[ale] BogoMips uselsess?

Vaidhyanathan G Mayilrangam vaidhy at loonys.net
Wed May 5 00:56:38 EDT 1999


On Tue, May 04, 1999 at 11:11:13PM -0400, Glenn R. Stone wrote:
> Byron A Jeff wrote:
> > 
> > >
> > >
> > >   I just built out a k6-2-400.  It reports 700+ bogomips.  Looks nice, but
> > > my Dual-P2-300 reports 600.  Should I be concerned that a 400 is faster
> > > than mine?
> > 
> > Bogo is derived from Bogus. The numbers only apply in the same chip families.
> > Since the P2 is completely different than the k6-2 the bogomip numbers cannot
> > in any way, shape, or form be related.
> 
> I have seen a K6-2-300 outcompile a PII-350... but the idea that one K6-2/400
> beats a PAIR of anything in the same class is just what Byron said.  Bogus. 
> Essentially it's a timing loop, and it's probably only getting a look at 
> one of the two Intels... 
> 
> Which leads me to the following related question:  is GNU make(1) smart enough 
> to take advantage of SMP?  I remember at GaTech we had a Sequent Symmetry we 
> could kick off make -P and snarf all ten processors on the thing.... emacs
> compiled in a trice.  One would love to turn a quad-Xeon loose on the kernel
> recompile.... can you say, way sub-sixty-seconds?  
> 
> -- Glenn

Just call it with make -j2 or make -j4 and it will use smp :)

Cheers,
Vaidhy






More information about the Ale mailing list