Advanced search

Message boards : Graphics cards (GPUs) : Work Fetch policy

Author Message
Profile Paul D. Buck
Send message
Joined: 9 Jun 08
Posts: 1050
Credit: 37,321,185
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 5068 - Posted: 30 Dec 2008 | 7:45:16 UTC

There is a discussion underway that may be of interest to those technically minded that have insight into BOINC and how it operates (you guys know who you are ... I have seen some of you around) who might be able to add some insight.

His proposal is shown here ... if you want my modification of that you will have to subscribe to the BOINC Dev mailing list or use the web page to scan the thread. There is a link on that page to the archive where you can see the discussion to date ... look for my name and click away ... it is down at the bottom of the page.

I am not sure how my suggestions will be received, history is not on my side ... but, comments are important for this as it will affect how GPU resources and other resources to come are handled by BOINC. If we don't get it right early we will be paying for it later...

One of my strongest suggestions is the end the one Non-CPU intense tasks per project per computer. I argue instead for a per core limit and allowing for the queuing of these tasks so that outages do not immediately halt work for these projects (FreeHAL, DepSPDR (now off line), etc.).

If you PM me with your e-mail address I will send my modified RTF file assuming I don't get inundated with requests.

ExtraTerrestrial Apes
Volunteer moderator
Volunteer tester
Avatar
Send message
Joined: 17 Aug 08
Posts: 2705
Credit: 1,311,122,549
RAC: 0
Level
Met
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 5478 - Posted: 10 Jan 2009 | 20:25:44 UTC - in response to Message 5068.

Sorry, still didn't get around to reading through this stuff. Just caught up on the forum posts of the last week..

MrS
____________
Scanning for our furry friends since Jan 2002

Profile Paul D. Buck
Send message
Joined: 9 Jun 08
Posts: 1050
Credit: 37,321,185
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 5484 - Posted: 11 Jan 2009 | 0:56:27 UTC - in response to Message 5478.

Sorry, still didn't get around to reading through this stuff. Just caught up on the forum posts of the last week..

MrS


Boring as heck ...

And terribly important ... though all signs to this point indicate that input is sought so it can be ignored...

At least to this point, as best as I can tell, none of the commentary and suggestions has caused Dr. Anderson to alter his path one iota ...

MarkJ
Volunteer moderator
Volunteer tester
Send message
Joined: 24 Dec 08
Posts: 738
Credit: 200,909,904
RAC: 0
Level
Leu
Scientific publications
watwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwatwat
Message 5713 - Posted: 17 Jan 2009 | 10:05:41 UTC

Rom has said they will be combining the work-fetch changes that Dr A has been working on in BOINC 6.6.2.

Profile Paul D. Buck
Send message
Joined: 9 Jun 08
Posts: 1050
Credit: 37,321,185
RAC: 0
Level
Val
Scientific publications
watwatwatwatwatwatwatwatwatwat
Message 5721 - Posted: 17 Jan 2009 | 13:25:59 UTC - in response to Message 5713.

Rom has said they will be combining the work-fetch changes that Dr A has been working on in BOINC 6.6.2.


The codicil to that is that this may delay the introduction of the 6.6.x version because of the need to finish the fixes ...

And as I have noted elsewhere the 6 version breaks OS-X systems with some projects because of the changes to CPU strings and I think I saw that with the linux system I used to have ... YMMV ...
____________

Post to thread

Message boards : Graphics cards (GPUs) : Work Fetch policy

//