| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
| World Community Grid Forums
|
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 41
|
|
| Author |
|
|
John Fontanella
Cruncher Joined: Nov 28, 2012 Post Count: 13 Status: Offline Project Badges:
|
Hey Ya'll,
I just set up a second computer. Boinc 7.0.52 I am running two Radeon HD 7770's (not sli'd) on an intel i5-2400 Windows 7 64-bit Can you help me set up another app_config? |
||
|
|
John Fontanella
Cruncher Joined: Nov 28, 2012 Post Count: 13 Status: Offline Project Badges:
|
Hey branjo,
I believe it is running 8 concurrent tasks. My GPU is about 95% most of the time. I haven't tried any other app_config setups because this one has been so succesfull. I'm not sure what cc_config is. Can you help me with that? I also don't know what shrubbing is. |
||
|
|
branjo
Master Cruncher Slovakia Joined: Jun 29, 2012 Post Count: 1892 Status: Offline Project Badges:
|
in cc_config you can set (on the top of lot of other things) to crunch on all GPUs.
----------------------------------------Check if you already have cc_config.xml in c:/ProgramData/BOINC If yes, check if it contains line SekeRob mentioned here. If you do not have cc_config.xml file, create one and the only lines you need are: <cc_config> <options> <use_all_gpus>1</use_all_gpus> </options> </cc_config> Now you can try 12 concurrent tasks on 2 GPU's + 4 CPU threads (6 per GPU): <app_config> <app> <name>hcc1</name> <gpu_versions> <gpu_usage>.166666</gpu_usage> <cpu_usage>.333333</cpu_usage> </gpu_versions> </app> </app_config> or 8 concurrent tasks on your 2 GPU's + 4 CPU threads (4 per GPU): <app_config> <app> <name>hcc1</name> <gpu_versions> <gpu_usage>.25</gpu_usage> <cpu_usage>.5</cpu_usage> </gpu_versions> </app> </app_config> or even 16 concurrent tasks on your 2 GPU's + 4 CPU threads (8 per GPU): <app_config> <app> <name>hcc1</name> <gpu_versions> <gpu_usage>.125</gpu_usage> <cpu_usage>.25</cpu_usage> </gpu_versions> </app> </app_config> (Don't forget to restart BOINC) Cheers ![]() Edit to add: shrubbing = crunching (in the language spoken by The Knights Who Say NI!) ![]() ![]() Crunching@Home since January 13 2000. Shrubbing@Home since January 5 2006 ![]() [Edit 2 times, last edit by branjo at Feb 16, 2013 9:16:23 PM] |
||
|
|
branjo
Master Cruncher Slovakia Joined: Jun 29, 2012 Post Count: 1892 Status: Offline Project Badges:
|
Edit to add: When you change app_config, restart your comp You don't need to restart your computer when you make changes to the app_config file. You only need to shut down restart BOINC. @branjo- It's he. ![]() Thanks for info nanoprobe, it will be easier to quote you Cheers ![]() ![]() Crunching@Home since January 13 2000. Shrubbing@Home since January 5 2006 ![]() [Edit 1 times, last edit by branjo at Feb 16, 2013 6:36:22 PM] |
||
|
|
jay_Orlando
Senior Cruncher USA Joined: Jan 4, 2006 Post Count: 189 Status: Offline Project Badges:
|
Hi.
----------------------------------------I have a question about the post 5th from the beginning. Is there a typo there? The last two lines look garbled - syntactically. I broke out the last line and 'prettied' it up. The '/' of 'max_concurrent' looks wrong - and I can't figure out the intent anyway.
