If we make the assumption that both pipelines could be kept completely
full, and since BogoMIPS is essentially a timing loop I hope the branch
prediction could get that one right, wouldn't it be reasonable to have one
completed instruction per pipe per clock cycle?

If there's an electrical engineer among us please correct me.

I just did a little looking into BogoMIPS and there are all kinds of
things that can make it give errenous numbers.  BUT it's as reliable as
any other benchmark.  And I'll use it to prove mine's faster than the
Jones's down the street. ;-)


On Wed, 12 Jul 2000, ryan hankins wrote:

> 
> I don't _think_ intel publishes the number of cycles each instruction
> takes.


---------------------------------------------------------------------
To unsubscribe, e-mail: tclug-list-unsubscribe at mn-linux.org
For additional commands, e-mail: tclug-list-help at mn-linux.org