I have a machine ("arf-pc") which not long ago had 2x980TIs running 3 WUs each on an i7-950 HT ( so 8 "cores"). I do not do SETI on the CPU, so it is completely free for GPU support. The command line I was using:

-sbs 512 period_iterations_num 10 -hp

To try to lower electricity use (it was doing 550-600 watts) I sold the TIs and got a 1080 and a 750TI (thanks, Craigslist!). This cut electricity usage to under 400 watts, but RAC suffered somewhat. So I added a 2nd 750TI. But things got VERY sticky at the keyboard and graphically, so I went to 2 WUs/card (a total of 6 simultaneously). That pretty much cured the sticky.

Questions:

1) given the GPU setup of 1080 + 2X750TI, and running 6 WUs, what percent of the RAC with the 2x980TIs can I reasonably expect?

2) since the cmd line applies to all cards, how can I change it to adjust my new GPU setup for better RAC?

Thanks for your help...

---

2) since the cmd line applies to all cards, ...

There is special file you can create:
MultiBeam_NV_config.xml
... but I don't know if it is possible with stock apps.
If the app searches for it next to the .exe i.e. in SETI@home directory (<BOINC_Data>\projects\setiathome.berkeley.edu\) it will find it.
If the app searches only in current directory i.e. in slots\ the file will not be there.


From ReadMe_MultiBeam_OpenCL_NV_SoG.txt
"
For device-specific settings in multi-GPU systems it's possible to override some of command-line options via
application config file.

Name of this config file:
MultiBeam_<vendor>_config.xml where vendor can be ATi, NV or iGPU.
File structure:

<deviceN>
	<period_iterations_num>N</period_iterations_num>
	<spike_fft_thresh>N</spike_fft_thresh>
	<sbs>N</sbs>
	<oclfft_plan>
	        <size>N</size>
		<global_radix>N</global_radix>
		<local_radix>N</local_radix>
		<workgroup_size>N</workgroup_size>
		<max_local_size>N</max_local_size>
		<localmem_banks>N</localmem_banks>
		<localmem_coalesce_width>N</localmem_coalesce_width>
	</oclfft_plan>
	<no_caching>
</deviceN>

where deviceN - particular OpenCL device N starting with 0, multiple sections allowed, one per each device.
other fields - corresponding command-line options to override for this particular device.
All or some sections can be omitted.

Don't forget to re-check device number to physical card relation after driver update and physical slot change.
Both these actions can change cards enumeration order.

---

 


- ALF - "Find out what you don't do well ..... then don't do it!" :)
 

I'm running a GTX 1070 with a GTX 750Ti using the Lunatics installer.

My mb_cmdline_win_x86_SSE3_OpenCL_NV_SoG.txt command line is
-tt 1500 -hp -period_iterations_num 3 -sbs 768 -spike_fft_thresh 4096 -tune 1 64 1 4 -oclfft_tune_gr 256 -oclfft_tune_lr 16 -oclfft_tune_wg 256 -oclfft_tune_ls 512 -oclfft_tune_bn 64 -oclfft_tune_cw 64

with 1 core reserved per WU & running just 1 WU at a time.
-period_iterations_num 3 tends to make the screen & keyboard input a bit laggy, but I can live with it.
Setting it to 10 (or higher if necessary) should result in little if any lag.

---

Grant
Darwin NT

The use of MultiBeam_NV_config.xml Needs to be defined in appinfo.xml.

---

With each crime and every kindness we birth our future.

One problem you will see is that one command line will work great for the 1080 but too aggressive for the 750's. I'm sure there is a way to use separate command lines, but I'm not sure how. (EDIT: OK that was explained)

You can look at my computers for RAC with 2 different 750s
http://setiathome.berkeley.edu/show_host_detail.php?hostid=7454279
http://setiathome.berkeley.edu/show_host_detail.php?hostid=7446133

I run this command line

-sbs 256 -spike_fft_thresh 4096 -tune 1 64 1 4 -oclfft_tune_gr 256 -oclfft_tune_lr 16 -oclfft_tune_wg 256 -oclfft_tune_ls 512 -oclfft_tune_bn 64 -oclfft_tune_cw 64 -hp