Riddle me this - FaD points

[BCinSC]

Timeate:Query:Job:NickName:Member:Hits:Moleculesoints:Time:GFlops
13:17 24-Jun-04 1a30-q2 m8839 BCinSC 4004561 403 9797 5382 0:001:11:21:30 43056
13:19 24-Jun-04 1uou-q1 m16208 BCinSC 1009897 221 10000 13242 0:001:11:18:28 105936

Two machines, roughly equivilent, slaving away almost exactly same time. The one that gets 403 hits only gets 5382 points, while the one with 221 gets over 13,000! And look at difference in GFlops. "Today was a bad day to give up glue sniffing." - Airplane!

 

[grendel8]

Check your *.tlg logs to see if a lot of the molecules on the low rated machine were discarded.

 

[dringdahl]

It's that black magic they use for calculating point/gflops...

I have noticed too that certain molecules type give better points, like UOU and YET. If you want more point crhunch those... THey do take longer (up to a couple days) but up make a lot more progress.

FYI: I think is am setting a think record... I have a A30 job running on a 248 opteron, that has been chugging for 9 days... and has 830 hits so far... Wonder how many point that will be..

 

[BCinSC]

What OS?

 

[Unforgiven]

Originally posted by: dringdahl
FYI: I think is am setting a think record... I have a A30 job running on a 248 opteron, that has been chugging for 9 days... and has 830 hits so far... Wonder how many point that will be..

holy crap dude, keep up the good work :beer:

 

[dringdahl]

Both Linux and Windows

 

[xComputerManiacx]

i thought points were based on the amount of gflops processed, not the number of hits. it could just be that the second job had a larger amount of gflops processed because it processed a bunch of near-hits that turned out to be misses near the end of the computation. more work done, but less hits total.