On WIN 10 I have a gpuLookupTable_v402.txt.
My GTX1080 Ti is not listed and is running with little load / in standard mode (stderr.txt).
Is there a table/values ​​that better utilize the GPU?

On WIN 10 I have a gpuLookupTable_v402.txt.
My GTX1080 Ti is not listed and is running with little load / in standard mode (stderr.txt).
Is there a table/values ​​that better utilize the GPU?

The default values should work well enough. The lookup table is for fine tuning and usually gives less than a 10% improvement.

If GPU utilization is less than 100%, you can try using the app_config.xml file to have it run more than 1 GPU task at a time. I have done that myself to raise my usage from 90% to 100%. Individual run times go up but you process 2 WUs at a time.

When you are running 2 tasks at a time how much CPU do you give to each task? Example 0.5

When you are running 2 tasks at a time how much CPU do you give to each task? Example 0.5

I think you are asking how to set the "cpu_usage" tag in the app_config.xml file.

This will vary depending on your cpu. I have mine set to .20. I got that by looking at what the cpu usage was for the gpu task when it was running a single task.

In case it helps, here are the contents of my app_config.xml file:

<app_config>
  <app>
    <name>GetDecics</name>
    <fraction_done_exact/>
    <gpu_versions>
      <gpu_usage>0.49</gpu_usage>
      <cpu_usage>0.20</cpu_usage>
    </gpu_versions>
  </app>
</app_config>

Thanks Eric, that was the information I was after. I tried it and I feel 8 minutes for one task when running 2 tasks is too long. When running 1 task I haven't seen runtime like that on my card. GPU percentage is 90% or above the majority of the time

Ok, thanks. with:
numBlocks = 9600.
threadsPerBlock = 32.
my GPU utilization is 99% and Power consumption around 175W.
With standard:
numBlocks = 1024.
threadsPerBlock = 32.
my GPU utilization is 98% and Power consumption around 160W = 10% less.

Mr Eric,

Is there a noobs guide for what you're talking about in this thread.

For years I've been just running a spare computer with whatever happens when you run BOINC by default. CPU only, no GPUs. Wondering if there's anything I can do to get more performance out of it.

Mr Eric,

Is there a noobs guide for what you're talking about in this thread.

For years I've been just running a spare computer with whatever happens when you run BOINC by default. CPU only, no GPUs. Wondering if there's anything I can do to get more performance out of it.

This link explains the app_config.xml file
https://boinc.berkeley.edu/wiki/Client_configuration

The app_config goes in the project directory. It gives extra control of the client. If not present, the default configuration is used, which in general is good enough for most users. I personally didn't need to use it until I had a much better GPU and wanted to run more than one task at a time.