Michael Bain wrote:
> hmmm, I´m not intimately knowledgable about the ins-andouts of newton
> programming in C++, but would it be possible to compile all the hard/64
> bit stuff in a library using a "real" C++ compiler, then statically link
> them in with the NCT code or do newton pkg's use some sort of wierd
> object format?
Short answer: yes. Long answer: it really hurts ;) The problem is that
the NCT uses the old AOF object format for object files. It is not
supported very well by other compilers such as gcc. One way to get
around that is to generate assembler files which then can be compiled
using the NCT assembler (if you want to use gcc, the target for
configuring it for that is "arm-semi-aof"). I tried that but something
went wrong :( I haven't had the time to look into that yet...
There is supposedly also a patched version of gcc 2.95 that produces AOF
files directly. All this won't work for C++ though since the name
mangling differs. The last resort is to get ARM's SDT for about $5000 ;)
Maybe I'll look into their evaluation version.
What I'm going to try next is Eric's suggestion for a more accurate 64
bit multiplication, and also a subroutine approach where the
multiplication is coded in assembler and called externally. This makes
optimization a bit more difficult for the compiler because I haven't
found a way yet to make the assembly code inline (gcc can do that).
Eckhart
PS: for anybody interested, the code in question is in the original MAD
source, file libmad/fixed.h
--
-- eck_at_40hz.org -- www.40hz.org --
--
This is the Newtontalk mailinglist - http://www.newtontalk.net
To unsubscribe or manage: visit the above link or
mailto:newtontalk-request_at_newtontalk.net?Subject=unsubscribe
This archive was generated by hypermail 2.1.2 : Fri Feb 01 2002 - 16:03:15 EST