The ***OFFICIAL*** F@H Harassment and discussion thread!!!

[Sparky]

Agreed on both counts. Not a good idea to go hijacking school computers.

I wish I was admin of the lab here at work - There are about 30 P4's in there that sit idle or used for web/MS word, imagine the numbers I could add up with all those. I can't do anything about it though

[Drunner611]

Yeah, i get your point. I was just curious, since I have access to probably over 500 p4's here. No admin though. Darn.

1 more question. Since I'm running dual instances, would it be wise to set core affinity's seperate? 1 instance on core 0 and the other on core 1? Or just let the 2 cores work together?

[angra]

Drunner, it's still a good idea - it just needs to be executed properly. I bet you could find the right teacher or professor or whatever to ask permission from.

regarding affinity, I set each instance on my dual processor windows machines to a separate affinity, but on my linux boxes, there is not an easy way to set affinity (unless the client does it itself?), and I've never noticed a significant difference in performance between the two.

[Drunner611]

Alright, I set them apart, so we'll see. Are they ever gonna develop an Nvidia gpu client?

[jimwah]

Someone mentioned to me in PM that setting affinity had improved frame time's on a PentiumD in windows (in the region of 10-15% iirc). I tried it on my C2D & it had absolutely no effect, I never had time to test it on my 805 and that machine is in pieces now Worth setting it just in case I suppose, certainly doesn't do any harm.

[angra]

I've thought about doing that Drunner . I do work with GPU enhanced scientific/signal processing about 15-25% of my work-time. I would love to have the spare time to look at porting one or two of the calculation kernels over to Nvidia. The ATI and nvidia architectures are organized differently enough internally that optimization choices for each are very different, so it's probably not trivial for Stanford to target both.

[embeejay]

as far as i can gather from the stanford folding forum, using nvidia is not an option for several reasons - the hardware isn't very wellmade for it (tho the new g80 might be) adn the drivers is another big problem since it doesn't return the "right" results in many cases - i know they have used quite a bit of time on it, but have given up since it was a lost cause.

Try to browse the stanford folding forum to find out the details there are several threads on the subject.

[Sparky]

WAKE UP CALL TO JIMWAH!!

3 days soon to be 4 days in a row with NO WU's, and NO points? Hello, wake up!

[jimwah]

I'm awake already! Just helping out our friends, & paying back some WU, at EOCF for the rest of January. I'll be back soon, I've got to crack 400K and catch Mr Cole!

[Sparky]

OK just checking I'm helping them out too - I have fold6 and fold7 on there, with another 3200+ going on soon - and I'm still putting some points on the F@H charts. I'm guessing you went all out to WCG?

[embeejay]

I must say i seriously hope the F@H slump is due to people helping out WCG and not a tell tale sign of the team dying - we have blues on the top 20 these days - it hasn't been that bad since i joined - i haven't even seen greens in top 20 for a while, and now we have blues.... *sigh*

Helping WCG is noble (and a good thing) i ran a client for them for a while as well, just don't forget to come back or TPR will overtake us again and that would just be, ehm, no i can't even finish that thought :P

[jimwah]

No I'm over here : EOC F@H Stats with a bit of luck I might hit 75k before the end of the month How's the WCG mission going man? I've just noticed I've dropped below you on the dailies, we can't have this

[Sparky]

Oh for some reason I thought you meant the WCG team I'm helping out the WCG'ers here for a spell. OK so you help out the EOC F@H team, I'll do the WCG Pretty much they are in a long race, trying to get to first place before another team does (easynews I think) but they are lagging behind by a lot of PPD. I may move more than I have over to them in a little while, we'll see.

@ embeejay, I'm guessing our output is lower because of many going elsewhere to help out as thank-yous. And our team really isn't dying, our 24 hour average is still a good 10k higher than it was before the race with TPR So once everyone is back on F@H I'm sure our average will stay up Only thing is the team "The Outcasts" is due to pass us in 3 months

[Sparky]

I was checking our EOC stats page and saw this newegg ad at the top... I think somebody needs the check their spelling!

[[XC] moddolicous]

Has anyone ever gotten some project 26xx? I got one on my A-XP 2200 and it is taking 6 1/2 days . They are taking forever.

[Sparky]

Has anyone ever gotten some project 26xx? I got one on my A-XP 2200 and it is taking 6 1/2 days . They are taking forever.

I've gotten a couple. My parents' PC (A64 2800+ and 512MB RAM) has a 2606 right now, it is a 292 pointer. The PPD for it is 222.28, so just slightly over a day's worth of time for this one.

[[XC] 4X4N]

I read somewhere about f@h using boinc in the future. Any news on this? If they do, I might come back and put a rig on this permanently. Boinc sucks in some ways, but it is a lot easier for me to manage on multiple computers.

[Sparky]

I read somewhere about f@h using boinc in the future. Any news on this? If they do, I might come back and put a rig on this permanently. Boinc sucks in some ways, but it is a lot easier for me to manage on multiple computers.

I haven't heard anything about that, but then I'm not active at the official F@H forums. You might want to check this folding manager out though, I haven't had a chance to mess with it yet but it looks like it simplifies things quite a bit by using a GUI to configure the console

[texnofobix]

i'd like to see a clustered client, like distcc works.

[Sparky]

i'd like to see a clustered client, like distcc works.

Any way to run something to cluster systems, then have one main system that you could put the F@H client on? Basically so the client sees just 1 CPU but that "1" CPU is the cluster of several? Hmm, I wonder how much faster that would be....

[Marvin_The_Martian]

I haven't heard anything about that, but then I'm not active at the official F@H forums. You might want to check this folding manager out though, I haven't had a chance to mess with it yet but it looks like it simplifies things quite a bit by using a GUI to configure the console

Hehe thanks for the commerial but the link is still down. Working on this alone is hard, especially with people coming in with suggestions which involve rewriting and especially rethinking allot off things. Btw, for people who just want a GUI to configure the console I believe their are some tools in this list which could be suitable. Especially this

Haven't treid it myself but the description looks rather good, something like my setup procedure

[davidletterboyz]

Someone mentioned to me in PM that setting affinity had improved frame time's on a PentiumD in windows (in the region of 10-15% iirc). I tried it on my C2D & it had absolutely no effect, I never had time to test it on my 805 and that machine is in pieces now Worth setting it just in case I suppose, certainly doesn't do any harm.

That's me PMing you LOL :P
I think it works on PentiumD bcos it's not a native dual cores CPU. I noticed that CPU load will go like 50-50 if i'm running one F@H. That means, the bus is pretty much cramped for bandwidth.
I have just upgraded to X2...manually setting affinity improved a few seconds LOL.

[jimwah]

Hey davidletterboyz, interesting theory - certainly the improvement in frame times you showed me proved it's worth setting imo

[GAM]

Here you all are!?!

I must live under a rock
(I get lost in my subscriptions and forget to check the boards )

[Sparky]

Here you all are!?!

I must live under a rock
(I get lost in my subscriptions and forget to check the boards )

yup, found us!