I uploaded in the files, directory "avr-gcc" two routines to handle
float numbers quicker that the library does.
(Do not report errors if there are, but just repair silently)

(You need to be a member of avrclub -- send a blank email to avrclub-subscribe@yahoogroups.com )