Follow-up Comment #4, patch #6649 (project avr-libc):
Thanks, Ruud!
This is fine improvements.
I have make a few changes (sqrt.S is attached):
1. low() is replaced to lo8() (to avoid compile error).
2. Optimize to size variant is slightly modifed to save 4 code words
(look, please).
3. __OptimizeSize__ is replaced to OPTIMIZE_SPEED, as last
is used in Avr-libc a long time.
4. Others are cosmetic: roll lines to 80 symbols, remove assembler
style comments from CPP statements.
In result (avr4, clocks are average for sqrt-500 test suite):
Optimize for size: 234 bytes, 541 clocks.
Optimize for speed: 252 bytes, 474 clocks.
Older versions:
Avr-libc 1.6: 276 bytes, 706 clocks.
Avr-libc 1.4: 606 bytes, 2354 clocks.
Testing in progress...
(file #16910)
_______________________________________________________
Additional Item Attachment:
File name: sqrt.S Size:5 KB
_______________________________________________________
Reply to this item at:
<http://savannah.nongnu.org/patch/?6649>
_______________________________________________
Message sent via/by Savannah
http://savannah.nongnu.org/
_______________________________________________
AVR-libc-dev mailing list
[email protected]
http://lists.nongnu.org/mailman/listinfo/avr-libc-dev
