Holger Veit <holger.veit at ais.fraunhofer.de> wrote:
These times are gone - assembler is irrelevant for larger projects, and
controller have become large enough to allow wasting space and tolerate
inefficient code.
Some controllers, including the rather lousy 8051 series, still persist,
but mainly because "system designers" have used and understood them for
20 years now and it is too expensive to train them another horse.
Holger
--------------------------------
I have to quibble a bit about this. "System Designers" have to design to
cost goals, as well as schedules. Most of them that I have worked with
would love to change and have the challenge of a new processor. Training is
not a factor.
It's the pointy hair managers that want the cheapest processor they can get.
And using an old war horse means not having to buy new development tools,
SDKs, etc. Plus there is a huge pool of existing applications and coding to
draw on. Moving to a new processor can add 18 months extra on a typical
program. That's a price few companies can afford.
And, of course, the leverage of buying bigger quantities of the old device
can lead to sizeable cost reductions.
So old core processors tend become embedded (pun intended) within companies.
It takes a significant new bauble to break through. The last major change I
saw in embedded processors was the ARM chip sets. They were slow to get
accepted but now are everywhere. Probably another 10-15 years of life
there. And of course the PIC devices are so cheap that they HAD to be
utilized. I don't see anything out there now that would be so superior as
to be worth the cost of change.
Billy