HDLcoder ignores coder.const

4 次查看(过去 30 天)
I'm tryint to compile some code into HDL. There are weights used in digital filter which I want to precompute at the compile time.
The way I did this is
persistent c buf tout pfb_weights;
if isempty(c)
...
pfb_weights = coder.const(get_pfb_weights(settings_Nfft,settings_Ntaps));
end
The reason I think it is not working is that the coding report says:
  • It lists sin function under Toolbox Function Usage: Supported List. Sin function is only used in get_pfb_weights function
  • It lists coder/matlabcoder under Toolbox Function Usage: Unsupported List
  • If I attempt fixedpoint conversion it complains that function "double" inside get_pfb_weights is not suitable for fixed point conversion and wouldn't proceed.
How do we fix this?

采纳的回答

David Fink
David Fink 2022-9-14
By making the call to get_pfb_weights opaque to code generation (so fixed-point conversion won't try to process it), you can call it extrinsically (in MATLAB), at compile-time, via coder.const and coder.extrinsic.
See the coder.extrinsic documentation for a coder.const example.

更多回答(1 个)

Kiran Kintali
Kiran Kintali 2022-9-25
You can find examples of HDL Coder friendly designs with coder.load and coder.const in the attachment.

类别

Help CenterFile Exchange 中查找有关 Fixed-Point Designer 的更多信息

产品


版本

R2022a

Community Treasure Hunt

Find the treasures in MATLAB Central and discover how the community can help you!

Start Hunting!

Translated by