On Thu, 7 Aug 1997 04:22:57 -0300 avrbasic@USA.NET writes:
>At 03:35 PM 7/8/97 +0100, you wrote:
>>HI guys...
>>
>>I'm working on a project with a 16C554, using the RB0 external
>interrupt to
>
>where did you get on hold of 16C554 with Interrupts?

The 554 is a 14-bit machine: 8-level stack, interrupts, in/out on A4, 80
bytes RAM in one page, the whole works that would come in say a '61.
Microchip insists on a rather muddy numbering scheme.

>

>there was some discussion about it, it looks like all commands that
>have the PORTB adress inside the command word do a portb read (even
>if command shouldnt) some RETLW could access portb and something like.
>
>It maybe that this feature of 'ghost accesses' to register file has
>been fixed by Microchip in some newer chips. PIC16C84 I think still
>has some commands that do a "ghost read".

Was there ever any clear indication of what chips and when this was done?
 I assume Mchip's party line on this (for all chips) is to use the RBIC
interrupt only to wake up from sleep to avoid this problem.

>>ps. I note that the Atmel AT90S8515 RISC processor has "120 powerful
>>instructions".

>its the way of counting. and marketing. If RISC is hot, we call it
>RISC.
>ST62 has real few commands but nobody calls it RISC.
>
>AVR "120" instructions is "high count" it is possible to count much
>less instructions than 120. just the way of counting.
>
>in 120 are counted 20 bit_branches on conditon,
>I would count then as one, and so on.

There are about 60 uniquely different AVR instructions.  A lot of this
results from breaking the AVR's data space up into registers, I/O, and
SRAM, then making specialized instructions for each space.