• Some users have recently had their accounts hijacked. It seems that the now defunct EVGA forums might have compromised your password there and seems many are using the same PW here. We would suggest you UPDATE YOUR PASSWORD and TURN ON 2FA for your account here to further secure it. None of the compromised accounts had 2FA turned on.
    Once you have enabled 2FA, your account will be updated soon to show a badge, letting other members know that you use 2FA to protect your account. This should be beneficial for everyone that uses FSFT.

Is UD/F@H multi-threaded?

Darth Bagel

Limp Gawd
Joined
Feb 1, 2005
Messages
383
So I'm trying to get this thing to chew on UD calculations as fast as it can, and it's only running as a single-threaded application (meaning only one processor at a time, which is annoyingly slow). I'd ideally like to just go for UD, but if there's no way I can use at least 4 of the CPU's at a time I'd be willing to go to F@H.
 
Neither program is multithreaded, however you can run multiple copies and assign each to a CPU.

The guide for F@H is here and
the link to a UD MutiCPU program (UD Tweaker) can be found here.

 
It appears it might be possible, the closest thing I've found to an answer is this thread from the grid.org forums.

Edit: It appears to be XP only, with fast user switching enabled.

 
Umm CPU usage is at 6%. why in the hell is that not pegged at 100 across teh board ;)
 
Because I can't figure out how to get it to use all 8 processors, maybe? Best I've done is 26% with UD across 2 processors.
 
MustangWorld said:
Run 2 instances of UD, the other 2 F@H. Best of both worlds

This is probably actually going to be the best, most efficient route. UD doesn't seem to lend itself beyond 2 CPU's or Windows 2K, so firing up F@H on the rest might be your best route.

I'll take a crack and say a quad HT boxen, or maybe 4 dual cores.... if only it were the latter1 :D

 
Compaq Proliant 8500R.
8x550MHz P3 Xeons, 1MB cache
4GB ECC Registered PC100/133 (I can't remember)
3x9.1GB 10k RPM ultra3 SCSI drives in a RAID 5 array (should have been 4, but it turned out one of them was dead)

Yeehaw. If anybody wants to send me memory and a hard drive I I can get another dual-processor machine up and folding in a month or so. Actually, if somebody wanted to be really awesome they'd send me a network card that didn't suck. I bought an Intel Pro/100 from a friend of mine to throw in my ML530 and the machine started bluescreening like a bitch.
 
If you could please start up a fold-server to test my code with, I'd love you forever. I haven't had an 8-way machine around to test the multiple CPU code over 2 processors. Fold-server will boot it over network, and then start Folding@home (theoretically with 8 instances :D ) on it.

Heck, for a chance to play with that beast I'd send you a fold-server box. Just have to put one together...

 
unhappy_mage said:
If you could please start up a fold-server to test my code with, I'd love you forever. I haven't had an 8-way machine around to test the multiple CPU code over 2 processors. Fold-server will boot it over network, and then start Folding@home (theoretically with 8 instances :D ) on it.

Heck, for a chance to play with that beast I'd send you a fold-server box. Just have to put one together...


Honestly, you want a fold server box all set and ready to go? Need Oldbenwa's permision but......

Sorry Mage, I unpluged [H]OldMarty today. Ben powered the server down nicely. It's been 110 F in the farm lately, and ~380 PPD from 20 amps just ain't all that great. I did get AC installed today back in the shop, but I don't think the power usage is really justified when I got 70 ~2.5 Gzh mahcines cranking away. My power bill is,,,,ahhh,,,high.

Been thinking of a way to tell ya, maybe this could be a good thing?
 
Hey, I can't blame you. It's hot here too.

Just don't take down the other 70!

 
I know F@H is multi-threaded and all - but I think I've read about people running two clients on the same machine without a multi-threaded processor.

Is there really any reason to do this? I don't see any possible gains that could be made running 2 clients vs. 1.

 
unhappy_mage said:
If you could please start up a fold-server to test my code with, I'd love you forever. I haven't had an 8-way machine around to test the multiple CPU code over 2 processors. Fold-server will boot it over network, and then start Folding@home (theoretically with 8 instances :D ) on it.

Heck, for a chance to play with that beast I'd send you a fold-server box. Just have to put one together...

Well I'm not really looking to do anything too extreme until September, because I can't really get to the machines easily until then (they're up in my apartment at school). Drop me a PM sometime after Sept. 5th and I'll be happy to test things for you.
 
blakex said:
I know F@H is multi-threaded and all - but I think I've read about people running two clients on the same machine without a multi-threaded processor.

Is there really any reason to do this? I don't see any possible gains that could be made running 2 clients vs. 1.

You'd have to split the amount of cpu time each took at 50, so in essence any WU would take twice as long as it would if you were running one instance. The only time a faux-multithreaded processor sees a gain is with the HT P4's and even then its incremental at best... 10-15%.

I am of the opinion that unless you have multiple physical processors or cores, more than one WU probably isn't the best idea as it can zap memory bandwidth, along with any number of possible errors. Stanford also prefers that in the instance of HT, that you run one instance only, as they prefer to have the results sooner.

 
Likewise my Linux folding service installer should create up to 8 instances of F@H for you. http://overclockix.octeams.com/foldinstaller.tgz

I've only been able to test this on single and dual CPU's, and have confirmed that single HT P4's are treated as two cpu's. My code should be fine for 4 or 6 or 8, as long as I don't have any typos....
 
Back
Top