Team CFC: Folding@Home project

[Genocidicbunny]

That this is alive and going so well makes me happy!

What nickname are you folding under?

 

[IglooDame]

Well, before trying the VM, I went and installed more stable drivers, and what do you know, Im up to almost 10,000ppd.

Stable drivers for the GPU card? And what is your CPU folding on the Q6600 churning for PPD now? (I'm trying to verify that mine is optimized.)

That this is alive and going so well makes me happy!

What nickname are you folding under?

Nicknames. Anyway, Abaddon gets credit for organizing TeamCFC in the first place, so we give him a pass on his current lack of points.

 

[Genocidicbunny]

Stable drivers for the GPU card? And what is your CPU folding on the Q6600 churning for PPD now? (I'm trying to verify that mine is optimized.)

Yep, more stable drivers for the GPU. I was using 185.20 ones, but now im down to 182. something, the latest full release ones.
As for ppd on the cpu, according to FahMon, on my 2.83ghz Q9550 im getting around 1800ppd.

 

[Genocidicbunny]

Just got the SMP VM's up, took a while to get the network stuff figured out.
============================
See the following guides:
http://www.maximumpc.com/forums/vie...&start=0&sid=80c9284a61dd8c920f2d7308c87f7690
http://www.hardforum.com/showthread.php?t=1410289

Mind you, I need to tweak the priorities, because as of right now, those VM's are making things a lot slower than they should be.

EDIT:

Its up around 4000ppd just from the VM's. The CPU is putting out nearly as much pdd as the GPU now.

 

[Izipo]

Quick question : I'm leaving in a few hours for a week, and was wondering if I could let my computer run for a whole week without supervision. It is very well cooled (as there are more than 7 fans), but I'm unsure about leaving my GPU on full blast for a whole week (GTX 260).

Any thoughts ?

 

[Genocidicbunny]

Does it run fine when you let it stay on overnight? Do you need to reboot often?
I suggest you set up a vnc so you can take care of the occasional glitches. Personally, if you discount the reboots I had to do to set up vmware player, my machine has been up and folding nearly 24/7 for the last 2 weeks with no problems. If it doesnt overheat on its own, then you should be fine.
YMMV, and something may go wrong, but if it does, it would have gone wrong either way.

I say yes, let it run. Otheriwse you lose a lot of points (7 days == 55000 or so points for you and I)

EDIT -- not more than an hour after this post I had a crash that corrupted both the VM's. So now I lost 12 hours of work on both of those. >.<

 

[plarq]

You run FAH on Virtual Machines?

 

[Genocidicbunny]

See my links. You can, and it can make quite a difference.

 

[Izipo]

Well, it seems that my computer crashed just a few hours after I left, since I've had 0 points for the past 9 hours...

I just hope it didn't set fire to my house.

 

[Genocidicbunny]

Hmm, yeah. Seems like it probably did.
I really should try to restart the SMP VM's, the only problem is they're gonna lose all of the work. I keep getting a out of bounds error when running them, so it looks like my vmdk files became corrupted.
This is fair warning to anyone who wants to set up a SMP VM, they're not 100% stable and its not a good idea if your pc is prone to crashing.

 

[plarq]

Why set up VM for folding, you have FAH for almost every platform you see on your PC.

 

[Genocidicbunny]

Why set up VM for folding, you have FAH for almost every platform you see on your PC.

The SMP client seems to be a bit inefficient, its not using all of the CPU that it can be, as obvious by the screenshot of FahMon. Setting up the SMP VM's basically doubles its production. Some people have managed to squeeze out triple production too. 1800ppd/cpu on a hyperthreaded Core i7 is amazing points, considering that without it it makes like 2000ppd.

It is a helluva lot more work than just running the SMP service though. Im pondering scrapping that idea and instead adding a second GTX 260+ next month. That will effectively give me another 7000ppd. Maybe even two if I feel like splurging and I find a good deal. Only problem with that plan though is my PSU. Here's hoping a 750TX can handle 3 GTX 260's

 

[plarq]

If SMP is inefficient, run several uniprocessor cores then.
Well, last time I check, it takes 90% of total CPU.

BTW, this is Linux.

 

[Izipo]

Wow, three GTX260s...

Would you mind handing out a spare one ? I've been dying to try that SLI thing.

 

[IglooDame]

The SMP client seems to be a bit inefficient, its not using all of the CPU that it can be, as obvious by the screenshot of FahMon. Setting up the SMP VM's basically doubles its production. Some people have managed to squeeze out triple production too. 1800ppd/cpu on a hyperthreaded Core i7 is amazing points, considering that without it it makes like 2000ppd.

It is a helluva lot more work than just running the SMP service though. Im pondering scrapping that idea and instead adding a second GTX 260+ next month. That will effectively give me another 7000ppd. Maybe even two if I feel like splurging and I find a good deal. Only problem with that plan though is my PSU. Here's hoping a 750TX can handle 3 GTX 260's

Hmmm... escalation...

I'm doing some pondering myself. I'm not sure I'm ready to drop out of third place yet.

 

[Genocidicbunny]

If SMP is inefficient, run several uniprocessor cores then.
Well, last time I check, it takes 90% of total CPU.

BTW, this is Linux.

Uniprocessor clients are even worse. Ive tried. The fact that its using "100% CPU" doesnt actually mean its using all of the cycles. Having the VM's run forces it to do that pretty much. Its why when I run the normal SMP client, at 100% CPU the core temps are around 62C. With the SMP VM's, at 100% CPU the core temps are closer to 70. The VM's push the CPU just about as far as it will go.

 

[Genocidicbunny]

Hmmm... escalation...

I'm doing some pondering myself. I'm not sure I'm ready to drop out of third place yet.

Well the question now is if you have bigger pockets. Im not gonna spend more than 300$ on upgrades, so if you've got more than that, you're golden. Plus losing your 3rd place is like a year away.

 

[plarq]

Uniprocessor clients are even worse. Ive tried. The fact that its using "100% CPU" doesnt actually mean its using all of the cycles. Having the VM's run forces it to do that pretty much. Its why when I run the normal SMP client, at 100% CPU the core temps are around 62C. With the SMP VM's, at 100% CPU the core temps are closer to 70. The VM's push the CPU just about as far as it will go.

Maybe it is just VM software takes additional CPU cycles.

 

[Genocidicbunny]

Maybe it is just VM software takes additional CPU cycles.

The pdd is irrefutable though. Its currently at 2000ppd with the regular client, with the SMP VM's it was also 2000ppd per VM

 

[IglooDame]

Well the question now is if you have bigger pockets. Im not gonna spend more than 300$ on upgrades, so if you've got more than that, you're golden. Plus losing your 3rd place is like a year away.

Thanks for taking all the fun out of it.



And congrats to VRWCAgent for breaking the half-million mark!!

What makes the points tally more remarkable is that he did it the hard way - he's been chugging away longer than pretty much any of us - IIRC he was among the first two or three posters to fold for TeamCFC.