Re: maximum/optimal # of SMP CPUs between 2.2 and 2.4

[Alan Cox]

> Aren't these quoted results from *static* web page tests?  And didn't 
> Linux use khttp?  If so, then unless NT is also using a kernel resident
> server, comparing these is perhaps useful for marketing but is not useful
> for engineering (and this is absolutely an engineering forum).

All of these tests are generally misleading. Linux holds web records
that include cgi but I'd hesitate to say that means we are the worlds best
web server. I could certainly go to a testing lab and get them to verify 
that we totally kick the other peoples backside, but as we know from the 
past in the other direction - its a bit dubious.

The only real test for a system performance is still how it runs real jobs
similar to the ones you plan to run. Unfortunately unless you are making large
purchases doing benchmarks first is rarely an option.

Alan