6 Aug
2006
6 Aug
'06
3:36 p.m.
On 8/6/2006 at 6:10 PM Ray Arachelian wrote:
When you attempt, however, to build another chip, the
8088 and still
keep compatibility with the 8080, that's where you're starting to make
the wrong turn. Now, the 8086 itself, is also reasonably ok, but then,
you get into weird issues where you can only use some registers for some
operations, but not others.
What drove home the difference of the 8086 and 8080 to me was the offer by
the local sales office to translate some 8080 code to 8086, with the
promise of being (a) smaller and (b) faster. After crashing the
translation utility for about 2 weeks, we finally got a good automated
translation that refused to give the correct answers--and it was over twice
the size of the 8080 code and a bit slower. We decided then that automated
translation wasn't reliable enough to give us something useful. We recoded
entire products. I can't say that the code was smaller, but it did run
faster (we were comparing 5 MHz 8085 vs. 5 MHz 8086).
Sorcim maintained the 8086 edition of SuperCalc as big hunks translated
8080 code, mixed with native 8086 code for critical sections, but they had
their own translator and development tools--and some sections were also
written in Pascal/M.
All in all, the 8086 might have been better off with an architectural
overhaul. I thought it genius that the 68K had 32-bit registers.
Cheers,