PE> Look at the generated object code to see whether the code
PE> actually generates worse assembler.

BL> I've been doing that, adding up clock ticks as well as running
BL> profiler tests, and Pascal impresses the shit out of me. It
BL> varies a bit, but 20% off optimum ASM is about the result for
BL> small functions. That's *very* good.

PE> Show me the comparison to the C code, and I'll show you what
PE> you did wrong in the C programming. 

BL> Eh? I was comparing Pascal and ASM, looking at the runtime library

You were saying that Pascal was better than C, before.  I told you to
compare the object code (which would show you to be wrong).

BL> ASM code and seeing how much faster it was than the stuff I could
BL> werite in Pascal itself. It varies a bit (especially with strings when
BL> ASM uses those instructions I can't remember offhand), but on average
BL> Pascal is only a little slower than ASM. I haven't bothered to check C
BL> (seeign it won't lead me anywhere) but from the tests Frank did with
BL> your Watcom I'd expect C to be nearly as good as ASM.

It's certainly close enough for 99% of the program.  BFN.  Paul. 
@EOT:

---