Folding on a 680.

[Keysplayr]

I'm using FAH GPU Tracker V2 to fold on a GTX680. Is 14,174 PPD normal?
Project 8010 at 2510 Points per WU. Around just under 6WU per day.

Thanks.

Keys

 

[Keysplayr]

Ok then.
30,760 with 2x GTX680s. Almost 60 views. No comments?

 

[VirtualLarry]

I would get 11K-14K on a single GTX460 1GB OC (715Mhz) card. with a v6 GPU client.

 

[Keysplayr]

Wow. Think it's safe to directly compare? I mean, would differences in projects and points per project have an effect?
What I have going on right now:
GPU0: Running 8010 Project, 2510 Points, 15,490 PPD
GPU1: Running 8010 Project, 2510 Points, 15,490 PPD

Anyone out there running the same projects/points?

 

[somethingsketchy]

How long have the 680s been running on F@H? I'm not sure if 15K/card is the maximum PPD or not, but it could be the card(s) need to run on F@H for a bit before you would see 40K/card or higher.

 

[Keysplayr]

How long have the 680s been running on F@H? I'm not sure if 15K/card is the maximum PPD or not, but it could be the card(s) need to run on F@H for a bit before you would see 40K/card or higher.

Hmm. Well, GPU0 has been running about 3 hours and GPU1 has been running about 2.5 hours (took some time to get more than 1 GPU folding at the same time. My dumb error)
And both cards are hovering at about 15.5K +/- 100.

EDIT: Wait a sec. Both cards just climbed to about 16.25K PPD. So I guess I'll let it run a while.

 

[VirtualLarry]

You do have SLI *disabled* in the NV control panel, right? That slows down CUDA apps.

 

[petrusbroder]

The PPD changes with WUs. Therefore it is good to run the GPUs for a number of days to see how the GPUs perform.
OTOH: the 6-series of NVIDIA's card have a worse double precision performance than the 4- and 5-series have. But I don't remember if that matters. In BOINC-projects, such as MilkyWay@Home and some others, i matters greatly.

 

[Keysplayr]

Yes SLI is disabled when folding. Ok, I'll run it for a few days to see what gives.
@petrusbroder: I don't "think" double precision is a factor in F@H.

 

[biodoc]

The kepler GPUs are not officially supported yet by F@H.

http://foldingforum.org/viewtopic.php?f=67&t=22287

However, you can join the beta team to get official support.

http://foldingforum.org/viewtopic.php?f=16&t=8

Your must be running the v6 gpu client with a forcegpu_fermi flag?

I suspect your ppd will increase when they release a v7 client with "official" Kepler support.