jesus, i've only just gone from 2 cores to 4 (775 ftw)
jesus, i've only just gone from 2 cores to 4 (775 ftw)
I think that's the point that Stanford are making. BIGADV is not for the home desktop market - too many incomplete pieces of data by people cancelling workunits in order to get higher scoring ones which leads to delays in getting rafts of data processed.
or if you are a conspiracy theorist like me, then Intels dominance is taking away sales of GPU's which is hurting NVidia's profit a little....
Either way, those of us who've invested in SandyBridge feel like we have just received a giant doo doo from the dove from above.
so much for Stanford's PR dept. Anyone would have thought that we were doing this for free...
Join the HEXUS Folding @ home team
Thats not good for them but I do see their point. Even so, feel for those that invested. I guess we would expect to see quite a drop in the number of machines running bigadv units now.
I don;t know of any running for Hexus. Looking over on Bit-tech users are still picking them up, but with periods of inactivity between processing while units become available.
I've turned off one of my machines, hence my points drop the past couple of days. I'm going to have a look at working on my HTPC tonight - depends how I feel when I get home. My motivation had plummeted since we lost BIGADV after spending my birthday money and then some on another rig rather than using my S775/Q6600, which is now being sold off....
Join the HEXUS Folding @ home team
I've asked exactly that question before, and a very kind soul responded with:
"That can be done with BCCD, Clusterknoppix, Quantian, Pelican HPC and Parallel Knoppix."
Iv'e not got any further details, but I was wondering about having a virtual machine on each of my 4x 8 core (well 4 cores + 4 threads) running this OS. If I were to only use 4 cores on each machine, then they'd still be only half utilised, leaving them free to use, and still chugging away through BIGADV or better still BIGBETA work units.
Great theory, but I bet it's as messy as hell.
Join the HEXUS Folding @ home team
I'd be interested. Do you mean "this could be done with any of the above" or "this can be done with all of the above combined". If so, messy, as you said.
I'm interested in running a dedicated rig to be more consistent (look at my production!!), but with new 16 core requirements i think i'll wait
It looks like they are all different types of Linux cluster OS builds - Folding works on Linux, so I'm guessing the former. I'll tell you when I've had a chance to play.
Join the HEXUS Folding @ home team
do the 'nodes' in the cluster have to be the same spec? could i pull old core2 rigs together to make a 16 'core' cluster machine... seeing as they'd all be physical cores, i'd say the CPU performance would be comparable to a 8/8 sandbridge config
EDIT: and overclocking?
EDIT 2: and having that kind of modularity would be advantageous in terms of future proof, because it would be relatively easy to upgrade a 16 core cluster to a 32 core cluster when the time comes for BIGADV etc. to move to 32 core.
....again, I've no idea. You know as much about this as me I'm afraid. Is there a more appropriate place to ask here on Hexus related to Cloud computing?
These are all questions that I'd like to investigate. Realistically it'll be Xmas before I'll get chance to look at this, but if anyone else wants to start a thread and get the ball rolling....?
I reckon that this could work if it can run in a virtualised environment on each machine, otherwise it means having all the machines dual boot into some sort of Cloud based OS.
I've got a gigabit switch which I need to swap out my HP managed switch, and get Gigabit networking... working. I seem to remember there being problems going above 10/100 for some reason... probably more to do with the fact that I was doing computer to computer and not using a crossover.
Join the HEXUS Folding @ home team
There are currently 1 users browsing this thread. (0 members and 1 guests)