Date   : Wed, 04 Apr 1990 21:09:49 GMT
From   : mojo!chuck@mimsy.umd.edu (Chuck Harris)
Subject: Non-Intel Bashing...

In article <9004040702.AA08985@ucbvax.Berkeley.EDU> JSHIN@HAMPVMS.BITNET
("No, me?") writes:
>
>Hmmmm.....
>
>This is a serious matter, you see.
>
>Personally, I think everything that Intel has ever come up with tend to
>be rather demented; especially the iAPX86 family.  The segmentation system
- I won't even touch this!
>is totally absurd.  The Queues on the lower processors are totally
- Think 1979.  Transistors on VLSI chips were rather expensive and limited in
- number. In 1980 the 68000 came out it didn't have any queues at all!

>mundane.  The processing time is totally outrageous.  The architecture
>(the way they have the registers set up) makes absolutely no sense.
>
- I assume that you are talking about the use of CX as a count register
- in loops, and the implied use of ES and DS in the movs instruction.
- The reason these instructions use "implied" registers is make programs
- using them use less memory.  An admirable thing to do in 1979 (and 1990 too)

>Aside from the compatibility, there is no reason why ANYONE should be
>*%&('ing using any of Intel's higher-level processors.  Intel-compatible
>op. codes are assembler programmers' worst nightmare - they are as
>far away from natural language as humanly possible.
>
- BURP! I feel better too.
- Binary Numeric OP codes are rather traditional among computers.
- I mean seriously, do you program in machine code? Opcodes are the numbers
- the machine uses to encode its different instructions.
- Do you mean the assembler's Mnemonics?  If you do, mnemonics like MOV, JMP
- INC, NOP had been with computers a long time before microprocessors existed.

>it out of production.  8086 makes as much sense as a pile of rocks
>in the middle of Interstate 91.
- BURP!
>level), I think there is a reason why:
>   -people abandoned 8080 for Z80
>   -people abandoned 8086 for 68000
>   -people never thought of using 8086 where they are using 32000 and Z80,000
>    these days.
>(oops; I forgot a "but" in the paragraph above.)
- This needs some comment.  In the CP/M programs that I have seen very few make
- use of the z80 extended instruction set.  Most people wanted their programs
- to work on as many machines as possible be they 8080, 8085, or z80.
-
- IMHO the reason the z80 was used in so many of the CPM machines had NOTHING
- to do with its instruction set.  It was used because of the hardware
- simplification it provided by giving DRAM support for free.  The engineers
- didn't have to do anything to make DRAM work.  Which was a good thing,
- because most engineers who tried to make DRAM controllers back then botched
- them but good. (think MITS ALTAIR)
- As to abandoning the 80x86 for the 68000, I can program each efficiently, yet
- I prefer to use the 80x86 family.  Why, because the hardware is simpler to
- design around, and I can fit very substantial programs in very small memory
- space.  I can't with the 68000 family.

>Someone should come up with the guts and money to speed up Z80;
- already been done (well not quite that fast) Think Hitachi.
- But seriously, speeding up the z80 won't scare Intel or anyone else.  What
- most people are looking for today is hardware that can handle LARGE complex
- programs with wide address spaces.  Both the 80386 and the 68030 do this
- very well.  A fast z80 couldn't even enter the race.

>with current AS-TTL technology, we can pull it up to 100MHz!!!  Look
>out Intel (Intel does, however, make some mean peripheral chips.)!
>
>     -John

Mild Ruminitions from,

               Chuck Harris
               C.F. Harris - Consulting