PC/104

Christopher Browne cbbrowne-Re5JQEeQqe8AvxtiuMwx3w at public.gmane.org
Wed Jul 18 19:05:04 UTC 2007


On 7/18/07, Lennart Sorensen <lsorense-1wCw9BSqJbv44Nm34jS7GywD8/FfD2ys at public.gmane.org> wrote:
> On Wed, Jul 18, 2007 at 06:10:04PM +0200, Christopher Friedt wrote:
> > Power is in the eye of the beholder... if I can twist such a saying to
> > illustrate my point here.
>
> You can do a lot with an 8 bit micro, but you can do a lot more
> (although probably with more development work too) with a modern 32bit
> embedded processor.

What is plenty interesting (that this demonstrates) is that having a
*LITTLE* bit of resources in the right places can be fabulously
useful.

Back in the dark ages, when we used to use modems to call ISPs and
BBSes to communicate, modem performance kept growing, to the point to
which it started to actually need to consume a fair bit of CPU to
manage it.  The "mystical, magical improvement" took place when we
upgraded to use 16550 UARTs that had 16 bytes of buffer.  That 16
bytes, in the right spot, caused downright spectacular improvements in
the stability of modem-based IP connections on busy computers.

Somewhat similarly, it was a Big Deal for disk performance with NSC
53c8xx SCSI host adaptors when they started having 256 bytes of tagged
command queue.

Surprisingly small buffers, if introduced in the right places, can
result in improvements wildly out of proportion with expectations.
-- 
http://linuxfinances.info/info/linuxdistributions.html
"...  memory leaks  are  quite acceptable  in  many applications  ..."
(Bjarne Stroustrup, The Design and Evolution of C++, page 220)
--
The Toronto Linux Users Group.      Meetings: http://gtalug.org/
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://gtalug.org/wiki/Mailing_lists





More information about the Legacy mailing list