Pro/Forums - WTH happened?

Pro/Forums (http://forums.procooling.com/vbb/index.php)

- Pro/Crunching - Distributed Computing Teams (http://forums.procooling.com/vbb/forumdisplay.php?f=48)

- - WTH happened? (http://forums.procooling.com/vbb/showthread.php?t=7267)

I've had my system connected pretty steady.

But for the last two days my system hangs on sticking points in the folds time after time.

I havn't contributed S#!T for the last two days!

I saw a post by JayDee that this could happen on occasion. But so many times for two days straight? I've done reboots, cleaned up junk files to maximize my RAM, nothing seems to have helped.

One last move, I bought a second 256 RAM stick tonight. I'm going to install it tomorrow while also cleaning the inside of case ect.

Any advice on anything else I should do to optimize?

My 3 computers have been putting out 30,000 or less a day. That's horrible. But I think once I get them all up to the 250 folding region, it'll go faster.

All you can do is sit it out. It sucks hardcore. I had one comp stuck for 12hrs once. The client will only use a maximum of 150megs of RAM so that shouldn't be to much of an issue. The best thing to do is get as much mhz out of your system as possible. The good news is there is a thing called the laxness levels. If you get stuck the lax levels rise and the next steps go by a lot faster.

Bah, better then my half ass explaination here is a quote from the Phase 2 FAQ

Quote:

My protein keeps getting stuck. Is my computer broken?
The new software uses a different approach to compute protein structures from phase I, (except for generation 0). Structure are computed by perturbing ever so slightly the best structure from the previous generation. Due to this restriction, there are a lot less places the protein can 'go' at each step and it is a lot more likely to bump into other atoms while building.

To counter this, the algorithm will 'give up' after a certain number of attempts (up to several hundred thousand in some cases) and start building the structure from the beginning again, in the hopes of avoiding the sticky point. If it continually gets stuck however, it will slowly increase the structure laxness levels. A given stuck structure will give up and restart folding after making 25N(log10 N)*CUBRT(gen)*SQRT(R) tries to place a residue, where N is the number of residues in the protein, gen the current generation number (or 1 for generation zero), and R is the residue number currently being placed. On a fast (2 GHz) computer this equates to 2-3 minutes real-time for a typical protein.

The structure laxness levels are three numbers which affect the quality of structures generated - basically how serious of an atomic clash is allowed and so on. The higher these get, the poorer the quality of the structure that gets built (in terms of geometry) but the less likely it will get stuck. Hence if you just leave it alone when stuck, it will eventually raise these laxness levels high enough to get unstuck. The levels are raised very slowly (for obvious reasons), however, the good news is, once raised, they maintain their levels so that you are not likely to get stuck again for nearly as long as the very first time it gets stuck. At the end of a generation the laxness levels are just slightly decreased, but will increase again if it gets stuck for awhile.

Thus getting stuck with this new algorithm is perfectly normal and to be expected, and although it may appear your computer is doing nothing, it is actually trying literally millions of placements trying to get all the atoms to fit together. Although it may still for several hours on one generation, once it gets through it, the others are not likely to take nearly as long, so do not get discouraged. We encourage everyone to let the client run and get up to generation 250, as that is where all the scientifically interesting stuff will begin to happen.

Blade cuts both ways... a couple of daya ago I had 12,000 in one go. *shrugs* considering I'm online all the time that was unsual.

Well then I guess that's just the way it is.

Agree with the start of JayDee's last post.

Sux

But at least a good excuse to add some RAM.

Quote:

Originally posted by Blackeagle

But at least a good excuse to add some RAM.

Hell yeah, and then another computer, and then another computer, and then payoff the credit card, then upgrade the computers, then buy another computer, then payoff credit card.... :D:D I am in the payoff credit cycle now. Most of it was truck parts this round though and not computer parts. :)

Picked up the parts to hook up the second puter.

Will be going down for "adjustments" soon.

Well I installed the new RAM and the nec cards in both systems and hooked both to the hub.

No joy.

Mine keeps saying the set up is complete but with errors--3X:mad: No clue what I'm doing wrong here. I follow all the steps without a hitch, but at the end it's always compete but with errors. Suggestions?

Other one doesn't seem convincable that a network card is there. Will check that one for proper and tight installation of the NEC card.

Going to just call it a night, have to work tomorrow. Will see if it loooks like the RAM sped things up at all when I get home and check my folding stats.

Good night.

Quote:

Originally posted by Blackeagle
Well I installed the new RAM and the nec cards in both systems and hooked both to the hub.

No joy.

Mine keeps saying the set up is complete but with errors--3X:mad: No clue what I'm doing wrong here. I follow all the steps without a hitch, but at the end it's always compete but with errors. Suggestions?

Other one doesn't seem convincable that a network card is there. Will check that one for proper and tight installation of the NEC card.

Going to just call it a night, have to work tomorrow. Will see if it loooks like the RAM sped things up at all when I get home and check my folding stats.

Good night.

Is it the NIC setup that is giving the errors or???

Need to know a bit more Info to be of any real help.

First thought though is a IRQ conflict.
If you have another open PCI slot try moving the Nic card to that slot. I have seen it happen many times with winblows 98 and 98se. Looking at device manager may also give you a clue ( look for the yellow question mark or exclamation point ). I have also seen times where ( according to device manager ) there is no conflict but the PC was slow accessing the internet, chaging PCI slots solved the problem.
You can also check in bios to see what IRQ the cards are assigned.
In most systems (Pre Windblows XP ) IRQ 11 should be reserved for Video card) then look at sound and the Nic, those 3 things should be on 3 different IRQ's.

Hope that helps
Aardil