I tought that i would get some better performance with CPU because of the non-stop load. That is my misinterpretation then.
Quote:
Your daily stuff will differ from mine. I've never had troubles running three tasks on a GTX970, or two on a GTX750Ti, or both. Plus sometimes some CPU tasks with it. But then I don't do video, or heavy 3D graphics or such.
then i think i do something wrong on my 'main' system. What would the best app_config be for me then for this system? i dont want to use CPU task, only GPU (plus offcourse the core's it need for the gpu). I also see that some of my workunits are pending for 5 days for now, is that normal at E@H?
Configuration? Sorry, I can't help you much with that. Personally I like to keep settings at standard if I can. With so many possible changes to make it's already hard enough to keep track and app_config is something that I don't feel comfortable with. It's for very special settings and I don't think it's useful in this case.
As I already said you can set the GPU utilization factor in the project settings, eliminating one reason to use app_config. When I change GPUs or a new application is released, I usually start processing single tasks and then increase the number of parallel tasks until the overall output doesn't improve any more. If in doubt, run a task less. Don't try to squeeze out the last few percents. If my GPU load is in the mid nineties that's fine with me, 100% looks too much like overloaded.
Einstein, unlike the few projects from your list that I'm familiar with, requires a quorum of two to validate, meaning two results need to be returned. This can take a while, even weeks in some cases. Up to a week is normal.
Thanks for the info. I
)
Thanks for the info.
I tought that i would get some better performance with CPU because of the non-stop load. That is my misinterpretation then.
then i think i do something wrong on my 'main' system. What would the best app_config be for me then for this system? i dont want to use CPU task, only GPU (plus offcourse the core's it need for the gpu). I also see that some of my workunits are pending for 5 days for now, is that normal at E@H?
The easiest way to not use
)
The easiest way to not use the CPU is to select under the project preferences to not use the CPU.
Your pending list looks pretty normal to me.
Configuration? Sorry, I can't
)
Configuration? Sorry, I can't help you much with that. Personally I like to keep settings at standard if I can. With so many possible changes to make it's already hard enough to keep track and app_config is something that I don't feel comfortable with. It's for very special settings and I don't think it's useful in this case.
As I already said you can set the GPU utilization factor in the project settings, eliminating one reason to use app_config. When I change GPUs or a new application is released, I usually start processing single tasks and then increase the number of parallel tasks until the overall output doesn't improve any more. If in doubt, run a task less. Don't try to squeeze out the last few percents. If my GPU load is in the mid nineties that's fine with me, 100% looks too much like overloaded.
Einstein, unlike the few projects from your list that I'm familiar with, requires a quorum of two to validate, meaning two results need to be returned. This can take a while, even weeks in some cases. Up to a week is normal.