Athlon MP Kernel Benchmark error?

[Red_Zealot]

I was really struck by the compilation benchmarks that showed single procs beating dual procs.

I dont have much experience with Linux MP, but I've read that kernel compilations under MP go much faster than single procs, but only when using the -j (or is it -k) switch for the number of compilation jobs. 2 or 3 jobs MP show significant improvements over single jobs.

Do you all think this is the case here?

"If you teach a child to read, then he or her will be able to pass a literacy test" - George W.

 

[Guest]

sure looks like it, if you compare the results:

http://www.tomshardware.com/mainboard/02q1/020211/dualathlon-09.html
http://www.anandtech.com/chipsets/showdoc.html?i=1570&p=8

In Anand's test, dual XP1900+ is about twice as fast than dual XP2000+ in Toms test.

drac

 

[poorboy]

It's -j, and I'd be betting that it is the problem. As usual, not too many words to go with the pictures, so we can only guess. But then I'm not a big fan of Frank's reviews...

Edit: <i>"We were not able to activate the MP mode of Quake 3 in Windows 2000 SP2. Result: the benchmark caused a system crash. For this reason, we only calculated the time-demo runs in single mode. As you can see from the chart, the data overhead in dual operation results in a drop in frame rate."</i>
I'd question the relevance of an OpenGL benchmark for a Mobo/CPU review anyway, but what is the F*N point of this?

I'd like to see a dualie chipset from NVIDIA. Over 4GB/s of memory bandwidth, with dual Athlon XPs. Now that would run my database just nicely...

<P ID="edit"><FONT SIZE=-1><EM>Edited by poorboy on 02/12/02 05:07 AM.</EM></FONT></P>