>>>Wasn't Turbo Pascal the real groundbreaker from Borland?
>>
>>Yep - bought it for my CP/M system. Compile and link in one operation. Small enough to fit on a single 240K 8" floppy, with space left over for your programs. So you didn't have to swap disks between your user files and the compiler/linker.
>>
>>A friend had a copy of Microsoft Pascal for CP/M, which came on multiple floppies for both the compile and link steps. You had to swap floppies a bunch of times to compile and link. And it was brutally slow compared to Turbo Pascal.
>
>I was working on a CP/M machine for a year or two, and had Turbo Pascal and Microsoft Cobol. Their relative speeds were about one order of magnitude. And I had to write a sort (yes, SORT) in TP to compensate for the M$Cobol's bug when the indexes would go bananas after 32767 records. I did it in a day and saved the day - to my boss's astonishment. He thought I was halfway between crazy and brave, neither of which were something to be admired in a programmer.
>
>TP knit circles around anything else I had on CP/M.

I started out as a hobbyist, I bought a friend's Jim Ferguson Big Board I (he had upgraded to a BBII). It originally came in kit form, and his girlfriend wire-wrapped it for him. That's still amongst the benchmarks I use when evaluating "acts of love" ;)

It originally was a 2MHz Z80, but on one of his vacations my friend went to one of the early West Coast Computer Faires and I got him to bring me back some parts. A Z80B processor, 64K of 200ns static RAM and a 5MHz crystal to run them. Just dropped them in, everything was stable. For a while I was the fast kid on the block :)

At work we had a weird resonance problem on the production line. It was widely considered intractable, but I figured out a way to brute-force simulate it, and devise countermeasures on my CP/M box using TP. At work I got them to buy TP for the early IBM PC we had. It had bitmapped graphics (Herc card IIRC) so I was able to add a module to graphically display the simulation rather than having to read columns of numbers (my BBI was 80x25 text only). For that application my CP/M machine was slightly faster than the IBM PC; that might have been just its 5MHz clock vs. 4.77MHz of the PC.

The simulation and graphics sold the powers-that-be, and the program got embodied into a dedicated microcontroller and rolled out to the plants nation-wide. The guy who did that work in Central R&D was named "Ken Tosh", nicknamed "Mac" whenever he visited our plant ;)