Toggle sidebar Toggle sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

What hardware are you folding with?

twoodcc said:
yes you should! well, 2 is better than 1! 😀
Click to expand...
Sad when the GPU costs more than the Mac Pro itself 😛

----------
Louis Wu said:
that optical bay PSU certainly is a more elegant solution than hacking wires into your case...

... but I guess the question is -- does the Titan work for folding when you've got it booted in Windows only? That's disappointing about the GPU not being accessible with a VM...
Click to expand...
Yeah I'm booted into windows now and it's folding away.
 
OrangeSVTguy said:
Sad when the GPU costs more than the Mac Pro itself 😛

----------


Yeah I'm booted into windows now and it's folding away.
Click to expand...

yeah, but only because Mac Pros are so cheap now! Just wait till the new ones come out!

all right! look forward to seeing you move up the ranks again 😀
 
SilentPanda said:
But I want 8900, 7810, 7811 units! I hear they are better ppd.
Click to expand...

One thing I noticed on your screenshot the other post that you have CPU:8 configured.
Try to set CPU:6 in your case. Problem is that under core 17 each NV card need one dedicated core. That is not the case for core 15.
The Standford server might not assign you core 17 because of that.

Give it a try.

I have it actually set to CPU:5, each GPU one core plus "one core for the can"
 
ChristianJapan said:
One thing I noticed on your screenshot the other post that you have CPU:8 configured.
Try to set CPU:6 in your case. Problem is that under core 17 each NV card need one dedicated core. That is not the case for core 15.
The Standford server might not assign you core 17 because of that.

Give it a try.

I have it actually set to CPU:5, each GPU one core plus "one core for the can"
Click to expand...

How do you set that?

Edit: Nevermind. Found it. 7 minutes till my 670 gets a new work assignment and it hasn't downloaded the next so guess I'll see.

Edit2: It snagged an 8018... lame. 1 1/2 hours for the Titan. I did restart the client prior to the 670 getting it's WU just in case.
 
Last edited:
After not getting core 17's with less CPU, I changed client-type from advanced to beta. Now I got core 17's! Huzzah!
 
SilentPanda said:
After not getting core 17's with less CPU, I changed client-type from advanced to beta. Now I got core 17's! Huzzah!
Click to expand...

Strange ... Seems the gamble in advanced to get "nice" WU is over. They might changed the priority to core 15 to get rid of the remaining WUs. You just need to accept that in beta mode you can loose WU (not very often, but happen). I had once a bad WU which I finally had to dump. Or some with extreme long runtime and low resulting PPD.

Assuming you got 8900: what is your TPF/PPD ?
 
The 670 got a 7810 and is currently at 2min 14s TPF.
The Titan got a 7811 and is currently at 2min 10s TPF.

I'm going to let it hang like this today then try turning the 670 off tomorrow. The Titan adjusts its speed based on the temperature of the card, so I think the extra heat from the 670 is slowing it down. Basically find out which gives better PPD.

Regardless, my overall PPD was around 50k and now it's at 158k which is at least a step in the right direction. Might take a week or two but I'll figure it out.
 
Grrrrrrr arghhh!

So FAH shows slot 0 as my 670 and slot 1 as my Titan (doesn't really matter which slot # it assigns). I used GPU-Z to inspect my cards and it shows that the Titan is performance capped because of temperature limit. Makes sense, that's what it's supposed to do. So I turned off in F@H what was labelled as the 670 to get a feel for "will it fold faster if the 670 temp isn't there". Went back to GPU-Z and it shows the Titan at 0% GPU load and the 670 at 99%! I also looked in Speccy (another program) and it too showed the 670 going and the Titan doing nothing.

So basically they appear to be labeled backwards in the F@H client for some reason. I even did a complete removal of the F@H client but nope... they're still backwards.

It doesn't hurt anything probably, unless at some point either card is supposed to get a specific type of WU that the other can't process. Just odd they're labeled backwards.

Whatever... I'm just gonna let them fold. I got passed costabunny and I can't let her catch up again.
 
What is everyone running on these days? It's been a while since I updated myself on PPD.

