On Thu, 11 Jun 1998, Vincent BARAT wrote:

> Yesterday, I tried to run the Byte Unix Benchmark on my Centris 650
> (NetBSD 1.3, 68040 30 MHz, slow SCSI drive, 24 Mo).
> 
> The score I obtained was ... 1.0 ... which corresponds to
> the score of a 360 SX 33 MHz under Linux :-( !

Unfortunately... 
> Even if raw CPU speed is better on my Mac (at least) the difference
> is not really significant (an average of 1.2 vs 1.0 for the 386 Linux box).
> 
> How do you explain that a 68040 30 MHz is only slightly quicker
> than a 386 33 MHz ?
I think their CPU takes an advantage of segmentation. Under UN*X I think
we simulate it.
> By the way, I saw the results for a NextStep box with a 68040 (apparently
> running at 25 MHz) and they were very close to those obtained on my machine
> (except for disk io of course). As you would expect it, my Mac was slightly
> faster in raw CPU (maybe by a factor of 30 / 25 ;-) !).
> ... polemic mode on ...
> So, my question is, is Linux really quicker than NetBSD or are Intel
> processors really quicker than Motorola's ones ?

My idea is BSD is a litle bit slow but MORE MORE MORE stable.

For ex. redhat linux with X at P200 16MB ram 32 user looses network
connections,
but a P100 16MB ram with FreeBSD 32 user, the system goes slow but never
loose connections. I don't like linux much. It's for children.I	 don't
like system V also. BSD is nice.  



PS: Can you send me these bechmarks. You can mail to seymen@gul.net.tr

******************************************
	M.Orhun SEYMEN			 *
	e-mail:o52931@sumela.ktu.edu.tr	 *
 		seymen@zigana.ktu.edu.tr *
		seymen@gul.net.tr	 *
******************************************