David Hess (davidwhess.delete@this.gmail.com) on January 9, 2020 8:27 am wrote:
> Anon (no.delete@this.spam.com) on January 9, 2020 4:24 am wrote:
> >
> > Intel presence in personal computers were almost 0 before IBM PC, market were dominated
> > by 6502 with Z80 as second, and I don't think translators worked for Z80.
>
> The translators could not have worked for Z80 simply because of the significantly expanded register and
> instruction set. But the only applications I saw which took advantage of the Z80 were custom ones.


I am not aware of Z80 to 8086 assembly language translators, but it would not have been difficult because the extra 8086 registers included the additional registers of Z80 and the 8086 instruction set also included most additional instructions of Z80.

The few extra instructions of Z80 that were not also provided by 8086, e.g. the bit operations, could be easily synthesized with the available 8086 instructions, in the worst case with a sequence of 2 or 3 8086 instructions. Still, most Z80 instructions could be mapped 1-to-1 during the translation to 8086.

I doubt that the fact that most Z80 improvements were also present in 8086 was a coincidence. It was most likely intentional, because until the early nineties, when Intel finally became comfortably ahead of its competitors, most new features added to every generation of Intel CPUs were present earlier in competing CPUs from Motorola or others, but Intel was quick to add them too.



>
> Intel's presence was only decreasing because CP/M systems were increasingly using the Z80
> in place of the 8080/8085 for increased performance, greater ease of use, and lower cost.
> These CP/M systems were the business computers of the time and available from so many sources
> that there was an entire "clone" industry going on running common applications.
>
> > Well... The importance of compatibility was not the same for Intel and IBM...
> >
> > ...
> >
> > For Intel customers which used 8085 (not in PCs) binary compatibility may have been important,
> > anyway, the number of 8088 sold for IBM PCs made all previous customers kind of irrelevant.
>
> The source code compatibility was so that CP/M applications could be easily ported to MS-DOS
> and CP/M-86. MS-DOS and CP/M-86 included backward compatibility features for exactly this.
>
> IBM did not have to chose MS-DOS, and CP/M-86 which was available as well, for the
> IBM-PC however there were *other* 8088/8086 systems available at the same time which
> ran these operating systems and benefited from the ported CP/M applications.
>


For most programs that I have used initially on MS-DOS, I had previously used a version of them on CP/M. For example I had used a large number of compilers and interpreters from Microsoft and also WordStar and other programs that I do not remember now, in their CP/M variants, before using them as MS-DOS programs.

So I have no doubt that porting programs from CP/M on 8080/Z80 to MS-DOS on 8088/8086 was very important in the beginning and that many might have used the assembly language translators, at least for an initial version.