🚀 go-pugleaf

>Klaus J. Koch heeft geschreven in bericht
<7k26wEf91hB@klaus.mr.he.schule.de>...
>It's not that easy. If you are using a pentium cpu, simply clock
>cycle counting won't help because certain instructions can be
>performed simultaneously. The guys at Borland who write the
>compilers know what code to generate to obtain fastest executables.
It would not help on something like an Inc() instruction. OTOH, if you have
two ways of sorting, e.g. Bubblesort and SelectionSort, it may be a nice
help. It is not a sure way to get differences between methods, but you could
use it as a cheap tool to point out the bottlenecks in your code.

>The other day I spent my time to convert 3 lines of pascal code into
>more than 50 lines of highly optimized assembler code avoiding slow
>commands as loops, jumps, calls, and pushs, and it ended up in
>taking 5 per cent _more_ cpu time than the compiler generated code.
If I'm not mistaken, then Pascal is converted to machine language. Is asm
not a one-on-one construction of machine language? The first thing I learned
about computer is that the fastest way to program is in machine language!

Now, it is assumed that Pascal is generally slower than C. And C would be
slower than asm. If your code is slower than Delphi, than this means that
the Borland Group has some good geeks stached in their basements!

>A hint that hasn't been given yet might be helpful:
How could I miss the obvious?

BTW, I don't (want to) know assembler - I want to be able to read what I've
written..