Mark Koek wrote:
> [...]
>
> Quick suggestion after a first glance at the patch: IIRC, 'teq' is
> slightly faster than 'cmp', so where you only test for equality it's
> better to use 'teq'. Maybe I'm wrong.
Not really. It is a simpler operation, but on all present ARM
implementations I know both instructions take exactly 1 cycle,
except when the shift amount is a register.
This can be important for next generation ARMs with longer pipelines,
but so far you can assume all ARM instructions to take only 1 cycle,
except load/store/mul/data processing with shift amount specified in
a register/FP instructions. This is valid for Arm2/250/3/6/7/Strongarm.
Peter Teichmann
--
Email: [EMAIL PROTECTED] WWW: rcswww.urz.tu-dresden.de/~teich-p
_______________________________________________
http://lists.arm.linux.org.uk/mailman/listinfo/linux-arm
