Simulate Fixed-Point on Windows

[Derek Price]

How do you simulate Fixed-Point math on a Windows PC?  For example, in the article Comparing Floating-Point and Fixed-Point Implementations on ADI Blackfin Processors with LabVIEW it describes how to do fixed-point math on the Blackfin DSP using the Blackfin Fixed-Point palette.  If you want to verify your algorithms on the PC side, these VIs don't work on the PC side.  How do you verify your results?  Are there corresponding Blackfin fixed-point VIs on Windows?  I know there's a FXP data type on Windows, but does it correspond to the Blackfin's (or any DSP's) fixed-point representation (Ex. 1.15 fract16)? 

Thanks

[smercurio_fc]

LabVIEW 8.5 and 8.6 have the Fixed-Point datatype.

https://www.ni.com/docs/en-US/bundle/labview/page/numeric-data-types-table.html

[Derek Price]

I saw the FXP data type, but how do you simulate embedded fixed-point VIs on Windows.  for example, the Blackfin pallete has fixed-point add, subtract, etc. using fract16.  How do you test this on Windows?

Thanks

[craigc]

If you want to emulate the blackfin Arithmetic block you will have to write code to emulate how it operates and use the basic maths pallette with a fixed point precision.  i.e to calculate a logarithm one valid way would be to use a cordic algorithm  However as i do not know exactly how the blackfin Arithmetic unit functions (Carries etc) you will have to find this out from somewhere then emulate it in LabVIEW.  Which in itself is probably not a small task.

[Jeremy_B]

Hi Derek,

If you create a numeric indicator, set it to FXP, and then go to the data type properties, you can set the integer and word length.  Set the word length to 16 bits and the integer length to 1 bit, and that should do it. 

[Derek Price]

Hi Jeremy,

Okay, so I can simulate the same fixed-point integer format on Windows.  What about simulating the embedded fixed-point palette?  We have our algorithms working well in Windows, but other than compiling and downloading to an embedded target, how do we verify the fixed-point conversion using similar VIs?

Thanks

[Jeremy_B]

Hi Derek,

I don't think there is any way to do this other than running it on the Blackfin itself.  The libraries are designed specifically for that processor.