imago@kabel065011.kabel.utwente.nl said:
> One easy (big?) optimalisation would be to compile the kernel and
> other stuff with -O3 since that inlines small functions and such and
> thus reduces the branch overhead significantly.... 

Except that calls through pointers can never be inlined (since, by 
definition, if we have to call through a pointer it's because we can't be 
sure where we will be going).  Calls to fixed functions can be inlined, 
but then for these we can predict the call (though not the return) and 
eliminate most of the overhead of the call.

Although by comparison with ARM7, the 32K Icache may seem huge (hell, my 
first BBC micro had 32K RAM *total*), lets not waste it by unnecessarily 
inlining things.

R.