Hello

As far as I'm aware, CPDN does not employ a quick return bonus scheme (if there is one please point me to the right direction :) ). Why can't CPDN implement say a 25 or 50% QRB on units that are returned ASAP or within a certain deadline? GPUGrid and F@H employ these measures to excellent effect, and I'm sure it could increase CPDN productivity/participation.

Just curious is all.

Regards

I personally would be against having a rapid turn around bonus. I contribute to this project because I believe in it. I think a rapid turn around bonus would encourage people to leave their computers on all the time adding to the global warming problem, rather than using spare capacity on a computer that is going to be switched on anyway.

Don't know if that is the reason or whether it has just not been thought about here before.

Dave

It's all to do with what this project is about: CLIMATE modelling.

Climate is weather over a very long period. Weather is a "chaotic system", meaning it can vary unpredictably, causing the climate derived from it to also vary widely from a given starting point.

People after credits often overclock their computers to complete work units faster, and therefore get more credits in a given period of time.

On other projects, most of which use integer maths, they can run the same data on 2 or 3 computers and then compare results. This can't be done here. So the climatologists use statistical methods to determine whether or not the models 'look' realistic. Those that don't aren't used for their work.

Overclocking makes a computer less stable than without it. And the more it's overclocked, the more chance there is of the processor's floating point unit (FPU), returning the wrong value, which makes the model at that point slightly different to what it should be, And this difference can lead to a different end result. Ultimately the overclocking causes the models to crash, but up until then, they may look OK to the cruncher.

And back to what you were asking: Handing out a bonus credit for fast completion may cause people to start overclocking to get more. So it's not going to happen.

The days when this project was exploring the vast area of parameter space with simple models is over. Now the climatologists want longer and/or more detailed models.
And only a few thousand are being released at a time. To be quickly snapped up by 30,000+ computers.
So there's no need to increase participation.

---

Backups: Here

I find this idea - bonus credits for fast processing - extremely destructive.
Because it is the same as bonus for the new hardware. The same as the credits, generated from the air.
The credits must be as indicator of useful job done and nothing more.
I hope, CPDN will never go to such scheme.
I hope, this project will never be a rich people club (like GPUGrid and, maybe, other projects).

---