Is the "/max_concurrent>" correct? Its not balanced anywhere. The final </app_config> is balanced to the <app_config> at the very first. To me if looks like: <app_config> -- several phrases of <app><name>[some name]</name>[stuff]</app> </app_config> Then, the garbled line. I was just trying to make sense of it - and couldn't. (Par for my course.) Jay ![]() |
||
|
|
jay_Orlando
Senior Cruncher USA Joined: Jan 4, 2006 Post Count: 189 Status: Offline Project Badges:
|
Branjo,
----------------------------------------Could you, please, tell me some sort of formula of how you come up with the <cpu_usage> ???? Hmmmm. Better yet, can you, please, explain the intent of what <cpu_usage> is supposed to be? I'm not clear why the value is some sort of inverse-fraction. Is it supposed to say how may CPU cores are allocated for reporting of GPU tasks? The preferences define the total number and, preferably, the percentage of you actual cpu cores that you want to allow BOINC to use. How does <cpu_usage> relate to that? I wonder why, in the examples that I have seen of <cpu_usage> , if it derived by a formula, why define it? - Let BOINC calulate it from the formula? Yes, I am confused. Thanks in advance, Jay ![]() [Edit 1 times, last edit by jay_Orlando at Feb 24, 2013 2:42:27 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
That last line *after* the closing </app_config> in that overlong, much redundant content having app_config.xml, looks like it just wandered in out of place. Ignore. You don't need <max_concurrent> unless you want to expressly limit the *total* number of CPU+GPU processes for hcc1. His .083333 for the GPU means there's 12 allowed to run on the GPU, the <max_concurrent> of 17 he uses on most sections, means that after subtraction of of whatever runs on the GPU, the rest of a science can run on the CPU. In case that were 10 running actually on GPU, that would mean on overflow of 7 allowed on CPU, but since [3] cores are already reserved for the GPU, that 17 would only be achieved in the situation where 12 are running on the GPU, the maximum as determined by the .08333 fraction, which by the cpu factor of .25 means that 3 full cores are reserved for the GPU, leaving 5 CPU only jobs i.e. it must be an octo.
|
||
|
|
branjo
Master Cruncher Slovakia Joined: Jun 29, 2012 Post Count: 1892 Status: Offline Project Badges:
|
Branjo, Could you, please, tell me some sort of formula of how you come up with the <cpu_usage> ???? Hmmmm. Better yet, can you, please, explain the intent of what <cpu_usage> is supposed to be? I'm not clear why the value is some sort of inverse-fraction. Is it supposed to say how may CPU cores are allocated for reporting of GPU tasks? The preferences define the total number and, preferably, the percentage of you actual cpu cores that you want to allow BOINC to use. How does <cpu_usage> relate to that? I wonder why, in the examples that I have seen of <cpu_usage> , if it derived by a formula, why define it? - Let BOINC calulate it from the formula? Yes, I am confused. Thanks in advance, Jay Ji Jay, <gpu_usage>Z1</gpu_usage> states you how many concurrent GPU tasks you would like to run (not sure if per GPU in multiGPU rigs, or per total GPU's), so <gpu_usage>.083333</gpu_usage> means I am running 12 concurrent GPU tasks (1 / 12). <cpu_usage>Z2</cpu_usage> is stating how many CPU threads you would like to assign to these concurrent GPU tasks, so <cpu_usage>.25</cpu_usage> is defining 3 CPU threads for 12 GPU tasks (3 / 12 = 0.25). If I would like to dedicate all my 8 CPU threads, with the same <gpu_usage>, <cpu_usage> in app_config will be 8/12 = 0.666666: <gpu_versions> <gpu_usage> .083333 </gpu_usage> <cpu_usage> .666666 </cpu_usage> </gpu_versions> And so on: 2 CPU threads: 2/12 = 0.166666, 5 CPU threads. 5/12 = 0.416666 ... Cheers ![]() ![]() Crunching@Home since January 13 2000. Shrubbing@Home since January 5 2006 ![]() |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
[moi] Did update the official wiki to give the answer to "(not sure if per GPU in multiGPU rigs, or per total GPU's)" which is when a max_concurrent of 17 would have led with 2 cards to one or the other card being underutilized, and not knowing at all [not a matter of sure or unsure], if that would be a random fill out or in the order of device 0, device 1 etc ... not in the least important to know since the put-off for most addicted crunchers is "underutilized" ;)
----------------------------------------edit: The wiki link [Berkeley is up again after 2 days down] http://boinc.berkeley.edu/wiki/Cc_config.xml [Edit 1 times, last edit by Former Member at Feb 24, 2013 8:59:41 PM] |
||
|
|
jay_Orlando
Senior Cruncher USA Joined: Jan 4, 2006 Post Count: 189 Status: Offline Project Badges:
|
Branjo,
----------------------------------------Thank you. Rob, Thanks again. Your patience is great. Jay ![]() |
||
|
|
|