AnsweredAssumed Answered

21479 single/double precision floating point

Question asked by simon.hutton@utas.utc.com on Jun 15, 2017
Latest reply on Jun 28, 2017 by Kader.M

Using Visual DSP++ 5.1.2 and running on a 21479 I have a problem with the resolution of single precision floating point accuracy. Could anyone point me to how under project options/processor the double size works. If I change the double size from 32 bit to 64 bit   the LDR produced increases from 298kB to 387kB even though I have no doubles defined in my code. My assumption is that all floats have now become doubles. This then obviously slows the code dramatically and my system is very time critical. 

There is also a tick box that says allow mixing of sizes, where is this defined, does this allow 32 and 64 bit, if so how.

Any help would be appreciated.

Outcomes