On March 25, 2004 21:04, mspgcc-users-requ...@lists.sourceforge.net wrote: > Today's Topics: > > 1. Some mail problems. (Dima) > 2. - (inbox.ru) > 3. hm.... 3.4 (Dmitry)
The mail problems seem to spread to include the mailing list. The second message is a .jpg image, presumably spam, and the third message, presumably real, is missing. Could you re-post this, please? > --__--__-- > > Message: 1 > From: Dima <d...@radio.rphf.spbstu.ru> > > Concerning array bug: Thanks Rick, fixed. the code is in cvs now. Thanks very much for the quick response. I imagine this was a bear to find, since it was so touchy about the exact code which provoked it. Your fix works well, not only in the artificial test case I sent to you, but also in my original code. > Concerning floating points: > > 1. GCC provides ieee-754 complaint FP library which operates either with 64 or > 32 bits floats. gcc uses 32 bit floats cause of memory limit. 32 bits FP has > 24 mantissa bits. > Check this > http://babbage.cs.qc.edu/courses/cs341/IEEE-754references.html > for more info. This is a wonderful reference, many thanks. I had already found out what I needed after Steve Underwood posted that the compiler was IEEE-754 compliant, since I have a (paper!) copy of the standard. Your reference provides first-rate material on all aspects of floating-point arithmetic. > 2. msp430-libc provides another FP library which optimized for fast > operations, but might be somehow incorrect or not precise. Yet this is much > smaller than gcc's one and also much faster. In the particular application I'm working on, I'm not pushed for time or space, but accuracy is a big deal, so I'll probably stick to the IEEE-754 library. It's great to have the choice. All the best, rick -- Rick Jenkins <r...@hartmantech.com> Hartman Technica http://www.hartmantech.com Phone +1 (403) 230-1987 voice & fax 221 35 Avenue. N.E., Calgary, Alberta, Canada T2E 2K5
