Native floating point setting in HDL coder simulink

Hello,

From some tutorial I came to know that if I set Native floating point setting in Simulink then Matlab will take care conversion of single data type to appropriate fixed point data type. It works for all the blocks except look up table, where it fails with the error that "native floating point is not supported by lookup table".

Please suggest me a solution or workaround for it.

Thank you

Answer 1

[Kiran Kintali]

I am assuming you are referring to native floating point capability available in HDLCoder.

With native floating-point capability you generate HDL code directly from floating-point model (in single precision) without requiring fixed-point conversion. This can be very useful when targeting high dynamic range algorithms onto FPGAs. The direct lookup table and lookupND blocks are supported in 17b release (and area available for preview in 17b pre-release software). If you do not have access to 17b pre-release please convert LUT data fixed-point.

If you do not need high dynamic range support features available with Native Floating Point please consider using fixed-point conversion workflow to convert your algorithms to fixed-point.

[Mohammed Shameem Hussain]

I tried to make the look up table to a fixed point, but that also does not work and through the same error.

It may not be possible for me to convert the whole project to fixed point as it is very time consuming and I cannot look each and every block for surety.

Hence, I omitted lookup table and decide to manually create the look up a table in Quartus (Altera) and then use it in HDL coder generated code.

looking forward to 17b. I hope it soon comes out.