I've played around with Rado's patches, and optimised them by by-passing
the gpiosysfs interface and just modifying IO ports directly.  Seems to
give a four-fold speed-up.  I won't even bother posting the link, 
since this is now well-and-truly RDC platform specific.

However, since I'm that way inclined I'm unlikely to stop there, 
and may go on to coding the spi loop in x86 assembler, just for 
kicks.  Joerg is sure to never let that anywhere near the avrdude 
SVN (rightly so), but this prompted me to wonder about something.

Has anyone considered a dlopen interface to the programmer code?  This
would be a nice way of allowing all sorts of extensions without bloating
the core code.  It doesn't seem too hard to implement.  It may even be
a better way of adding the GPIO SYSFS stuff.

regards,
Biff.







_______________________________________________
avrdude-dev mailing list
[email protected]
http://lists.nongnu.org/mailman/listinfo/avrdude-dev

Reply via email to