"Mark Brinicombe" writes:
>>(String copy is a bit slow though)
>
>Not surprising if you look at the C code that does it ;-) Things should
>improve when I get ASM versions of the string functions in libc.

Ah.  I'm still waiting for the bcopy() stats Mark said he'd produce for me.
Since I know Mark's usually very busy, and I have plenty of other tasks to
keep me occupied, I've not pestered him about them.

So far I have written faster versions of some of the more common string.h
functions.  I've also chopped a few cycles out of the strlen() I wrote a
while back.  I can send these to someone if they want to integrate them into
libc.  Currently they're in objasm format, since I've failed to find a
RISC OS version of gas.

The stats I wanted are frequencies of (length,from&3,to&3) triplets in bcopy
calls.  If you (Mark) have too much to do, let me know and I'll try to get
around to installing RiscBSD so I can generate my own.  Anyone in Cambridge
willing to lend me the sets on floppy or CD?

Cheers,
Olly