Re: Avatar

[toddpw@cco.caltech.edu (Todd P. Whitesel)]

ltchean@iss.nus.sg (Lim Thye Chean) writes:

>The question is: Is it really IMPOSSIBLE to write a fast, optimized C compiler
>for 65816? I hope NOT! Where are all the Apple II wizkids that can solve the
>problem instead of saying "it can't be done"?

It is not impossible. It just takes more work than most people realize -- but
this is true for ANY chip; the 65xx series is the only one that did not have
solid manufacturer backing for HLL development. Orca/C 2.0 (for all its bugs)
is amazingly better, and at this point I think the real limitation is Mike
Westerfield's personal view on how 65816 compilers should work.

Where are all the wizkids? They've got well-paying jobs on compilers for newer
machines and are busy enough already. Writing compilers is a lot tougher than
writing games or applications, if only because the compiler has a _huge_ list
of rules that it has to follow & uphold or else it will be buggy as hell. This
includes the language definition (plus extensions) as well as the machine and
O/S hardware & software coding and usage rules.

Right now I see Orca/C and maybe Sanyo if they decide to run with their 65816
derivative.

>It looks like it is THE computer for me. I don't agree that Avatar will
>definitely fail. I think there is a high chance for it to succeed. It might
>be a computer similar to NeXT, not the mainstream, but many people still 
>buy it.

Note: NeXT has announced that they are going to stop making hardware and stick
to being a software company from now on. A wise decision, since they have
already ported NeXTStep to the 486 and it has much more potential than NeXT's
hardware had (this is not a swipe on the 68040, but on NeXT's system design,
which was too slick for its own good in too many ways).

>I think a Super fast Apple IIGS-compatible which has 1024 x 768 resolution,
>4 MB standard RAM, 16 bit sound, disk coprocessor, and a completely new
>architecture compare to other computer is very exciting, and should appeal to
>many Apple II users. Instead of owning a old-technology computer, they can
>have a machine they can be really proud of.

True, but that's about the extent of the potential market. Unless the machine
offers something that is unique _and_ desirable to modern users, it isn't
going to win over anyone new.

What I'd like to see out of Avatar is full exploitation of the coprocessor
design, in effect creating a cheap multiprocessor machine with response time
that whomps the current crop of monolithic workstations into the ground, even
if its raw computation specs aren't impressive at all. 90% of what I do on my
sparc at work is either graphic-intensive or text-intensive, and the
interactive response on most modern workstations still leaves something to
be desired.

Todd Whitesel
toddpw @ cco.caltech.edu