On Thu, Sep 07, 2006 at 08:46:16AM -0600, Jack J. Woehr wrote:
> On Sep 7, 2006, at 4:13 AM, Sander van Dijk wrote:
> >Optimization is far overrated,
>
> About a decade ago I was working with a really good older programmer
> named Frank on a huge realtime system he'd been coding for 18 years.
>
> One of our junior programmers who was helping him proudly optimized a
> loop.
>
> Frank looked at the code change and said, "Great! Now if the system is
> running 24 hours a day, you'll save about one second a year in execution
> time. How long did it take you to code that?"
Wise words. My OS professor (he's around 70 years already) told
me about his time at Munich in the early 60s when they developed
a time-sharing operating system for some Telefunken computer
(yea they did it at the same time as MULTICS and later UNIX
evolved in US), that they spent around 20 men years to optimize
the system (which never got finished, similiar disease as with
MULTICS, - well the main architect died in a car accident which is
the main reason for dropping the system by that time).
But they learned from this development (and Moore's Law) that
they spent 18.5 men years too much, because if they had waited
only 15 months, the hardware would have been fast enough, that
the system would have performed in the same speed as 20 men
years of useless optimization achieved. And they couldn't win
the challenge against Moore by that time.
Regards,
-- Anselm R. Garbe ><>< www.ebrag.de ><>< GPG key: 0D73F361Received on Thu Sep 07 2006 - 17:40:17 UTC
This archive was generated by hypermail 2.2.0 : Sun Jul 13 2008 - 14:31:08 UTC