Folding percentage used ceiling control.

Feature Name:
Folding percentage used ceiling control.

What does this feature do:
A user would have the option to input a percentage value such as 90% and the CPU and/or GPU would then throttle itself to not go above that percentage used while folding. Folding would not disrupt or slow down the end user applications like in an office environment.

Why would this feature be useful to be included in the tracker:
Folding would not disrupt or slow down the end user applications like in an office environment. The current automated client method of this feature offered by Stanford leaves much to be desired. An example is when Microsoft Word 2007 is started it responds sluggishly while a Work Unit is also being processed. If 90% of a CPU and/or GPU is dedicated to Folding then at least 10% of the CPU and/or GPU could be used for other tasks.

brcromer wrote:Feature Name:
Folding percentage used ceiling control.

What does this feature do:
A user would have the option to input a percentage value such as 90% and the CPU and/or GPU would then throttle itself to not go above that percentage used while folding. Folding would not disrupt or slow down the end user applications like in an office environment.

Why would this feature be useful to be included in the tracker:
Folding would not disrupt or slow down the end user applications like in an office environment. The current automated client method of this feature offered by Stanford leaves much to be desired. An example is when Microsoft Word 2007 is started it responds sluggishly while a Work Unit is also being processed. If 90% of a CPU and/or GPU is dedicated to Folding then at least 10% of the CPU and/or GPU could be used for other tasks.

There is actually no way to "throttle" GPUs to only use some percentage of the resources. The GPU can't be doing compute and graphics at the same time. Even if you write a kernel that uses only uses a small portion of the GPU the rest can't be used to render the desktop environment at the same time. The second best option is to reduce the kernel execution time, which allows the desktop to update more often, but that can't be done with 3rd party software.

For multicore CPUs leaving 1 core free (for example using 7 cores on an i7 which has 8 logical CPUs) combined with the idle priority should solve any lag issues. Limiting CPU usage is actually possible though, unlike with GPUs. However, considering that desktop lag can easily be eliminated with the current settings this would be redundant.

End result: never going to happen.