My shameless self-promotion knows no bounds
Alan Burlison
My shameless self-promotion knows no bounds
Alan Burlison
Completely awesome article, Alan!
I use fixed point in AVR quite a lot for calculating things like bezier
curves and PID values. The only problem I have when using fixed point is
remembering what units I used for the variable.
was that a 6:10 or a 4:12 ?
I need to review my fixed point code, after reading that there is every
chance is it is slower than the compiler’s floating point math… /grumble
grumble.
Also floating point code eats up the ROM space in a micro.
well done, superb work.
BazOn Sun, Jul 15, 2012 at 9:00 PM, Alan Burlison alan.burlison@gmail.comwrote:
My shameless self-promotion knows no bounds
http://hackaday.com/2012/07/15/a-detailed-explanation-on-speeding-up-avr-division
–
Alan Burlison–
You received this message because you are subscribed to the Google Groups
"HAC:Manchester" group.
To post to this group, send an email to hacman@googlegroups.com.
To unsubscribe from this group, send email to
hacman+unsubscribe@googlegroups.com.
For more options, visit this group at
http://groups.google.com/group/hacman?hl=en-GB.
Completely awesome article, Alan!
Thanks
straight
I use fixed point in AVR quite a lot for calculating things like bezier
curves and PID values. The only problem I have when using fixed point is
remembering what units I used for the variable.was that a 6:10 or a 4:12 ?
Yeah, found that myself. Plus keeping the simultaneous scaling and
division-by-shifts straight did make my head spin a bit
I need to review my fixed point code, after reading that there is every
chance is it is slower than the compiler’s floating point math… /grumble
grumble.
Yes, those divisions are a killer.
Also floating point code eats up the ROM space in a micro.
Yes, quite a bit Mind you, integer division & multiplication uses
library routines as well, although they aren’t as huge as the FP ones.
well done, superb work.
Ta
Alan Burlison