Now and then people seems to face the need of faster floating-point
computing on integer hardware, such as pxa.
There are of course more elaborated (and complex) solutions, but I wish
to point to ARM's application note 33, document name DAI0033A, that
gives hints about fixed point arithmetic and how it compares to floating
I found it to be a nice introductory reading, since it also comes with
code examples, both as C macros and assembler code (arm & thumb).
I'm not sure if this application note is easily found on the ARM's web
site or the internet as a whole; I got it from ARM on the technical
publications cdrom, which can be ordered from the site for free. There
are also other application notes I suspect are worth reading but I
couldn't find the time (I'll get there eventually).
Get latest updates about Open Source Projects, Conferences and News.