Originally posted by: MichaelD
I have noticed that I've got a lot of points (not complaining!) for relatively few WUs. Others have just as many points as I, but way more WUs.
Do you get awarded more points, the faster you turn in the completed WU?
Thanks.
You have been getting boffo points lately, MichaelD, keep it up!
The point credit system is a little weird, though... there is a huge amount of variety among the WUs.
Basically what Stanford does is
benchmark all WUs on a single Pentium 4 2.8 GHz box running Linux. Based on how well that particular machine can crunch the WUs, points are assigned. You can check point values for all the proteins
here.
I've noticed that my Athlon box running Windows gets results wildly divergent from Stanford's Intel/Linux baseline. It tends to hinge mostly on the difference between the two F@H cores (Tinker is the older one, Gromacs is newer code). My box just finished a 242 point Tinker in 27 hours (about 9 pts per hour rate), while two days ago it took 8 hours to get through a 46 point Gromacs (only 5.75 pph). And that was with SSE and other speed-enhancing options turned on. YMMV, of course.
-----------------------------------------------------------------------------------
-Adam in Philly
Athlon 64 FX-51 @ 2.4GHz/1.625vcore - ASUS SK8V
Liquid cooling by Koolance