Attn GPU Folders

Biffa · 13 Nov 2012 at 12:51

Looks like the beta testing is still underway for the QRB WU's, so its worth setting your clients up to get them, all for the good of the team!

Here are some tips and tricks from our friends over at [H]ardforum:

musky said:
If you are folding an nVidia GTX400 series or better video card, make sure you have your client set up to receive the quick return bonus (QRB) work units now in beta. You mainly need two things (3 with the v7 client):

1. A passkey - this is the same passkey you use for SMP work. I assume most of you have one already.

2. You need to set the beta flag. For the version 6.xx client, you just add -betateam to your shortcut to start the client. In v7, it is under the Configure button - Slots tab - Extra slot options. Add a new option, Name is client-type, Value is beta.

3. (v7 only) add another Extra slot option, Name is next-unit-percentage, Value is 100.

If you have any issues setting this up, please ask. The point value on these units (8057s) is very high, and other teams have taken advantage of them. We need to as well.

On my GTX470 the point value for a 8057 is 72K that means if I can get all the 470's on weebeastie folding a 8057 the full output would be:

CPU0: - 17K
GPU0: - 72K
GPU0: - 72K
GPU0: - 72K
GPU0: - 72K
Total: - 305K

Of course, now I've set it up I have one gpu with a 8057 and the others with 12K 8018 WUs.. but hey

That being said, these are beta team WU's and as such you can't get any help from Stanford for them or have any recourse if they set your PC on fire etc etc.

As to the ethics of the practice of doing beta WU's when not on the beta team I leave it to you individually to decide. :cool:

Biffa · 13 Nov 2012 at 15:40

Are you using the v7 client?

Biffa · 13 Nov 2012 at 20:25

ozaudio said:
yep , well 7.2.9 , is this correct?

https://fah-web.stanford.edu/projects/FAHClient/wiki/BetaRelease

that one

So you need to be in advanced mode and add the flags above to the gpu slots

Biffa · 14 Nov 2012 at 15:14

overclocked? I've found I have to dial back to stock speeds to get 100% success rates on both the 8057 and the 8018's (High/Low WU's)

Biffa · 15 Nov 2012 at 08:09

No you should only have the two flags from the first post:

client-type: beta
next-unit-percentage: 100

The client-type: advanced will conflict

And you only get a 8057 sporadically, 8018 is the norm.

Biffa · 16 Nov 2012 at 00:30

More info here, especially about switching to the 2.22 gpu client for around 25-30% more PPD on Fermi cards.

FYI my 470's jumped from 63K PPD to near 100K just by changing to the 2.22 client from the 2.25 client.

Interesting to note a 580GTX class card can pull nearly 250K using the 2.22 client if hitting a good high stable overclock.

Unfortunately the Keplar cards (6* series) only work with the 2.25 client, so although yes you still get a great boost on those cards its still relatively small compared to a equivalent price or cheaper Fermi. For reference a 660Ti gets around 110K so the same as a stock 470

If you can overclock then you really start to see some returns.

Now this might not last for long, they may re-align the QRB or we just might be benefitting from the fact these are test WU's with only 900 atoms. But for now, lets catch some teams and leave some others in our dust because it won't be long before some other teams catch on and then we will all be even again.

Biffa · 16 Nov 2012 at 11:29

I've just put my 590's on the "cove of drugs", along with the unused water blocks for a very palatable sum allegedly.

Biffa · 17 Nov 2012 at 00:06

You have the worst luck loudbob

Biffa · 17 Nov 2012 at 15:51

Nope, well Folding works but not with any sort of performance to write home about, and these QRB WU's are Nvidia only

Biffa · 17 Nov 2012 at 16:14

Thats the right amount of credits for a 8018, they are not the QRB work units, those are 8057's

PPD for a 8018 is probably around 19-20K for a 670, not surew what you are using the the CPU but I get around 19K on a [email protected] while gpu crunching on the same machine.

Biffa · 17 Nov 2012 at 16:47

Sounds about right then, overclocking the CPU will near double its output, putting a 460 in there will help more right now with the new WU's but the more gpus you have the less cpu power you have left over to crunch.

Biffa · 19 Nov 2012 at 22:47

It should download and update to 2.25 automagically, people have been complaining that it does.

Biffa · 23 Nov 2012 at 00:01

Same here

Biffa · 23 Nov 2012 at 01:11

Just got three

Just so you guys know, the reason the 2.25 client and 2.22 client are in different directories is because the only beta WU's are the 8057's and they will automatically use the 2.22 client, wheras any other WU will use the non beta 2.25 client from the non beta directory.

Also if you are on a Keplar device you *have* to use 2.25

Biffa · 6 Dec 2012 at 21:33

Unified GPU/SMP benchmarking scheme implemented

The unified GPU/SMP benchmarking scheme will be rolled out today on FAH. All the future GPU projects would now be benchmarked using the new scheme. Most of the existing GPU projects have been re-benchmarked to reflect the changes in the benchmarking scheme. We are still in the process of re-benchmarking some old projects assigned to ATI and G80 GPUs. The uniform benchmarking scheme significantly boosts the base points for all GPU projects. However, Quick Return Bonus for the GPU clients has not been introduced at this stage, but will be introduced once we work out an issue on our side. We would like to thank all the beta testers who participated in testing the unified benchmarking scheme.

Biffa · 8 Dec 2012 at 08:33

QRB is turned off at the moment.