Joined: 19 May 99
One of many examples: http://setiweb.ssl.berkeley.edu/workunit.php?wuid=2929795
My Dual 1 GHz claimed 29.58 credits while the next machine, a Pentium 4 Dual 3 GHz claimes 40.07 credits although it finished 7,000 seconds faster. Wasn't the idea of this complicated credit system that slower machines get more credit but faster ones compensate for this through more WUs per day? When looking at Windows/Mac relations it looks exactly the other way.
|ID: 38643 · |
Keith J. Schultz|
Joined: 6 Apr 01
One answer might be that your machine has a higher ranking when running the benchmarks!
Also, if I understand the system correctly the computer who claims less credit is given credit !!
See other posts.
If you look and the granted credits you were given 40.07 credits. More than you claimed.
So why worry !
Keith J. Schultz
|ID: 38764 · |
Joined: 11 Apr 03
> If you look and the granted credits you were given 40.07 credits. More than
> you claimed.
> So why worry !
I have the same problem. I switched back from the optimized client to the stock Seti version of 4.13 and am still getting the same.
Work Unit I.D. - 04mr04aa.10196.31970.479830.128
Other participants computer
-AMD Athlon XP-M Pentium
-Measured floating point speed 2,374.19 million ops/sec
-Measured integer speed 5,704.99 million ops/sec
-Work Unit completed in 8,210.11 seconds or about 2 1/2 hours
-Power Macintosh PowerMac4,2
-Measured floating point speed 330.58 million ops/sec (-2,043.61 or 7 times slower compared)
-Measured integer speed 998.55 million ops/sec (-4,706.44 or 6 times slower compared)
- Work Unit completed in 42,792.10 seconds or almost 12 hours (5 times slower compared)
Other participant claimed 38.61
I claimed 32.91 (-5.7 less)
Why worry about claiming a low credit?
A claims 10
B claims 20
C claims 30
They all get 20.
If A claimed the actual work time of say 40 instead of an incorrect 10
then they all should have gotten 30.
Everyone on that work unit loses 10 because there is something wrong with how A claims credit.
Because A is a Mac, then on all Units a Mac works on not only do they lose credit, but everyone on those units lose too.
This needs to be fixed and is not because of the optimized client. This example is from a stock Seti version.
FYI, I reset the project and re benchmarked before these results were received.
<IMG SRC=\"http://www.boincstats.com/stats/banner.php?cpid=dd63d11af421f351945d4c6f8f23ba7b\" WIDTH=\"161\" HEIGHT=\"90\" BORDER=\"0\" ALT=>
|ID: 45242 · |
Joined: 19 May 99
> > If you look and the granted credits you were given 40.07 credits. More
> > you claimed.
> > So why worry !
> I have the same problem. I switched back from the optimized client to the
> stock Seti version of 4.13 and am still getting the same.
> Work Unit I.D. - 04mr04aa.10196.31970.479830.128
> Other participants computer
> -AMD Athlon XP-M Pentium
> -Measured floating point speed 2,374.19 million ops/sec
> -Measured integer speed 5,704.99 million ops/sec
> -Work Unit completed in 8,210.11 seconds or about 2 1/2 hours
thanks for the support! That's exactly what I am talking about! Furthermore the benchmark results you mention are most probably cheated!
My brand new G5 Dual 2.5 GHz machine shows:
Measured floating point speed: 1325.05 million ops/sec
Measured integer speed: 3752.9 million ops/sec
and needs appr. 8-9000 seconds/unit.
Same time but only half of the benchmark results! If I understand correctly the time is multiplied with the benchmark result (plus some more adjustments). Therefore this AMD machine automatically claims twice as many credits as my Mac would - and then your argumentation kicks in.
|ID: 46561 · |