23 Mar
2004
23 Mar
'04
8:40 a.m.
From: "ben franchuk"
<bfranchuk(a)jetnet.ab.ca>
Antonio Carlini wrote:
Hi
We make CPU chips and I can tell you that they are so complex
that I doubt that we or anyone else in the world has ever tested
one completely enough to say that there isn't still a bug in
it someplace. It isn't that we and our customers don't try.
We throw everything we can at them. It is just that the number
of combination of events is so astronomical that it truly
can't be full tested for bugs within the time of the remaining
life of the universe. We do test them to see that they act
as designed but that doesn't expose bugs. That is a completely
different problem.
Dwight
I think you've missed what AXE does.
I give you a spec for the chip (in this case DEC STD 032, the
VAX Architetcure Manual). You build the chip.
I did miss that point that it is a pre-production rather than
a post-production test.
I want to know if you've built a chip that
meets
the spec or just one that "almost" meets the spec.
Don't look at me , my CPU's never meet the spec,
but then again I am not doing a commerical product. Like who would invest in a NEW LS-TTL computer?
I tend to be over 20 ns my clock period or way too
many chips but I have fun unlike people in the real
world desiging CPU's.
CPU diags are there to check that the chip was
stamped properly
at the factory and still works after you spurt coke all over it.
Umm what about the cold pizza too?
That's not AXE's job.
Thank you for
the insight.
Ben.