Credit where credit's due

Post by **UBT - Chris Suddick** » Mon Mar 05, 2007 5:57 pm

There have been a few postings that have talked about credit awarded for various projects and it has prompted me to do a few calculations on the credits awarded by different projects that I crunch (I'm a bit sad that way I'm afraid). I gather it can vary depending on the kind of processor you have and the OS that you run although why it should I can't imagine. Anyway, I have an AMD 2400 running Linux and for the 6 projects I currently have connected I get the following daily credits:

Simap... 460
Einstein... 405
Rosetta... 215
Malaria... 184
LHC... 164
SAP... 160

Whilst talking to my daughter the other day, we did the calculations for her machine which runs Windows XP but I don't know the spec. I can't remember the figures but I do recall that Rosetta came out higher than Einstein. It seems strange that there should be such a large discrepancy between all the projects and even stranger that relative order should vary between processors.

Anyone else ever done the calculation?

Chris.

UBT - JohnR · Post by **UBT - JohnR** » Mon Mar 05, 2007 6:16 pm

SAP is very memory intensive. The bigger the cache on the processor the better, and fast memory is also needed. I have a Celeron that is 50% faster than a AMD 3000+ with slow memory
I have always found Einstein gives good credits, Simap is more variable but at times is better than Einstein. Your results agree with my findings.

Rockinfroggi · Post by **Rockinfroggi** » Mon Mar 05, 2007 6:36 pm

Might be interesting to do a few comparisons, but what method did you use to get your daily totals?
I just did a quick tally on SIMAP which at the moment is running on my HT 3.0 over-clocked to 3.8 with 2 Gigs Ram with winXP and my total for yesterday was around half yours at 245, that was just adding the granted credits on my results page.

Gary.

Post by **UBT-mark3346** » Mon Mar 05, 2007 7:01 pm

I would say take quite a few granted credit results then
total credit/total hours = credit per hour (bigger the sample the more accurate the figure will be). There may be other ways of doing this but you would have to deal with pending credits and do they grant what is claimed etc, this way gives an hourly figure you turn to daily if you wish.
With CPDN you can check the time and credits per trickle, I have 2 on a dual core at the moment and get around 500 credits/14 hours with the two togeather. I ran SAP on the same machine (pentium) and on an AMD and that was one of the few projects for me where the AMD did not get the higher credit per hour so I am not going to try any CPDN on the AMD.

Post by **UBT - Chris Suddick** » Tue Mar 06, 2007 9:06 am

Rockinfroggi wrote:Might be interesting to do a few comparisons, but what method did you use to get your daily totals?
I just did a quick tally on SIMAP which at the moment is running on my HT 3.0 over-clocked to 3.8 with 2 Gigs Ram with winXP and my total for yesterday was around half yours at 245, that was just adding the granted credits on my results page.

Gary.

The method I use is 86400 [No of secs in a day] / (time taken to crunch n WUs) * (credit granted for n WUs). I usually take n=5 for no readily apparent reason, it just seemed like a good idea at the time. I tend to ignore the pending WUs and just use the most recently granted WUs.

Chris.