"A Fundamental Turn Toward Concurrency in Software"

Philippe C. Martin philippecmartin at sbcglobal.net
Sat Jan 8 04:18:04 EST 2005

I remember a _few_ year ago when all specialists (Intel's) included
agreed that the 100MHZ barrier would never be passed - so, at least, we
did get free lunch for a couple of years :-)

I also must add that in my 17 years of realtime/embedded programming,
the problem usually was not the CPU speed - since you know _prior_ to
the design of your software whether your CPU will change context on time
when the hardware IRQ comes in - but rather the fact that all the
software/peripherals had to share the same bus ==> that usually was
where the bottleneck was.

I guess my point is to not focus the performance of a system solely on
its CPU's.



Philippe C. Martin
SnakeCard LLC

More information about the Python-list mailing list