Originally posted by: ShawnD1
Cool. I'll check this out.Originally posted by: taltamir
if you wanna do folding... or any DC, just use BOINC. berkeley open infrustructure for N-something computing.
it works like a charm, and they actually have two seperate protein folding projects:
http://boinc.bakerlab.org/rosetta/
http://boinc.fzk.de/poem/
it will handle more cores than you have, and can even do multiple GPUs, all from within the same interface. (you can select how many cores and GPUs it is allowed to use on each computer)
Alright I've been dicking around for about an hour and it's honestly not any better than F@H. F@H pisses me off because I need to configure it 4 times on a quad core, but each configuration is pretty easy. BOINC has a lot more settings to go through and I need to add myself to the Anandtech team for every single project I want to be involved in. Another issue is that Rosetta does not have any GPU client at this time. Actually, most of the BOINC projects don't support GPU. This thing doesn't even work on my Phenom, it just keeps saying "communication deferred" and it has yet to download any Rosetta work units.
I don't know why distributed computing needs to be so unintuitive and difficult. BOINC doesn't work, F@H's easy to install client only uses 1 core, and F@H's SMP client has deadlines that are almost impossible to meet (a Phenom X4 overclocked to 2600mhz never meets the deadlines). The only thing that seems to work ok is installing F@H's console client 4 times.
