[ On Friday, April 6, 2001 at 11:26:27 (+0200), "Jaromír" Dolecek wrote: ]
> Subject: Faster pipes from FreeBSD
>
> I've made a stab at porting the FreeBSD sys_pipe() implementation to
> NetBSD (that implementation was written by John S. Dyson). The
> basics are finished, the thing works quite nicely and indeed about
> 1.5-7x faster than out current pipes (wow!). I've tested it very
> lighly on i386 so far, my simple pipe push program gave these results:

VERY COOL!  I started doing this as well but have been side tracked with
trying to build a new release for a customer and upgrade their machines.

I am very interested in trying your changes!  As soon as I have my
second development machine upgraded I'll build a kernel with them and
give them a good workout (and maybe on the sparc too).

This should breath an enormous amount of life into any older machine
that's got adequate amounts of RAM (eg. my Sun 3/280 w/128MB).  (I
wonder what that MicroVAX-II and the VAXserver 3000 I was given the
other day have for RAM in them!)

> The system with this boots to multiuser and seems to work OK,
> so hopefully there is no problem here.

Have you done a "make build" or similar with "COPTS+=-pipe"?  That would
be a major workout and sanity check!

> The code does not yet do the direct memory write stuff; I hope it
> would be possible to do it with UVM, since that should yield another
> 30% performance increase, at least according to the comments
> for FreeBSD VM.

Yes this would be very nice to have too!

-- 
							Greg A. Woods

+1 416 218-0098      VE3TCP      <gwoods@acm.org>     <woods@robohack.ca>
Planix, Inc. <woods@planix.com>;   Secrets of the Weird <woods@weird.com>