Mail Archives: djgpp/1997/09/17/07:15:52
Herman Schoenfeld <you AT somehost DOT somedomain> wrote in article
<341f7974 DOT 0 AT 139 DOT 134 DOT 5 DOT 33>...
> In article <01bcc27c$add7b720$2b40cbc2 AT russnt>, >
russ AT algorithm DOT demon DOT co DOT uk
> says...
[...]
> >I think that was the point. Terminal Velocity / Fury 3 aren't likely to
be
> >significantly different, are they? The inner loops will still be
assembly,
> >so the compiler will make sod all difference.
> >
>
> No russ. The compiler will make a drastic difference. If not then the
games
> should be equal in performance. Even when the watcom version tested on a
> machine 5 times slower than the other test machines it still manages to
> outperform fury 3!
HOW IS THIS THE COMPILER'S FAULT? (Yes, I do mean that as shouting).
Do you have the faintest idea of how games are written? The inner loops,
which would take about 90% of the CPU time are likely to be written in
assembly. The compiler cannot affect the speed of this (since very few even
attempt to optimise assembly - end even then, only inline stuff).
Therefore,
the compiler can only affect the speed of 10% (by time) of the game - and
obviously, that won't make a big difference.
---
Russ
- Raw text -