Subject: Re: Compile speed wars [was Re: FreeBSD Bus DMA]
To: None <tech-kern@NetBSD.ORG>
From: John S. Dyson <firstname.lastname@example.org>
Date: 06/12/1998 12:10:45
Manuel Bouyer said:
> Just a data point: my system at work (PPro 200, 64Mb RAM, IBM DORS 32160W
> on an 2940UW) has a NetBSD 1.3/Linux 2.0.30 multiboot.
> Compiling a linux kernel, using linux binaries
> (chroot /emul/linux under NetBSD) took exactly the same time (8 mn 02 s)
> under NetBSD or linux.
> One could argue that the 2 systems were not set up the same way: under
> NetBSD the FS were not mounted async :)
> Now, for these kind of benchmarck to be valid, we should repeat it on
> different hardware, to differentiate 'general system speed' from
> 'this scsi driver speed'.
Your results only show that there are times that NetBSD
can perform okay. Now, what about the situation that
it doesn't? Should NetBSD perform as well as FreeBSD or
Linux, in your situation, or should it really work well
in a significant production senerio?
Of course, I am not describing a situation that is posed
on "different" hardware.
Maybe, by Wed or Thu, I'll be able to post some info
about the differences. However, it is lunacy that
certain latency improvements make for a significantly faster
This is valuable work in the FreeBSD kernel that is
available for inspection, however, there is a long way
to go for NetBSD to be able to effectively utilize
some of that work.
John | Never try to teach a pig to sing,
email@example.com | it just makes you look stupid,
firstname.lastname@example.org | and it irritates the pig.