For users with multiple hosts, not all of which are going to EVER be capable of running CUDA tasks (laptop with NVS 135 has 128MB free after video buffer is allocated), we need a way to specify whether to use a GPU (ATI or Nvidia) on a host by host basis not on a project basis. My other PC has an Nvidia GEForce 560 with 1GB of memory. It will run just about ANY GPU wu that anyone cares to imagine - tho not as fast as some other NVidia GPUs, it gets the job done, unlike the laptop with 128MB of memory, in which nothing will run.
So for Einstein, I wish to be able to specify which host to send Nvidia GPU tasks to. It seems right now, that ALL hosts run with GPUs or NO Hosts run with GPUs
Copyright © 2024 Einstein@Home. All rights reserved.
Some GPUs are NOT ready to host Einstein WUs
)
Create a different profile for your laptop then?
You have four different
)
You have four different venues available to you, Default, Home, School and Work, just move your Laptop to one of those venues, and create Computing Preferences and Project Preferences for that venue,
Or if your running Boinc 7, you can use the option to exclude the GPU from certain projects:
Client configuration
Claggy
RE: unlike the laptop with
)
Collatz and i think PrimeGrid apps work on GPUs with less than 128Mb Memory - I tried Collatz's app on my 128Mb 8400M GS, proved that it worked, then got them to lower their requirements,
Claggy