What sites do you use to read up on rigs? Thinking about building something around xmas 😀 Budget would be around $600-1000 and I would just need CPU - Mobo - GPU.
 
Right now folding with 2 GTX 780 and on weekends I add one GTX 660TI. All on one Asus P8Z77 with 8GB RAM and from SSD. OS is Ubuntu 13.04.
One GTX 780 makes 130kPPD; the 660TI around 61kPPD.

For little stuff I have one CPU:3 and one CPU:4; both together netting in 20kPPD.
Of course all those PPD depend on the WU assigned.
 
DeSnousa said:
What is everyone running on these days? It's been a while since I updated myself on PPD.

What sites do you use to read up on rigs? Thinking about building something around xmas 😀 Budget would be around $600-1000 and I would just need CPU - Mobo - GPU.
Click to expand...

this thread on the [H] forum is pretty timely. most of the posts refer to projects other than f@h but there's a good reply from Zagen30.

http://hardforum.com/showthread.php?t=1786969

GPUs are really the most flexible way to go these days; build the rig for whatever else you want to do and then stick in a decent card and forget about CPU folding unless you're going with a dedicated multiprocessor rig...

Let us know what you decide to build!

Louis

----------
SilentPanda said:
Grrrrrrr arghhh!

So FAH shows slot 0 as my 670 and slot 1 as my Titan (doesn't really matter which slot # it assigns). I used GPU-Z to inspect my cards and it shows that the Titan is performance capped because of temperature limit. Makes sense, that's what it's supposed to do. So I turned off in F@H what was labelled as the 670 to get a feel for "will it fold faster if the 670 temp isn't there". Went back to GPU-Z and it shows the Titan at 0% GPU load and the 670 at 99%! I also looked in Speccy (another program) and it too showed the 670 going and the Titan doing nothing.

So basically they appear to be labeled backwards in the F@H client for some reason. I even did a complete removal of the F@H client but nope... they're still backwards.

It doesn't hurt anything probably, unless at some point either card is supposed to get a specific type of WU that the other can't process. Just odd they're labeled backwards.

Whatever... I'm just gonna let them fold. I got passed costabunny and I can't let her catch up again.
Click to expand...

You can always, if you want, set the GPUs to "Finish"; wait for the WUs to finish and then delete the slots, then add them again through configuration. I found that sometimes helps the client identify the GPUs properly.
 
DeSnousa said:
What is everyone running on these days? It's been a while since I updated myself on PPD.
Click to expand...

Oh, I'm running =:

rig 1: 2 x AMD 7970 OC on a Celeron chip, basic Gigabyte board.
rig 2: 1 x AMD 7970 OC and 1 x nVidia 660ti on a Pentium chip; Gigabyte board.
rig 3: 4 x AMD Opteron 6128 (8 core, 2.0 MHz overclocked to 2.31 MHz) on a Supermicro board.

rigs 1 & 2 are GPU only; rig 3 is CPU (with bigadv) only.
 
OrangeSVTguy said:
Just wondering how that titan would bring that in? Maybe I don't have it configured correctly? GPU and CPU are at roughly 75k combined.
Click to expand...

Client-type = advanced should give you access to public beta Core 17 work units. Those together with a qualified passkey get you quite some points. Our Panda had some suboptimal experience though: even with the right client-type he got only core 15 WU. Give it a try.

My Titan Mini (GTX 780) gets me each 130kPPD.
 
ChristianJapan said:
Client-type = advanced should give you access to public beta Core 17 work units. Those together with a qualified passkey get you quite some points. Our Panda had some suboptimal experience though: even with the right client-type he got only core 15 WU. Give it a try.

My Titan Mini (GTX 780) gets me each 130kPPD.
Click to expand...

It's luck of the draw I think. I just got a core15 with the beta flag too. No biggie... it's doing SCIENCE.
 
SilentPanda said:
It's luck of the draw I think. I just got a core15 with the beta flag too. No biggie... it's doing SCIENCE.
Click to expand...

yeah, I just got a Core16 which is what they're trying to finish up.

I've heard on other boards that a Titan on Core17 should pull in 180-200 K PPD on its own...
 
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back