Interpreting credits

Walter Kraslawsky
Walter Kraslawsky
Joined: 11 Feb 05
Posts: 5
Credit: 84103
RAC: 0
Topic 191284

Over the last 5 days, for 29 granted observations on EINSTEIN only, my average claimed credit is 41.32 and average granted credit is 34.20. Of these, 4 observations were below 50%, and 12 were above 90%, with an average of 78.9%.

Why would granted credit percent vary so widely, and be so far from full credit?

I understand about comparing results from other computers. But doesn't your algorithm for computing claimed credit account for computer speed differences already, leading to a narrow result range?

Charity begins @Home...

SETI Classic: 06 June 2002 / 51 WU / 2,554 Hours

Erik
Erik
Joined: 14 Feb 06
Posts: 2815
Credit: 2645600
RAC: 0

Interpreting credits

Quote:

Over the last 5 days, for 29 granted observations on EINSTEIN only, my average claimed credit is 41.32 and average granted credit is 34.20. Of these, 4 observations were below 50%, and 12 were above 90%, with an average of 78.9%.

Why would granted credit percent vary so widely, and be so far from full credit?

I understand about comparing results from other computers. But doesn't your algorithm for computing claimed credit account for computer speed differences already, leading to a narrow result range?


Concerning your lower average of granted credit compared to claimed credit, it is most likely attributable to the fact of some other participants using optimized Windows applications and/ or calibrated/ optimized clients. So they are crunching in less time and/ or have increased benchmarks compared to someone who has the "stock" BOINC client and Einstein application. Also there are Beta applications for Windows, Linux, and FreeBSD x86 apps that are faster than stock. I also believe I've read where Linux doesn't get benchmarked as well as Windows on identical machines.
I'll take a look at your machines and post here in a few about giving you a heads up on a faster application that will be good at least until the end of the S4 run.

Erik
Erik
Joined: 14 Feb 06
Posts: 2815
Credit: 2645600
RAC: 0

RE: I'll take a look at

Message 35002 in response to message 35001

Quote:
I'll take a look at your machines and post here in a few about giving you a heads up on a faster application that will be good at least until the end of the S4 run.

First off you'll need to download and run CPU-Z to find out what instructions your computers' processors use (MMX,SSE,SSE2,etc.)

Second, here are a couple of links to Akos' optimized apps and the Window Beta application. The CPU-Z is not needed if you decide to go with the Beta. Warning: The Beta is a beta and Akos' apps are caveat emptor (buyer beware), use at your own risk. Installation directions are found in the first post of the thread in the first link of this paragraph. The newest post in that thread has the most recent updated applications listed with the applicable computer processors.

Thirdly, here are the links to the various observation threads you should read before installation.

C41.XX
D41.XX
S41.XX
U41.XX

I'm not going to go into the optimized or calibrated client issues here (I need sleep...) but if you read through the Cruncher's Corner (and elsewhere) forum threads you'll come across discussions about BoincStudio, Truxoft, and Crunch3r clients. gl



Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.