'Newbie in deep pooh re C17 and 32 bit maths'
|I've got that awful feeling that goes with throwing a week's wages out of
I'm developing a product which requires 32-bit floating point arithmetic.
For some reason I got the impression that Microchip's C17 'C' compiler
supported floating point math, and yesterday blew 300 UK pounds of my own
money on it.
It arrived today, and I was horrified to see, on page 38 of the Compiler
User's Guide that none of the 32 bit data types (long, unsigned long, float
and double) are supported! Aaaaaaarrrrggghhh!!
300 bloody pounds down the drain!!
Then I was deeply confused when I saw on page 181 a description of the "32
bit Integer and 32-bit Floating Point Math Libraries". It lists a number of
functions which include goodies like "32-bit floating point division", and
Now I'm totally confused. How can you use these math libraries when 32 bit
data types are not supported?
I'm a very distressed (and destitute) person at the moment!
Incidentally, can anyone recommend a DECENT 'C' compiler that does the job
properly and gives full 32-bit support in terms of data types and math
Finally, a message for Microchip: if you're compiler really won't support
32-bit data types then I'm DEEPLY unimpressed! Can I have my money
Many thanks in anticipation,
Web Site: http://www.btinternet.com/~stevethack/
|>Incidentally, can anyone recommend a DECENT 'C' compiler that does the job
>properly and gives full 32-bit support in terms of data types and math
I use the HiTech PICC compiler. It supports 32 and 24-bit floats (to save
space). I've been using it since the original beta releases (before you
could buy it) and have been very happy with their tech support.
I'm sure that ByteCraft's does as well. While I don't use the Bytecraft
compiler, it has been around a long time, too, and is well supported.
Sometimes Walter even helps with other problems, even though he knows I
don't own it - very appreciated!
Both companies subscribe to this list, so I'm sure you'll hear from them.
Andy Kunz Life is what we do to prepare for Eternity
rc-hydros.com andyhttp://www.rc-hydros.com - Race Boats
montanadesign.com andyhttp://www.montanadesign.com - Electronics
|On Fri, Aug 27, 1999 at 08:18:51PM +0100, Steve Thackery wrote:
> Now I'm totally confused. How can you use these math libraries when 32 bit
> data types are not supported?
In a similar manner to how a compiler can support other
higher-level data structures such as matricies without
having a corresponding data type; all access to the data
type simply goes through the library calls. Since the
PIC itself doesn't support floating point in hardare,
support for floating point has to be done in software. I
grabbed the demo copy and I found that, for example, the
fpm32.asm file contains the source code for a mechanism
to multiply two 32-bit floating point numbers each stored
in four 8-bit memory locations. This is more or less the
way that a compiler would have to do it anyway even if
floating point was "supported" with an explicit data
type. I did not, however, note the presence of e.g.
higher-level transcendental functions and such, but I
could just have missed them. ;-)
> I'm a very distressed (and destitute) person at the moment!
Not to be unsympathetic (we've all made such mistakes),
but uChip does post a time-limited demo and the full
documentation for download. (Yeah, I know, everything is
easy in retrospect...)
> Incidentally, can anyone recommend a DECENT 'C' compiler that does the job
> properly and gives full 32-bit support in terms of data types and math
Take a look at Hi-Tech C, it supports the 17Cxx series,
and 32 or 24-bit math; the manual is available on-line.
> Finally, a message for Microchip: if you're compiler really won't support
> 32-bit data types then I'm DEEPLY unimpressed! Can I have my money
I'm guessing that you'll want to contact
your local representatives on that one. Check
Bob Drzyzgula It's not a problem
drzyzgula.org until something bad happens bob
More... (looser matching)
- Last day of these posts
- In 1999
, 2000 only
- New search...