I need some help/instructions on how best to optimize my new rig.

I just purchased an i5 8600k with a GTX 1080ti, you should be able to see it on the list of my computers.
You will notice I have tried to run a few WUs with Lunatics so that information is available.

I would also like to know how I could stop CPU tasks on the laptop while allowing them on the desktop.

Also, how many WUs should I expect to run on the card at once and how, also if I should keep a core or two free for
processing/feeding he GPU.

Any xml info or cmd lines you feel would be best served would also be appreciated greatly.

I haven't had a desktop to run SETI for 7 years now... It's time to get dedicated again.

---

With the GTX 1080Ti, many people have found running more than 1 WU at a time gives better throughput, on my GTX 1070s I've found 1 at a time is just as good.
I'd suggest getting it to run 1 at a time with the best possible crunch times would be best, then give running multiple WUs a go.

I would suggest making sure you have the Lunatics Betav6 installer.
Make sure you select AVX for the CPU, and SoG for the Nvidia GPU. These applications will give the best output from your hardware.

After running the Lunatics installer and installing the SoG application, there will be a
mb_cmdline_win_x86_SSE3_OpenCL_NV_SoG.txt
file in the Seti project data directory.

If you put
-tt 1500 -hp -period_iterations_num 1 -high_perf -high_prec_timer -sbs 1024 -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
in there & save it, when the next WU starts processing it will use those values, and you should see a significant reduction in your crunching times.
If the screen lag/keyboard input lag becomes too great, increasing the value for -period_iterations_num to 3, 5, 7, 10 etc will reduce that lag.

Also in the Seti project data directory, if you create a file named
app_config.xml
and put the following in there

<app_config>
 <app>
  <name>setiathome_v8</name>
  <gpu_versions>
  <gpu_usage>1.00</gpu_usage>
  <cpu_usage>1.00</cpu_usage>
  </gpu_versions>
 </app>
 <app>
  <name>astropulse_v7</name>
  <gpu_versions>
  <gpu_usage>0.5</gpu_usage>
  <cpu_usage>1.0</cpu_usage>
  </gpu_versions>
 </app>
</app_config>

(you can leave out the astropulse section if you don't do AP work)
You need to click on the Manager Option, Read config files for those values to take effect (or exit and restart BOINC).

That will reserve 1 CPU core for each GPU WU being crunched, which will give a boost to GPU output. When the GPU runs out of work, those cores will be released to process CPU WUs until more GPU work comes along.


In your account page, under Seti@home preferences, it's possible to have 3 different sets of preference (called school, home and work).
Set up one for your main system, to process everything all the time, and use another location for the laptop. You can either disable CPU processing, or just reduce the number of cores, or the CPU load when processing CPU work.

---

Grant
Darwin NT

Thank you for the information.

I just changed to tthe SoG app.

Could you please explain to me what "-period_iterations_num" means?
Other wise I do not understand what what changing the value means.

---

Could you please explain to me what "-period_iterations_num" means?

My understanding is spotty, at best.
Basically the lower the number, the faster a WU can be processed. However, particularly on lower performing video cards, that can result in a significant impact on system responsiveness. Dropped characters when typing, stuttering in video playback, windows sticking when you move them around etc.
On my GTX 1070s I get some of those issues, but not enough to be annoying- hence I stick with the aggressive settings. The higher the value, the less the performance, but also the less any impact on system responsiveness.
On a low end card you might need a value of 50 (or higher) to avoid any user issues. But with highend hardware such as yours, even at a value of one I doubt there will be much impact. If there is, a value of 3-5 will probably be more than sufficient to make the system responsive enough to not be annoying.
The other value that you can vary to improve system responsiveness, that doesn't have as much of an impact on performance is the TT value.
Instead of 1500, you could set it 800 or so, and see if that improves things (while keeping -period_iterations_num at 1).

---

Grant
Darwin NT