On Mon, Jun 17, 2002 at 08:49:36PM +0200, Frank van der Linden wrote:
> Could you perhaps include actual performance figures (and of course
> how you obtained them)? I'm always very interested in performance
> comparisons, but plain statements like "it's on par with" or
> "FreeBSD is faster" are hard to verify, or indeed makes it
> hard to fix if there is a problem there.
> 
> On Mon, Jun 17, 2002 at 11:31:50AM -0700, Jonathan Stone wrote:
> > Myself, I just want a driver that runs at (or near) wire speed on some
> > of the current-generation (under US$100 and trending down) boards.
> 
> Wire speed is Hard to obtain still. Numbers vary greatly, especially
> with gig-e over copper. If you're doing 50-60 MB/sec over IP, userspace
> to userspace, and using normal framesizes, you're doing alright.
> That's still only half the wire speed. If you want to go higher,
> you'll need jumbo frames, and a faster/wider PCI bus.

FWIW:

At least two vendors manage to get 90MB/sec or more using Tigon-II cards
with modified firmware, even with slow CPUs (e.g. MIPS R10K at 170MHz)
and wide-but-slow (64/33) PCI buses.  This is with *standard* size frames.

The comments at the head of the FreeBSD BCM570x and Intel Gig-E drivers
appear to state that they can deliver similar performance on a modern
machine.  But the experience of SGI and others with old hardware and slow
CPUs does seem to imply that even with small frames, it can be done.

-- 
 Thor Lancelot Simon	                                      tls@rek.tjls.com
   But as he knew no bad language, he had called him all the names of common
 objects that he could think of, and had screamed: "You lamp!  You towel!  You
 plate!" and so on.              --Sigmund Freud