• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

AMD folding

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

flounder43

Senior Lawyer-clocker
Joined
Sep 20, 2001
Location
Minneapolis
Is there still a role for an AMD person to fold?

There was a time when my farm was top 20. I used to have many more machines up and folding and the numbers have since dwindled, in part because I felt that the Folding people (Stanford) were making it tough on me to keep up. They started with the Gromacs and the requiring of "flags" to fold fast and keep up. The "Force flags" here and there, I just don't think it should be required.

I have upgraded some of the computers on my old farm to some XP's, and still, I am supposed to use some "flags" to fold?

I am offering the use of my cpu, to it fullest extent, but they seem to prefer that it be an *Intel*, seems to me... I feel that they just don't even care if I *totally* leave.
 
Last edited:
I think the flags are used to save hassle. For example, -advmethods which is required even on Intel based PCs to get the maximum output, allows you to get workunits that arent reckoned to be 100% ready for the masses yet. -forceSSE is a flag because if SSE were enabled by default there would be a lot of crashed PCs, frustrated folders, and people would leave the project.

The flags can add some instability, which is why they are flags. If you set up the shortcuts right then you just need to fire up the shortcut with the flags added in for it to work to its full potential, you dont need to type them manually every time.
 
How much of an increase would I see with the right flags on with a 2.3Ghz Barton and Windows XP? Does it really help that much?
 
As a devout AMD folder, I still believe that there's a role for AMD in the folding project. If I remember right, isn't the SSE issue with folding a result of the programming of the core or client? IF that's the case, then it should be relatively easy to resolve and if that happens, then it's smooth sailing. I like AMD, I use only AMD and I'm willing to go through the testing process to see if each new cpu I get can run with SSE enabled or not. I don't think Stanford really cares what cpu we use to fold as long as the work gets done.

Quailine, if you're not using any flags right now, your production could double if you used -advmethods and -forceSSE with the 4.0 client.
 
First off: Greetings Flounder, you old time folder;)

I agree with David on this. The client has to be geared toward the mass of people that just start folding and forget about it(like google folders). They are not worried about or follow stats. They just don't want there pc's to lock up because they don't have the first clue about troubleshooting. These are the people that buy a Dell or HP plug it in and expect it to work period.

The flags and beta clients are for us devoted people that are willing to deal with some problems in order to get the most out of our pc's. We are willing to spend some more time(or some of us just don't have a life;)) on the project and its ins and outs so we get some reward in the way of increased production.

I have 5 rigs currently folding that are intel. They range from a 400 celly to a 1Ghz machine. Most of my points come from my AMD rigs.
They range from a 1333 tbird to a XP2200.
Just my outlook on things, but I'm gaining on the 33rd position:D
 
Flounder, if I lived close to you I would gladly go over and set up the flags for you on your machines just to keep you from quitting the project.:)

Like other people have stated here, the flags are for the more dedicated folding people who want to maximize their efficiency for the project. There have been some issues with the client and SSE on some Tbred and Barton procs, so Stanford had to come up with a way to keep those rigs folding and at least using the 3DNow! assembly loops, which don't give problems. The way they did this with the FAH3 version was just a cobbled together workaround until they released the new version 4 of the client. With the new client, there are separate flags to use either 3DNow!(-forceasm or no flag) or SSE (-forcesse) on the AMD procs. I've also noticed a marked decrease in problem Gromacs WU's that lock up using SSE the last 2 months or so also. I don't know if Stanford has been reworking the cores or the clients for added stability with AMD rigs, but I have definitely noticed an improvement.

The greatest majority of my farm is still AMD and the use of the flags has definitely helped my folding production (and point production;) ) tremendously. I can crank out around 5000 points/week when using the -advmethods and either -forceasm or -forcesse flags on the clients with my little farm. With the new additions and upgrades to my farm since Christmas, I would expect to see my production climb closer to 6k points/week in the near future too(no borged machines, all my own machines).:)

It doesn't take very long to set up your shortcuts with the flags, Flounder, and it is definitely worth it as far as for maximizing your folding production.
 
Well, i do know how to set up the shortcuts and all, but I was really disappointed when I found my new 2500 Barton was using 3DNow! instead of SSE by default. The SSE, as I understand it, is a lot faster with the folding client.

On a P4, it would use the SSE by default, would it not?

If the Barton is defective, according to Stanford, maybe they should call AMD and tell them so. My guess, though, is that the Barton is fine and it is the Folding people who have the problem.

Anyway, i did set up the -forceSSE flag on the Barton, and it is working fine.
 
Last edited:
flounder43 said:
Well, i do know how to set up the shortcuts and all, but I was really disappointed when I found my new 2500 Barton was using 3DNow! instead of SSE by default. The SSE, as I understand it, is a lot faster with the folding client.

On a P4, it would use the SSE by default, would it not?

If the Barton is defective, according to Stanford, maybe they should call AMD and tell them so. My guess, though, is that the Barton is fine and it is the Folding people who have the problem.

Anyway, i did set up the -forceSSE flag on the Barton, and it is working fine.

AFAIK its to do with the Gromacs implementation of SSE or something, and Im sure stanford have informed AMD.
 
How would I get to run the flags evry time folding starts if I'm running F@H 4.0 as a service?
 
flounder43,

Try switching over to the newest 4.x client. It seems to run better on AMD than the 3.x clients did.
 
Back