>https://www.theregister.com/2024/01/04/niklaus_wirth_obituary/?td=rt-3a
>Some of my earliest programming was using Turbo Pascal on both a Z80-based CP/M homebrew machine and an original IBM PC 5150.
>Interestingly, the performance was nearly identical on both platforms. My CP/M box had a hot rod 5Mhz Z80B vs. the stock 5150's 4.77MHz 8088.

Yupp, Turbo Pascal was a beast. My entry into bytes happened on a C64
(bought as I was a lousy typist), but programming started soon afterwards
as I went crazy trying to calc primitive statistics with a programmable calculator -
I rewrote some basic tests and a regression package in BASIC.

Soon afterwards I was asked to create a "profesional" program,
for which UCSD and a specific C64 Pascal were to slow compiling.

"Tested" a Compaq for a long weekend and the program was done,
sold and used the money for a memory upgraded Atari ST,
which allowed me to code in linear adress space - but with traditional
compilers from Prospero in Pascal (with the benefit to link FORTRAN).

Moved to OS/2 with Prospero on 80386, ported there later to JPI Modula2
to allow linking of C, then to Extended DOS, which allowed FORTRAN again,
but already had bought the C version of the Numerical Recipies,
which I coded under NT4 with JPI (including a SQL interpreter library)

JPI was a (mostly) memory based 2 pass compiler, but the blistering speed
of compilation able with early Turbo single pass was never reached -
programs were larger but compiler ran on 486, then Pentium, then K6-3 -
which was a beast running vfp6 (floating point not necessary)
with 2 Adaptec and 7 SCSI HD with nearly 16 GB.

If compiled programs fit into a single bank, compilation into and
running from that bank felt like working with an interpreter with
very slow start up but crazy execution speed
(if all errors were eliminated - but with Turbo, the lousy typist
could weed out many when compiling, not later in runtime ).

Ported the a few dbMan dbase compatible routines to calc
mostly taxes later to Fox - and with Fox 2 linked
my "multithreaded" Modula2 CoRoutines to Fox C runtime.
Which funneled me into "optimizing Foxpro"...

I owe a lot to the basic stuff in languages Wirth created -
my early versions of sorting arrays of pointers to records in memory
were helped a lot by the bounds checking immanent in all his stuff.
CoRoutines under MS-Dos in Modula2 pointed me early in a direction
nobody thought about before Threads or Timers were available.

Sold at last ported to C as Wirth was not "industry compatible".