Getting small WU's on a Geforce Titan

[kiore]

Ghot, I think this issue will sort itself out in a while, many people seeking the new cores and not that many available, they are new afterall..
People with 'lesser' cards as you call them or any kind of card not running regularly will probably not be pleased with the output due to the quick return bonus.
Right now a whole bunch of people are wanting to test them out to see if they are worthwhile or not so this should be less of an issue soon.
Any restrictions related to 'fairness' on what cards can do these beyond how it is already would probably be to restrict to high end AMD cards only as they have the most productivity gains..
Hang in there fold what is allocated, so you get a core 15 you don't particularly want, it will be finished soon and you go back into the draw for a core 17.

[halucigenia]

I'm currently getting "ERROR:WU01:FS01:Exception: Could not get an assignment". Is that because of the shortage of core_17 WUs. I am testing two "low end" Kepler cards at the moment and would like to keep going on linux - core_17 WUs are the only ones I can use for GPU folding.

[Napoleon]

The core_17 WU shortage will sort itself out once new projects graduate from beta to advanced. If you're in a pinch and also want to test with core_15, you could use WINE and v6 console client - see Installation instructions for v6.41 GPU in Linux. That method isn't supported officially, but it seems to work just fine nevertheless.

In my experience, low end NVidia cards run hotter with core_15 WUs, so they might actually suit your testing purposes better.

[Nathan_P]

Core 17 projects are just starting to come out, there are a limited number of WU's available compared to the ones from the older projects using Core 15. Conversely there are large numbers of people with folding systems trying to get Core 17 WU's. So at times there will not be any available when your system requests a new WU and you will be assigned one from another server with work available. As more projects are developed using Core 17 the supply will improve.

Well I can't see a reason to run a TITAN full out, and get FahCore 0x15 WU's, it's a waste of electricity for a ridiculously low Credit. Yes I DO want to help fold. I DO want to help cure diseases.

There are plenty of lesser machines out there to fold small WU's. If I can't GET large work units, I'm just shortening the life of my TITAN for WU's that could be run on a far far lesser GPU.

If you are serious about the wanting to help, fold what you get from the servers. we have all been in the position of powerful hardware folding "lesser" wu. Every WU is important, you never know which WU presents a possible cure for a disease

I suffer a 80k PPD variance on one machine, but I don't let it worry me any more. each WU is once step closer....

[bruce]

In case you've forgotten Core_17 is still under development. It seems that two very similar proteins can fold at radically different rates yielding radially different PPDs. I doubt that anybody can explain why yet. Most likely some critical optimization steps might not only accelerate folding, but narrow the variations that are currently being observed. If that's true, it will take lots of work and lots of time to incorporate a lot of minor improvements before anybody actually notices that things have improved.

Some of those changes will apply mainly to big GPUs and some will apply mainly to small/slow GPUs. Every one will probably help both, just one more than the other -- but nobody can predict which -- or when.

[n_w95482]

I'm under the impression that GTX 460 level GPU is the watershed where you start to gain any benefit from GPU QRB.

Based on my experiences with my GTX 460 1 GB, I agree with that. It gets about the same PPD with core 15 WUs as it does with core 17 WUs, and that's with it overclocked a bit. My GTX 570 and 680 went up with core 17, substantially with the 680.

Hmm, now that I think about it, maybe I should drop the advanced flag on the 460 and let it go back to core 15. That will let me run SMP 2 on the CPU again.

Edit: Switched back to core 15, the 460 gets ~5k PPD more than the core 17 WUs it worked on, plus ~1100 for SMP 2. Oh well, it was worth a try.

Well I can't see a reason to run a TITAN full out, and get FahCore 0x15 WU's, it's a waste of electricity for a ridiculously low Credit. Yes I DO want to help fold. I DO want to help cure diseases.

There are plenty of lesser machines out there to fold small WU's. If I can't GET large work units, I'm just shortening the life of my TITAN for WU's that could be run on a far far lesser GPU.

Maybe we need to rework this system, so it assigns large WU's to folks with better GPU's and smaller WU's to those with lesser GPUs. The same holds true for multiple TITAN systems. I restored from a July 10th backup and am back running an FaHCore 0x17 Project. I see no point in the larger electric bills and shorter life span of my GPU on projects that could be done with a 6870 or some such.

As others have said, 65k PPD on a core 15 WU is very good, I got about half of that with my GTX 680. Your temperatures are also very good. I wouldn't worry about it. Whether it's doing core 15 WUs or 17, it's doing good work very quickly .