Newcomer to Folding@Home

[Thoreau]

It's been ages since I've done anything in the distributed computing world and figured that now is as good a time as any to change that.

I used to crunch for the SETI team here but that died off for me for still unknown reasons.

Right now I'm only running FAH on one machine, but I'm anxious to see how well it can perform. The rig is a dual core Opteron 170 (each core is @ 2.0GHz) with a couple gigs of ram. The system is a dedicated server in Dallas that always seems to be damn near idle even with the rather hefty traffic and amount of hosting it does, but since it doesn't cost me anything extra to heat those little cores up and suck up a little more power, I figured why not =)

Just curious tho, approximately how long should such a rig take to crunch work units? Or is work even measured in that method (old SETI mindset is coming back.) It's running two instances, one on each core, so I would hope that it can pound out the work, but who knows. OS is Redhat Enterprise 4 and the client version is the latest, 5.04 beta. I didn't even realize that I had picked a beta version until after it was already running, but if it is known to perform less than 5.02, I can always switch to that version.

Anywho, expect to start seeing work units coming in from ThoreauAZ for team 198 =)

Edit: Now also running on an A64 3200 (s754 variety) w/2GB RAM, WinXP.

 

[Thoreau]

On another note...

sda5: write failed, user block limit reached.

The above seems to be popping up every once in a while on both FAH instances. Any thoughts?

 

[trevinom]

Welcome to F@H.
Nice rigs you're running.

Work Units come in a couple of flavors, you have the Gromacs and the Tinkers. Normally, they are cut up into chunks of 100 pieces, when you finish all the pieces it sends the results back to the servers and you get credits. On my AMD 2100+ rigs with 512MB of rams, it can take anywhere from 1 day to a couple of days for the 200 or point WU. But they come in so many different point values, from 20 to 600, that it becomes an incosistent way of measuring. F@H crunchers have resorted to PPD (Points Per Day) as a measuring stick. Again, my AMD 2100+ with 512MB of ram averages about 100 PPD.

You're server should be pumping out upwards of 200 PPD easily on each core, I would think.

As far as the error message, I've never seen it before.

I"ll let some of the more experienced crunchers answer that PPD and error message more fully.
Just wanted to say 'Welcome to F@H'

 

[Spacehead]

Welcome to the F@H TeAm Thoreau. I remember that name from the SETI days

You might be able to get a little info on PPD from this thread.
It's hard to compare different WUs so unless you're working on one that's already listed it's kinda tough.
I think there are other comparison pages somewhere, but i don't have the links.

The 5.04 beta client isn't beta anymore, they just never changed the name, so it's fine to run it.

 

[GLeeM]

Originally posted by: Thoreau
On another note...

sda5: write failed, user block limit reached.

The above seems to be popping up every once in a while on both FAH instances. Any thoughts?

Welcome back to DC, Thoreau

Your computers should do awesome on F@H!

I've never seen that error on Windows, it must be a Linux thing.

? Maybe there is a limit on how many blocks can be written (to memory?) by a limited user?

I checked Windows Task Manager on my dual Xeon system and there are quite a few IO Writes made by the F@H core. For comparison though, it is a very small fraction of that made by the MalariaControl program

I'm not sure but I would guess that the F@H client will not finish a WU if it has a "write fail" error. Does it get a new WU after a "write fail"? (Check the FAHlog.txt file for this info.)

I think we need a Linux expert. There are a few on the TeAm. So be patient, someone will be along to help, if not I will add this to the stats thread

 

[Thoreau]

I actually just went ahead with what is probably a giant no-no and re-ran the FaH executables under root and bam, no errors =)

The server uses Plesk (ie: server is mainly for web hosting) so quotas for user accounts are all managed thru Plesk itself. I think I'll try just manually creating a user account to run F@H under and hope that Plesk doesn't somehow get it's grip on it and try to enforce any quotas.

Of course, this is all assuming that it is indeed quota related as I am hoping it is.

Edit: PS - the home PC (a64 3200) spit out it's first WU today, 53 points. Still waiting for the dual core opty to do so. Both units that it is crunching are in the 30-37% range for completion, so hopefully soon =)

 

[Madwand1]

My main desktop is an X2 3800 (which used to be OC'd to 2.4 GHz for a long time, but is now back down to 2.0 GHz for SATA reasons). It runs 2 F@H instances all the time at 100% low priority, and browsing the logs casually I see that single WU's take between 12-24 hrs. There's a lot of variability. Around 24 hours is a good time to check on your progress.

 

[yelo333]

Originally posted by: Thoreau
I actually just went ahead with what is probably a giant no-no and re-ran the FaH executables under root and bam, no errors =)

The server uses Plesk (ie: server is mainly for web hosting) so quotas for user accounts are all managed thru Plesk itself. I think I'll try just manually creating a user account to run F@H under and hope that Plesk doesn't somehow get it's grip on it and try to enforce any quotas.

Of course, this is all assuming that it is indeed quota related as I am hoping it is.

Edit: PS - the home PC (a64 3200) spit out it's first WU today, 53 points. Still waiting for the dual core opty to do so. Both units that it is crunching are in the 30-37% range for completion, so hopefully soon =)

Sure does sound like a quota issue to me... Does df -h report an acceptable amount of free space? Ext2 & Ext3 are configured by default to reserve 5% of space for root, though this will show up as 105% under df -h. That /could/ be why you are getting the error... I've never heard of/used Plesk, though.