ajp166 at bellatlantic.net
Thu Apr 19 17:00:12 CDT 2007
>Subject: Re: CP/M survey
> From: "Chuck Guzis" <cclist at sydex.com>
> Date: Thu, 19 Apr 2007 09:25:46 -0700
> To: "General Discussion: On-Topic and Off-Topic Posts" <cctalk at classiccmp.org>
>On 19 Apr 2007 at 9:48, Allison wrote:
>> >Maybe, but it couldn't run JRT Pascal. AFAIK, the only commercial
>> >product that ever used the bizarre coding sequence:
>> > LXI SP, PROC-1
>> > CALL PROC
>> JRT Pascal was Z80 code if memory serves. But LXI SP, value is
>> valid as the arithmetic is done at compile time not execution.
>Nope--8080. I've still got the 1.0 disk. One can precede this
>sequence with (and I think that JRT did):
I have V3 and never had a problem, guess they fixed it.
> LXI H,0
> DAD SP
Only way to get the SP on 8080. Liked the Z80 because they fixed that.
>to get the old SP value into the HL register pair prior to the LXI
>SP. And it is a V20 bug--I remember calling the NEC technical guy
>in Natick and getting about 10 words into the report and having him
>say "JRT Pascal, right?". If anyone's interested in the V20 errata,
>I've still got the stuff.
Was that Charlie or Ted? I'd like to see the report or erata mostly
since it would fit nicely in my NEC file.
>JRT was one of the earlier attempts at "virtual" 8080 code; it
>swapped procedures from floppy. Of course, it was miserably slow,
>but at something like $30 for a Pascal, it looked like a great deal.
>Of course it was buggy as the dickens. I think the oddball calling
>sequence was to keep the stack adjacent to the procedure for
>subsequent swapping, rather than having to deal with a single stack
>that might well overflow without special handling routines, given the
>"virtual" nature of JRT Pascal.
Doing anything on 8080 that was virtual was slow. V3 was still $30
and slow but it did work.
>> In the end running an 8080 (V20)
>> when I have Z80 or even fast(6mhz HmosII) 8085s is sort of
>> less than interesting.
>Maybe, but you use what you have at your disposal, even if it is an
>8080. And most professional apps for CP/M used the 8080 instruction
>set initially--only later did a bunch of Z80-specific (e.g. ZCPR)
>code come out. I never could understand this--in general, little to
>be gained in speed by using Z80 codes.
By time the V20 hit the street I was running hand upd780s at 8mhz
and had at least three s100 crates going.
>FWIW, I still use 22NICE on Win2K. It nicely integrates old CP/M
>apps into the Windows environment without having to create virtual
>disks or such stuff, so using apps under emulation is no harder than
>using native ones.
Still have it and use it, interesting tool.
More information about the cctalk