On 09 Feb 1997 11:43, Jeff Guerdat (1:2613/313@) wrote:
 JG>EB>Mistake nr 1. The operating system and all other tasks 
 JG>running will also
 JG>EB>use that single processor. On a multi-processor system all 
 JG>threads (over
 JG>EB>all processes) will share the CPU's. Even a single threaded 
 JG>application
 JG>EB>will show a performance boost because it can get more 
 JG>cycles.
 JG>
 JG>Not extensively benchmarked but we have tried this a few times 
 JG>and have found
 JG>only a ~10% (the exact number is no doubt variable) performance 
 JG>gain.  One
 JG>would hope for at least a 50% gain given 2 CPUs vs. 1...
It depends on how you define, and measure, performance. I have some heavy I/O 
intensive applications running on a number of servers and all multi-processor 
machines show an almost doubled performance. 
It's normal for a simple application to show a relative low boost on a 
multi-processor machine because you are measuring one application only. If 
you would measure the overall performance then you measure a much better 
improvement. This is also one of the reasons for developing multi-threaded 
applications because each thread can perform better on it's own, if the 
application has been designed properly.
Cheers
Ernie
--- GEcho 1.00
---------------