Question about ranking

[Starfire_MK2]

Hi all. Well, the contest is underway! I took a look at our team scores, and have a question about the scoring setup. Here's a sample from ealier today:

358 88 (member name 1) 27 8.8 0.0 3.2 5.0 0.0 30.8 25+ 26 0.02

and here was a random selection of someone else's scores:

294 94 (member name 2) 24.5 5 0.0 0.0 5.0 0.0 47.49 25+ 22 0.03

I'm curious about the apparent inconsistancies. I know there's a reason for them, I'd just like to know how the scoring works. How can 22 wu get a 25+ rank, and get a score of 47.49 when 26 wu gets 30.8? I take it different protiens are worth more points, but who determines what machines get what protiens? Member 1 had both a higher 168 and 24 hour score than member 2, and had completed more wu's, but a much lower score and ranking. How does this work?

 

[Ploaf]

The scoring is the important part of the equation. WU's mean very little in telling how much work your system did. It's not totally consistent on a micro level but it evens out over time and it's pretty close overall. Member 1 may also have a faster cpu than member 2.

A protein A that takes hours or days or even weeks for some is worth much more than a bba5 which is complete much more quickly. The scoring is based on how long it takes a 400MHz Celeron to perform the work and the scoring for a 400MHz Celeron should be about 1 point/day.
Is that what you needed?

 

[Starfire_MK2]

So it really comes down to the complexity of a protien and the speed of a cpu, in that a slower cpu gets a higher score for processing a sample? Therefore, if Stanford arbitrarily assigns the complexity of a protien to person A, then they get a higher score for completing that particular wu than someone who completed a less complex protien? Is the assignment of protien types arbitrary?

 

[Ploaf]

So it really comes down to the complexity of a protien and the speed of a cpu, in that a slower cpu gets a higher score for processing a sample? Therefore, if Stanford arbitrarily assigns the complexity of a protien to person A, then they get a higher score for completing that particular wu than someone who completed a less complex protien? Is the assignment of protien types arbitrary?

Yes and no. Stanford does send out higher scoring wu's to some but that work unit is worth more because of it's size. Faster CPU's just do the work faster. There is no weighting based on cpu speed.

The score of the protein only represents the amount of time that it took to complete the protein. For example using a celeron 400. If Protein X takes 3 days to complete you will get 3 points. If Protein Y takes 1 day to complete you will get 1 point. Protein Y is worth less than Protein X in that it's only worth 1 point as opposed to 3, but in 3 days you can do 3 Protein Y's and still get a total of 3 points for the same amount of processing time.
This translates to all cpus, but some cpu's will get more points. A 1G Athlon may be able to do a Protein X in 1 day. That Athlon will then get 3 points for 1 days work and in 3 days it will have 9 points because it did 3 times the work that a celeron 400 did.

 

[Starfire_MK2]

I see now. Thanks, Ploaf.