How to write a Verilog to implement -MULT(27x27) so that QPRO map the full functionality to 1 DSP?

[DNguy11]

Using QPRO it is possible to build an IP to implement -(X*Y) where X and Y are of 27 bit-width. However I find NO verilog writing style such that QPRO is able to map the full functionnality within 1 DSP

[Kenny_Tan]

What device that you were using? You will first need to understand the architecture of the DSP. You can look into it in the resource property editor.

With that said, you can look into how to write the code in right click the verilog.v files -> insert template -> Verilog HDL -> Full Designs -> Arithmetic -> DSP features

Thanks

[DNguy11]

Issue is with any stratix10 device. I already used the resource-property-editor to implement -(aXb) using the negate of the mult + the load_const of the DSP. So the solution is feasible. I already looked at the possible templates but nothing is proposed to implement -(A*B) within a single DSP

[Kenny_Tan]

If that is the case, you might have to use the IP from the IP catalog. Can you look into using those?

[DNguy11]

For portability reason I need to go though Verilog. Does it mean that QPRO is not able to infers such a basic DSP functionality?

[Kenny_Tan]

That is correct, if you cannot find anything from https://www.intel.com/content/dam/www/programmable/us/en/pdfs/literature/ug/archives/ug-qpp-design-recommendations-19-3.pdf?wapkw=12+recommended+hdl+coding+styles

Which means Qpro was not able to infer the DSP functionality base on your requirement. We can certainly make enhancement on this.