Javelin doesn't support floating point!!!!!! OMG

WuLabs

I just figured out that the Javelin doesn't support floating point calculations or types! I guess i purchased the Javelin assuming it would have it, as both the BasicX and the Atom Pro both have it, and the Javelin definitely seems much more powerful. Especially for a language such as Java, I don't understand why the Javelin doesn't support FP!!!!

Now I have to find another microcontroller, as this isn't usable.

Jon Williams

Neither the Basic-X or Atom Pro support Java, so I don't quite understand why you would make the comparison. Indeed, the Javelin has a very small JVM, and clever programmers have created floating point classes to use when actually required in an embedded project. Most embedded applications don't require floaing point mathematics, so to add it would have been an inefficent use of resources.

Just a note: We make all of our product information avialable to download at no charge.

▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔▔
Jon Williams
Applications Engineer, Parallax
Dallas, TX· USA

WuLabs

Yeah, from using FP for a long time, I had assumed that it would come with such a supposedly "more powerful microcontroller", even after pre-reading the manual. Anyways, the IEEE745 Floating Point library which is under development, does it actually work at this point? I can't find the forum that relates to that.

Singularity

wuLabs,

I share your pain. It appears that I am the only other embedded developer that needs native floating point
If the device supported native 32 bit signed int then maybe it would not be so painful for those in need of precision.

There is the micro fpu coprocessor. despite being an add- on device it is fairly fast. there is another thread on it elsewhere on this forum.

Kindest regards

Peter Verkaik

There is the FloatLite6 (AN013) here

http://www.parallax.com/javelin/applications.asp

that gives 6 decimals of precision. Range is -32768.999999

to +32767.999999

On the same page there is the Int32 class (AN011) for signed long types.

The Float32 class can be found here.

http://www.parallax.com/dl/appnt/jav/class/float32.zip

As far as I know all classes are functioning.

regards peter

WuLabs

I took a look atboth of those, but they aren't powerful enough. For instance, FloatLite only allows addition and subtraction. I need a full FP library, including trig functions, div and mult. In Parallax, theres a section for apps under developement. ONe of the things under developement is a full IEEE745 32 bit FP library which includes all these things. There is no documentation(because its under dev)? I was wondering if anyone knew more about this.

Also, i saw some FP math coprocessors. They are ~$20-$40.....which is ok. but i don't want any external stuff.

Andy Lindsay (Parallax)

The Javelin Stamp does support floating point.· I think this class will have the computing power you need:

[*]Float32.zip
It is avialable from www.parallax.com -> Products -> Javelin Stamp(TM) Module -> Applications -> Apps Under Development

Post Edited (Andy Lindsay) : 1/14/2005 5:55:00 AM GMT

WuLabs

that's the one i was talking about. before I load it, has anyone had experience with it? Theres no documentation!

Andy Lindsay (Parallax)

Run the demos programs; they tell all. Seriously, they're comprehensive, and easy to understand.