Page 5 of 5

Re: Need help selecting a Processor

Posted: Wed Feb 17, 2021 1:01 pm
by Neil-B
FaH uses minimal bandwidth so issues with streaming probably less due to network load but the fact that FaH is maxing out your CPU and causing issues that way ... You would need to be on a really slow connection to have issues with network load contention between machines used for FaH ... FaH only uses bandwidth during download and upload of WUs/Cores - If your streaming issues are happening most of the time and not just when thelog shows upload or download is happening then it isn't FaH overloading your network bandwidth

Re: Need help selecting a Processor

Posted: Thu Feb 18, 2021 11:33 am
by Jupis_folder
inOr wrote:So, if you want more power than a one-processor machine then a GPU is the best way to go?
Yes. Just compare these lists:
https://folding.lar.systems/gpu_ppd/overall_ranks
and
https://folding.lar.systems/cpu_ppd/overall_ranks

Re: Need help selecting a Processor

Posted: Fri Feb 19, 2021 4:38 pm
by v00d00
inOr wrote:Besides, I couldn't stream "Tenet" on my notebook machine without experiencing interruptions when the F@H was running a simulation. I have only one cable modem. I had to stop F@H before I could stream without interruption. How am I going to use 4 machines dedicated to F@H at one time? Won't they bog each other down?
More than likely graphics lag and not related to network issues. Sometimes you get lag while running gpu folding. I've seen it a number of times on PC's over the years that fold. Thats why I use onboard graphics on my systems for day to day usage and keep the Nvidia GPU's for dedicated folding (ie no montors connected to them). On a notebook it probably doesnt work that way unless you have an Optimus enabled notebook, in which case, switch the primary gpu to the onboard Intel card and leave the Nvidia card to fold.

Re: Need help selecting a Processor

Posted: Wed Feb 24, 2021 8:47 pm
by bruce
While the interruptions are potentially bad, the fact is that network speed is much, much slower that RAM to RAM data transfers so the cluster is terribly inefficient over the time required to complete a WU.

Adding the time to transfer a WU across your LAN once per every step makes it orders of magnitude slower. It all depends on your perspective of what is "fast" and what is "slow."