Re: [NTLK] [OT] SETI WU

From: Sushi (Sushi_at_ragingbull.com)
Date: Wed Apr 21 2004 - 08:13:11 PDT


>On 4/21/04 @ 3:54 AM, Richard Kilpatrick wrote:
>
>Meh, I wasn't grumping about methodology, I was just
>pointing out that WUs are quite a variable measure of
>performance - my fast results on the G5 could just
>have been very simple units.
>
>IIRC the actual ratio was the G5 completing 4 units in
>the same time as the G4 completing 1. Logging on to SETI
>gives me an average of 5 hours per work unit, which is
>the G5 I'm sure, but it appears to record double the CPU
>time for dual processors, without seemingly using both
>CPUs. I'm not sure exactly how to use SETI as anything
>other than a guideline :)

Richard, yes you are correct. SETI WUs are different and thus vary in
the time required to complete them.

I generally use a range: fastest WU to slowest WU to get a feel for a
machines performance. While not scientific in the least, it does give a
rough comparison between two systems concerning running SETI software.

The only way you could completely compare two computers using SETI is to
have each computer complete the same WU and record the time it takes.
Then again, this would not be a perfect test because of the different
angles and calculations required. However, since I do not have the
source code, I do not know their methodology for calculating a WU. It
could be that their methodology favors a certain chip architecture over
another. Then again, it may not.

So for me, I just use it as an unscientific rough comparison, though a
fun one nonetheless. That's all.

Looking forward to seeing you on the Newton SETI team contributing all
that G5 power! :-)

Sushi

-- 
This is the NewtonTalk list - http://www.newtontalk.net/ for all inquiries
Official Newton FAQ: http://www.chuma.org/newton/faq/
WikiWikiNewt for all kinds of articles: http://tools.unna.org/wikiwikinewt/


This archive was generated by hypermail 2.1.5 : Wed Apr 21 2004 - 08:30:01 PDT