With two GTX 750 Ti, How do I do: Cuda50 4/gpu (1or2 GPU) and NV_SoG 2/gpu (max 1 GPU)

Message boards : Number crunching : With two GTX 750 Ti, How do I do: Cuda50 4/gpu (1or2 GPU) and NV_SoG 2/gpu (max 1 GPU)
Message board moderation

To post messages, you must log in.

AuthorMessage
Profile Stubbles
Volunteer tester
Avatar

Send message
Joined: 29 Nov 99
Posts: 358
Credit: 5,909,255
RAC: 0
Canada
Message 1816039 - Posted: 10 Sep 2016, 0:41:18 UTC
Last modified: 10 Sep 2016, 0:45:09 UTC

Hello "Boinc client file" gurus
After reading one of Richard's post,
it got me wondowering about doing a similar scenario on my rig with two GTX 750 Ti.

I already have a semi-automated script to move tasks from CPU queue to any GPU app queue.
I tried editing app_info.xml to add parts of the Cuda50 setup that was in:
C:\ProgramData\BOINC\projects\setiathome.berkeley.edu\oldApp_backup\
but that failed and I guosted many tasks and screwed up my Lunatics NV_SoG setup.

So I reinstalled Lunatics with Cuda50, and right now, I'm only trying to modify app_config.xml in a similar way to another of Richard's post.
Unfortunately, the Boinc Wiki doesn't even mention some of the fields that Richard used: http://boinc.berkeley.edu/trac/wiki/ClientAppConfig

( FYI, with Lunatics v0.45 Beta4, I can move tasks with my script to any of the GPU <plan_class>. ex: <plan_class>cuda50</plan_class>
and it will only use the GPU app selected during the Lunatics setup no matter what the GPU <plan_class> is. )

Unfortunately, I'm not getting the desired GPU processing that I want:
- Cuda50 4/gpu (1or2 GPU), and
- NV_SoG 2/gpu (0or1 GPU)

I get the following:
- 8cuda/2gpu (Good!)
- 2SoG/2gpu (Bad: I want 2SoG/1gpu, and 4cuda/1gpu)
- 4cuda/1gpu, and 1SoG/1gpu (1st part is right but second should be 2Sog/1gpu)

Here's my: app_config.xml
<app_config>
	<app>
		<name>setiathome_v8</name>
		<gpu_versions>
			<gpu_usage>0.25</gpu_usage>
			<cpu_usage>0.25</cpu_usage>
		</gpu_versions>
	</app>
	<app_version>
		<app_name>setiathome_v8</app_name>
		<plan_class>opencl_nvidia_SoG</plan_class>
		<avg_ncpus>1.0</avg_ncpus>
		<ngpus>1.0</ngpus>
		<gpu_versions>
			<gpu_usage>0.50</gpu_usage>
			<cpu_usage>0.50</cpu_usage>
		</gpu_versions>
	</app_version>
</app_config>

note: I've removed the "astropulse_v7" section in order to make it shorter.

[edit]I'm guessing it has to do with:
<avg_ncpus>1.0</avg_ncpus> 
<ngpus>1.0</ngpus>
[/e]

Once I get the desired behaviour, I'll ask afterwards how to modify app_info.xml in order to use the following two GPU apps .exe
- cuda50: Lunatics_x41zi_win32_cuda50.exe
- NV_SoG: MB8_win_x86_SSE3_OpenCL_NV_SoG_r3500.exe

Any help is appreciated.
Cheers,
RobG :-)
ID: 1816039 · Report as offensive
Richard Haselgrove Project Donor
Volunteer tester

Send message
Joined: 4 Jul 99
Posts: 14650
Credit: 200,643,578
RAC: 874
United Kingdom
Message 1816116 - Posted: 10 Sep 2016, 7:50:32 UTC - in response to Message 1816039.  

Go back and read the documentation for the extended version of app_config.xml - the syntax for the <app_version> is very different from the simple <app> case.

Actually, there's a working example a couple of posts earlier than the one you linked - message 1813746.
ID: 1816116 · Report as offensive
Darth Beaver Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Avatar

Send message
Joined: 20 Aug 99
Posts: 6728
Credit: 21,443,075
RAC: 3
Australia
Message 1816597 - Posted: 12 Sep 2016, 5:35:09 UTC - in response to Message 1816039.  

Stubbles doing Cuda 50 and SoG at the same time I have to ask WHY?

SoG file nonvlar 2 at a time on my 680 and 970 approx. 9-12 mins witch means approx. 4-6 mins a unit

Cuda file 2 at a time approx. the same times as the Open CL SoG files after you use 1 tweek in the mb_cmdline_win_x86_SSE3_OpenCL_NV_SoG file

And you have loaded latest Lunatics version and then after that it will only do SoG as you won't get any Cuda if you didn't chose when you installed it .

lunatics I thinks only lets you do 1 or the other
CUDA or SoG

SoG Vlar 2 at a time maximum time on 680 is 27mins and the 970 approx. 20 mins

Cuda vlar 1 at a time 1 hrs plus up to 3 hrs
ID: 1816597 · Report as offensive

Message boards : Number crunching : With two GTX 750 Ti, How do I do: Cuda50 4/gpu (1or2 GPU) and NV_SoG 2/gpu (max 1 GPU)


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.