> >Or look at the PDP11's Unibus (or Q-bus). It's elegant. It works. And
> >there are no IRQ or DMA conflicts. The PC bus is a kludge from start to
> >finish. If you read the PC/AT TechRef (I have), you'll find there is an
> >official way to share IRQ lines in a PC/AT - which IBM then ignored when
> >writing the BIOS. And that's what became the standard.
>
> I can expect a kludge from a hacker... because even if it's weird, it works
> when it's not supposed to. That takes a lot of imagination.
Hmmm... Working when it's not supposed to is not a kludge. It's plain good
design IMHO. In other words, a good design should do the right thing under
all circumstances.
>
> A kludge of this magnitude from the number of people assigned to design the
> expensive IBM machines of that day is a crime!
The point is that the IBM ISA bus doesn't _ever_ do the right thing. If
there were 2 ways to do something, then they picked the wrong one. It's
amazingly bad.
[CoCo3]
> Thumbs up everywhere! I have one (and still use it at least weekly) and
> despite it's (now I'm getting technical for the sake of conversation) 1.78
I liked the design so much that I shipped one over here from the States. I
ordered the service manual first (One nice thing about Tandy machines was
that the service manuals were always available), read it, and ordered the
machine.
Tandy never sold the CoCo 3 in the UK, for all they'd designed a PAL video
version, according to the manual
> Mhz clock speed, I pitted it against a True-blue IBM 286-10 AT machine in
Which true-blue machine was a 10MHz 80286?
> several benchmarks (which I wrote) which included number & string sorting,
> prime number creation and getting the factors of a number and..... WON!
> People don't realize that the Intel x86 chips are *highly* inefficient,
> especially in their addressing and branching instructions, which gave my
Indeed. I once read the data sheets on (I think) the 80486, and was amazed
by the fact that some quite simple instructions took 60+ clock cycles. It
appeared that either Intel couldn't design a CPU, or were shoving the
clock rate up (marketing?) but taking so many cycles to do anything that
the real speed was very low.
> Roger Merchberger | Everyone complained to me to change my .sig,
--
-tony
ard12_at_eng.cam.ac.uk
The gates in my computer are AND,OR and NOT, not Bill
Received on Fri Apr 11 1997 - 09:43:21 BST