Is 2689 -bigadv out??

wall_street_bailout_protest.jpg


What do we want?

FAH BIGADV EXPONENT EQUITABLY DECREASED!

When do we want it?

NOW! ;)

@punchy - No problem - I do get excitable about bigadv .:eek:
 
Punchy said:
I agree that a nonlinear curve is unsustainable - but remember, the bonus curve is not a full curve. The function is truncated at 10x so that bonuses don't become infinite.
Click to expand...

When I was doing some research, I found a post on FF that said the 10x bonus cap had been removed. I don't know if that's still true or not, but I think it is based on comparing the bonus formula with actual points granted. Regardless, even if the bonus was capped, the PPD would still show exponential growth as machines get faster, just with a slightly more gradual curve.

In my opinion, some sort of bonus system is fine, but the current one was not thought out very well. This is partially due to the model, but the main culprit is PG's "out of whack" benchmarking. If the points were adjusted based on good benchmarks so that a P2692 gave the same points as a P2684, this would be less of an issue.
 
The point system gives PG a huge amount of control over what hardware people buy to fold with, SR-2 for example. I don't think PG necessarily saw that the exponential bonus rate would result in a flock of SR-2s, I think they need to consider the consequences of giving points more carefully. Would it have been better for the research PG would like to run if, for example, the folks dedicated to folding splurged on 500 GTX 460s instead of 50 SR-2s? Points can do these things. If PG decides they could use more high end GPUs for big molecules WUs next gen they could simply give the big GPUs WUs more points and put what is essentially hundreds of thousands of dollars into the ideal hardware.

The trick is to maintain the integrity of the point system to maintain the competition and therefore the motivation to fold for the people who don't base their hardware purchases around WU and client releases. It is in the best interest of the research to retain casual folders rather than drive purchases from those with farms. My personal opinion is that the bigadv bonuses break the SMP point system. If someone running an overclocked Phenom II only earns one sixth the points of my two year old i7 they see this in the forums and the competition, the main motivation to keep clients running, goes away.
 
Last edited:
All I know is that the PG have a few things up their sleeves to improve the points situations and one of the most important is the QRB (quick return bonus) which will be applied on all client versions as soon as the WS code get updated. With that, quick return of workunits will get rewarded and the issues we have with WU dumping and other bad behaviors will disappear.

The second thing is that when the DAB is fully settled, be assured that we will be discuss the benchmarking system to find a better method which will get a very predicable points output for any hardware vs the capabilities of such hardware.

That's why we need to take things you all saw with a grain of salt for now. The PG is working hard to improve the situation with the points.
 
You must log in or register to reply here.
Back
Top