Folding Forum

see psummary page http://fah-web.stanford.edu/psummary.html and found this list

Yes. It also takes 3 days to run on a dedicated 8-core box. If you have such hardware and are interested in helping with these work units, let me know.

This WU shouldn't be displayed on psummary ... also, the server doesn't appear on the server status page ...

These projects study how influenza virus recognizes and infects cells.

This is real important work with the viral climate in the world today. I am happy to get the regular Influenza WUs on a recent foray into VMware an ubuntu running SMP

Are the Linux machines the only ones getting these WUs?

Leoslocks wrote:Are the Linux machines the only ones getting these WUs?

For now, it's Linux and MacOS. It turns out that there are a fair number of high-powered Macs containing dual quads. I don't think there many machines of that class running Windows that could be dedicated to FAH, but if there are, I'm sure Kasson would be happy to hear about them. If the resource is out there, the Pande Group does a good job of figuring out how to use it.

How are these WUs being assigned ???

how long would this take on a mac mini?

You can't run it on a mac mini. It requires 8 or more cores. Sorry.

Those units look insane. According to the math, 265 points an hour. I'm going to guess a half gig of RAM per thread, maybe more. If they're pushing the hardware as hard as I think, it's worth the points :>

The only thing I can think of that would run these is stuff in a film/animation studio or hooked up to a CT/MRI scanner. The highest end Mac systems would run them.

Brian Redoutey wrote:Those units look insane.

Maybe, but science keeps asking questions that are harder and harder to answer.

call me dumb.. but can't these sorta work units be 'cut down to size' ? Like i would think it'd be better to have work units that take less than a day to process continually being returned rather than large work units which only increase the chance of it not being returned (errors during processing... power outages etc.). I know with the PS3 client i'd rather receive 4 x 250 point WUs than one of the 32hr 1,250 point WUs simply if i have a power outage.. my PPD goes way down.. but also, the WU has to be re-dealt and keeps delaying it's return.

Yes, but splitting up the WUs in to smaller chunks increases the work load at Stanford. First to split them, then the server load to send them out, and get them back, and then to reassemble. If you split that big WU in to 4 chunks, the load increases exponentially. And what happens if 1 of those 4 pieces (or all 4) goes to a slower computer? Or goes to a fast computer that is overclocked just a bit too much, and EUEs that same WU 3 times...? So it isn't always faster to send out more or smaller pieces.

Look at it this way... if the SMP client, which currently defaults to using 4 cores is better and faster at processing larger chunks of data than a single core CPU client, then a work unit which requires 8 cores probably improves the size and speed again.

Also consider that after folding this Really Big WU for 3 days, you get 25K+ points. What's your SMP client get in 3 days time?

smASHer88 wrote:call me dumb.. but can't these sorta work units be 'cut down to size' ?

I notice that this project does not show the number of atoms. I suspect that the number is very large, in comparison to the projects that we're used to running. There's no doubt that many interesting proteins are much larger than we've been processing and there's no way to fold half a protein. If I'm right about the number of atoms, the simple answer will be: No.

Time to buy some new hardware.

is there some available ?

i am interested to try

16 cores under linux

Wow. That is a NICE mobo.....and I'm pretty sure it's not consumer level. Which makes me want to ask, is it a Tyan? What northbridge are you running? The RAM slot positions are somewhat a give away you don't play Solitaire on something like that