BOINC Version 7.0.42 is released

Message boards : Graphics cards (GPUs) : BOINC Version 7.0.42 is released

Author	Message
Rantanplan Send message Joined: 22 Jul 11 Posts: 166 Credit: 138,629,987 RAC: 0 Level Scientific publications	Message 27711 - Posted: 16 Dec 2012 \| 22:05:56 UTC Last modified: 16 Dec 2012 \| 22:06:40 UTC
	Hello, not only a new software have been released, there is more. http://boinc.berkeley.edu/trac/wiki/ClientAppConfig This will allow multiple gpu tasks per gpu, or didnt understand that right ? But how exactly do i this. Any guesses ? Thx, greetings from germany
	ID: 27711 \| Rating: 0 \| rate: / Reply Quote

skgiven Volunteer moderator Volunteer tester Send message Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level Scientific publications	Message 27712 - Posted: 17 Dec 2012 \| 11:42:47 UTC - in response to Message 27711. Last modified: 17 Dec 2012 \| 12:32:28 UTC
	At present the app_config file isn't really applicable to GPUGrid. Perhaps there will be tasks in the future that can use this, but most current tasks have high GPU utilization and when you run more than one task at a time there is no increase in overall throughput. Another concern is memory capacity of GPU's. Here some of Nate's tasks require large amounts of GPU memory. Most GPU's would not have sufficient memory to run two tasks. Not that Nate's tasks have poor GPU utilization anyway. The main benefit of this <app_config> file is to replace the app_info files that people have to create (containing much more project information), and maintain; when a new app is released. app_config should be much easier to create and maintain. The AMD/ATI GPU's are generally more adept at running multiple GPU tasks. While I wouldn't recommend running multiple tasks on GF200 cards, you can run multiple tasks on Fermis. I think Kepler cards are more on a power with the AMD cards in this respect. ____________ FAQ's HOW TO: - Opt out of Beta Tests - Ask for Help
	ID: 27712 \| Rating: 0 \| rate: / Reply Quote

Rantanplan Send message Joined: 22 Jul 11 Posts: 166 Credit: 138,629,987 RAC: 0 Level Scientific publications	Message 27713 - Posted: 17 Dec 2012 \| 11:56:55 UTC - in response to Message 27711. Last modified: 17 Dec 2012 \| 11:57:38 UTC
	17.12.2012 12:52:04 \| GPUGRID \| This computer has reached a limit on tasks in progress Getting this message, no tasks in quere , what can i do ? Running on kepler.
	ID: 27713 \| Rating: 0 \| rate: / Reply Quote

skgiven Volunteer moderator Volunteer tester Send message Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level Scientific publications	Message 27714 - Posted: 17 Dec 2012 \| 12:35:12 UTC - in response to Message 27713. Last modified: 17 Dec 2012 \| 12:36:26 UTC
	It means your Boinc cache is saturated (has enough tasks in queue and won't get any more until one task is competed). At GPUGrid, each GPU can only be allocated 2 tasks, no matter what your cache settings are (0.05days or 5.0days). This is because GPUGrid requires a fast turn around of work. So no more than one running (per GPU) and one in the queue. ____________ FAQ's HOW TO: - Opt out of Beta Tests - Ask for Help
	ID: 27714 \| Rating: 0 \| rate: / Reply Quote

Rantanplan Send message Joined: 22 Jul 11 Posts: 166 Credit: 138,629,987 RAC: 0 Level Scientific publications	Message 27716 - Posted: 17 Dec 2012 \| 13:09:31 UTC - in response to Message 27714. Last modified: 17 Dec 2012 \| 13:10:59 UTC
	What i see is, there is no benefit in running 2 tasks per gpu. Nothing less i need to know how i bring BOINC to load the tasks earlier that i had uploaded one tasks. I dont what to get a task 12 hours earlier then i get no 24 hour bonus. The waiting time is so high, that i even crunsh a task by 5-6% in uploading time.
	ID: 27716 \| Rating: 0 \| rate: / Reply Quote

Rantanplan Send message Joined: 22 Jul 11 Posts: 166 Credit: 138,629,987 RAC: 0 Level Scientific publications	Message 27717 - Posted: 17 Dec 2012 \| 15:02:30 UTC - in response to Message 27716. Last modified: 17 Dec 2012 \| 15:03:21 UTC
	ok here is my app_config.xml file: Using Q6600 + GTX 670 <app_config> <app> <name>acemdlong</name> <max_concurrent>4</max_concurrent> <gpu_versions> <gpu_usage>.5</gpu_usage> <cpu_usage>1.25</cpu_usage> </gpu_versions> </app> </app_config> I hope i will get then wus when uploading, right !?
	ID: 27717 \| Rating: 0 \| rate: / Reply Quote

skgiven Volunteer moderator Volunteer tester Send message Joined: 23 Apr 09 Posts: 3968 Credit: 1,995,359,260 RAC: 0 Level Scientific publications	Message 27719 - Posted: 17 Dec 2012 \| 17:49:42 UTC - in response to Message 27717.
	There is no benefit in using app_config. ____________ FAQ's HOW TO: - Opt out of Beta Tests - Ask for Help
	ID: 27719 \| Rating: 0 \| rate: / Reply Quote

ExtraTerrestrial Apes Volunteer moderator Volunteer tester Send message Joined: 17 Aug 08 Posts: 2705 Credit: 1,311,122,549 RAC: 0 Level Scientific publications	Message 27721 - Posted: 17 Dec 2012 \| 21:40:14 UTC
	Sounds like you should adjust your cache size: not too large, so that the next WUs is not fetched too early. But not too low, so that you don't run dry upon uploading a result. MrS ____________ Scanning for our furry friends since Jan 2002
	ID: 27721 \| Rating: 0 \| rate: / Reply Quote

Post to thread

Message boards : Graphics cards (GPUs) : BOINC Version 7.0.42 is released

	About	Science	Volunteers	Performance	Forum	Join us	Donate