Advanced search

Message boards : GPU Users Group message board : Work

Author Message
Zalster
Avatar
Send message
Joined: 26 Feb 14
Posts: 211
Credit: 4,496,324,562
RAC: 46,448
Level
Arg
Scientific publications
watwatwatwatwatwatwatwat
Message 48896 - Posted: 11 Feb 2018 | 4:46:13 UTC

Seems like I'm picking up work from Linux machines that have errored out the work units.
____________

Keith Myers
Send message
Joined: 13 Dec 17
Posts: 676
Credit: 883,394,807
RAC: 849,439
Level
Glu
Scientific publications
watwatwatwatwat
Message 48897 - Posted: 11 Feb 2018 | 7:08:16 UTC - in response to Message 48896.

As rare as the short ACEMD tasks are . . . you are welcome to mine.

I'll not let them go without a fight once our app gets fixed.

Zalster
Avatar
Send message
Joined: 26 Feb 14
Posts: 211
Credit: 4,496,324,562
RAC: 46,448
Level
Arg
Scientific publications
watwatwatwatwatwatwatwat
Message 48936 - Posted: 14 Feb 2018 | 11:05:00 UTC - in response to Message 48897.

Been a good run this week. Seems like almost everyday I get a few work units.
____________

Keith Myers
Send message
Joined: 13 Dec 17
Posts: 676
Credit: 883,394,807
RAC: 849,439
Level
Glu
Scientific publications
watwatwatwatwat
Message 48968 - Posted: 16 Feb 2018 | 2:12:26 UTC

Just got the new 9.19 application and can crunch gpu tasks again on the Linux cruncher. Amazed to see a large batch of both short and long tasks in the RTS buffer. Grabbed half a dozen for both Ryzen machines. Always nice to pad the overall BOINC credit stats with the big payouts from GPUGrid. But I don't want to lose my new rankings at SETI with too much time away from crunching for SETI.

Zalster
Avatar
Send message
Joined: 26 Feb 14
Posts: 211
Credit: 4,496,324,562
RAC: 46,448
Level
Arg
Scientific publications
watwatwatwatwatwatwatwat
Message 50760 - Posted: 30 Oct 2018 | 4:59:36 UTC - in response to Message 48968.

Wow, been a while since we posted in here. Not that there are many of us here anymore.

Keith has correctly determined that the SWAN_SYNC=1 will work in Linux but you need to be running BOINC 7.4.44 which is available at Cruncher's Anonymous website in the all in one download. Look in DOCS in the boinc folder and you will find the ZIP file for BOINC 7.4.44

I make the move to Linux once Petri got his special cuda for Seti refine. Bonus is Ubuntu also works great here and at Einstein as well. If you have any questions. Please post and we will get back to you as soon as we can.
____________

Keith Myers
Send message
Joined: 13 Dec 17
Posts: 676
Credit: 883,394,807
RAC: 849,439
Level
Glu
Scientific publications
watwatwatwatwat
Message 50871 - Posted: 15 Nov 2018 | 16:45:32 UTC

Yes, this thread sure was stagnant. I was happy to do the detective work to figure out swan_sync when another forum inhabitant pointed out the anomaly on one of my hosts. Why one but not the others? Sure does make a difference in run times. I would have thought it not that important. Goes to show that GPUGrid,like Einstein are high bandwidth users of memory.

Keith Myers
Send message
Joined: 13 Dec 17
Posts: 676
Credit: 883,394,807
RAC: 849,439
Level
Glu
Scientific publications
watwatwatwatwat
Message 51467 - Posted: 12 Feb 2019 | 1:09:24 UTC

If I have installed the anaconda python environment for something else on a system, would that satisfy the requirements for the cpu tasks? Or would it have to go through the standard project installation to get the requisite programs installed other than the obvious science application?

Zalster
Avatar
Send message
Joined: 26 Feb 14
Posts: 211
Credit: 4,496,324,562
RAC: 46,448
Level
Arg
Scientific publications
watwatwatwatwatwatwatwat
Message 51468 - Posted: 12 Feb 2019 | 2:12:49 UTC - in response to Message 51467.

I don't know if you have to install that program. The error message that appears is a report to the developer to update his program to the newest anaconda. It should just work. I've never installed anything, just downloaded the program. I have however, had to put a max concurrent in to limit it to 3 on one machine and 2 on another. You can also change how many cores it runs on. I let it use 4 threads but other limit it to 2.
____________

Keith Myers
Send message
Joined: 13 Dec 17
Posts: 676
Credit: 883,394,807
RAC: 849,439
Level
Glu
Scientific publications
watwatwatwatwat
Message 51469 - Posted: 12 Feb 2019 | 3:59:20 UTC

The cpu app definitely needs the miniconda environment installed to allow the app to run. Most of the code runs on python. The app installs the necessary resources to run the app when it downloads the first cpu task for the app.

I was just asking whether the fact I have already installed the full anaconda platform negated the need to install anything automatically by the first running of a cpu task because everything was already in place other than the science app needing to be downloaded.

Keith Myers
Send message
Joined: 13 Dec 17
Posts: 676
Credit: 883,394,807
RAC: 849,439
Level
Glu
Scientific publications
watwatwatwatwat
Message 51500 - Posted: 16 Feb 2019 | 2:52:19 UTC

Well I just dumped 3 gpu tasks on one host. They were in progress when I shut down to upgrade to RTX 2080 Hybrids. Forgot to exclude the gpu from the project.

Caught the other one in time before it trashed work. Also remembered I can't use a max_concurrent statements anymore or it stops cpu tasks from processing.

Have to limit cpu usage the old way with local preferences.

Sure wish they would fix the client.

Zalster
Avatar
Send message
Joined: 26 Feb 14
Posts: 211
Credit: 4,496,324,562
RAC: 46,448
Level
Arg
Scientific publications
watwatwatwatwatwatwatwat
Message 51539 - Posted: 21 Feb 2019 | 18:28:20 UTC - in response to Message 51500.

Good News,

My house guest are leaving so I can now fire back up my machines. Hope to have them all crunching away this weekend.


Z
____________

Keith Myers
Send message
Joined: 13 Dec 17
Posts: 676
Credit: 883,394,807
RAC: 849,439
Level
Glu
Scientific publications
watwatwatwatwat
Message 51543 - Posted: 21 Feb 2019 | 22:11:42 UTC

Well I should have a nice bump in my completed today. I lost my internet connection in the middle of the night and couldn't get any work or report till I rebooted the gateway. So all of the hosts were out of work and also that all my caches from other projects were completed once Seti ran out.

Zalster
Avatar
Send message
Joined: 26 Feb 14
Posts: 211
Credit: 4,496,324,562
RAC: 46,448
Level
Arg
Scientific publications
watwatwatwatwatwatwatwat
Message 51563 - Posted: 23 Feb 2019 | 20:46:36 UTC

Hey LUCY.....I'm home!!!
____________

Keith Myers
Send message
Joined: 13 Dec 17
Posts: 676
Credit: 883,394,807
RAC: 849,439
Level
Glu
Scientific publications
watwatwatwatwat
Message 51564 - Posted: 23 Feb 2019 | 22:03:27 UTC

Welcome back Ricky!

Post to thread

Message boards : GPU Users Group message board : Work