K-factor, the i5 CPU 750 @ 2.67GHz, and Moore's law

[wuffy68]

A couple of questions about the way folding@home determines PPD points

The Folding@home Points System

1. Does the K-Factor used in the determination of points PPD relate to the kinetics and thermodynamic properties in any given protein molecular model?
2. The i5 CPU 750 processor, used as a baseline for simulations, was first released in 2009; does Folding@home account for Moore's law in other ways by increasing difficulty through fewer points being awarded as time goes on, or are points based on a linear projection off the baseline processor, with the QRB (Quick Return Bonus) used to account for Moore's law (the newer the processor or GPU, to higher the QRB awarded).

Hopefully those questions make sense - just trying to clarify this for some of our team members.

Thank you,

-wuffy68

[PantherX]

1) K-factor is this:

The k-factor, a coefficient in awarding bonus points, is currently set to a baseline value of 0.75, but may vary depending on the scientific value of a project.

https://foldingathome.org/faqs/points/b ... etermined/

2) The scale is exponential, not linear as that's how the QRB system was designed.

[wuffy68]

Thank you PatherX, So the K-factor is determined more by scientific value of the WU - rather than anything directly related to formulas used in protein folding dynamics equations?

[PantherX]

Yep, that's my understanding

[Joe_H]

In the past the K-factor was used to adjust the level of bonus achieved by faster systems, now it is being used as a constant. So currently all projects use a K-factor of 0.75.

One of the main reasons behind the QRB formula in the beginning was that on multiprocessor systems running separate WUs on each processor would give more points per day. But that meant each WU took longer and delayed the return and generation of the next WU for each PRCG. This part of the reasoning for the QRB was to make up for the fact that there is overhead involved in multiprocessing, doubling processors does not cut the time in half.