Main Site Documentation

Would a floating point co-processor make sense?


#1

I’ve been thinking on and off about this floating point co-processor from Sparkfun: http://www.sparkfun.com/products/8129

I like playing with external chips but I’m actually wondering if this makes sense on a board as powerful as the ChipWorkX. I guess my Panda2 would benefit from it, especially the gps features look interesting… but I’m not sure it would do much on a CWX or Hydra.

What’s your opinion? Does the ARM CPUs already have a floating point unit inside? Plus you’d have to consider the time it takes to send the request and get it back from the chip via the serial interface (SPI or I2C). An RLP driver might help though.


#2

The chip you pointed to is basically a micro (proably 8bit PIC or AVR) micro that is programmed to handle floating point.

Any on our board is a lot more powerful by many folds. Even if we do not have the functionality you may need, you can add it using RLP and still be much faster.

The value of that chip is in the software provided, not in speed. Of course if you are using a 16Mhz arduino then that chip will make things faster but not on 200Mhz ARM9 :slight_smile:


#3

Thanks Gus, that’s actually what I thought but still wanted to ask to make sure… I still use avr cpus like the attiny2313 or atmega328 from time to time so it might come in handy one day but if I ever get one I won’t bother writing a C# netmf interface for it… :wink: :wink:


#4

You can do ti for the fun of it but not to make ChipworkX faster. I t will actually make it slower :slight_smile: