Folding Forum

As a new GPU folder myself, we have to remember that it is really important to note which WU is deviated from the benchmark.

If we note that the benchmark GPU [GTX 460] should earn 9787PPD - that puts the 76xx projects pretty much bang in line with the benchmark. However, because 8020 earns considerably more than 9787 PPD then the conclusion has to be that it is over valued. As far as I see 76xx should have re-emerged with the same ppd as before if it's bang on the benchmark value. We should probably be discussing this issue in the thread of projects that are overvalued, rather than this one.

As to why the previous project(s) are overvalued - I'm not entirely sure why it was higher and why points weren't brought more in line. I don't know whether a different GPU was used to benchmark those projects or perhaps PG just decided that the higher PPD on the project was just a nice short term perk for GPU folders (perhaps devaluing them would have had a big impact on more powerful GPUs). Perhaps they just considered that project more scientifically important - say because they wanted a quick return of points. Given the current situation that is an entirely plausible method...

Unfortunately I think we can both agree that high PPD is just as dangerous as low PPD, it just comes out long term as a lot of users, such as DoctorsSon, and other users who start folding with those projects, get use to the higher PPD and then feel like the accurate PPD is too low and not doing enough science.

Of course that does to some degree depend on whether people feel that a point is worth a point and should not change in value. That has since been modified to say that a point is now based on the 'scientific value', which is much more dynamic.

What doesn't change, in my mind, is that complaining about these projects being 'undervalued' relative to other projects will see the other projects devalued, rather than these projects being increased in value...

k1wi wrote:...

What doesn't change, in my mind, is that complaining about these projects being 'undervalued' relative to other projects will see the other projects devalued, rather than these projects being increased in value...

Seems like someone said something like that back on page 1.

There is nothing wrong with these projects netting ~10K PPD on a 460 GPU. This is exactly as designed according to the benchmark standard. There is nothing more that can be said about these projects that would motivate Pande Group to change the points on these projects. There is nothing wrong, so PG would change nothing.

If there is a problem, it's with the 8020 project having too high a score, and you would need to start a new topic about that project.

I was getting great points on 8020s on my recently added gtx460 [aiming at getting steady 15k ppd] though this was not the max, just reliably stable.
I then hit the 76xx's a few days back and thought there was something amiss in the card works, as the drop was as described by Halo.
Now I've seen the overview here, I can concur that I assumed that the 8020 were either overvalued for science or throughput reasons rather than the 76xx undervalued.[Why? The 8020 was high historically from what I'd seen reported]
Doing the work for points only is a mistake in my personal view, though I can see the gaming side getting understandable interest in what can be a slog of experimental number crunching from the local side of things -as 'just' a PC provider.
I am always aware that points for projects vary anyway- this example does seem to be a large variation, even so -and why it has attracted comment.
Just so long as all is well with the blessed shaders- sucking the amps
The 'Science' must always win- I wouldn't like a job looking after all those servers ,either as Post BSc,MA/PhD or whatever!