56 UTC

Somehow strange for a single threaded application :

05:10:53 (3448): WU CPU time: 5418.81
Elapsed wall-clock time: 0 4958.1 sum = 4958.1

Message boards : News : New application


Main page · Your account · Message boards


Copyright © 2017 VGTU