Re: ALTFP_CONVERT in Modelsim

Altera_Forum · ‎04-14-2012

I'm currently using the ALTFP_CONVERT megafunction in my design to convert a 10 bit integer to a 32bit FP. It has a clk_en (convert_en) and an aclr port (convert_rst).

Unfortunatly the output stays at high impedance (as shown in the attachment).

I get the following warning during the simulation setup:

--- Quote Start ---

# ** Warning: (vsim-8607) c:/altera/11.1/modelsim_ase/win32aloem/../altera/verilog/src/cycloneiii_atoms.v(6155): Non-positive replication multiplier inside concat. Replication will be ignored.

--- Quote End ---

which I don't really understand at all. Could that be related? Or is there something else I'm missing? Thanks in advance :)

Altera_Forum · ‎04-15-2012

You are not showing the output port of the altfp_convert megafunction. I don't think, that the output of altfp_convert is tri-stateable, apparently it's sourced from a register.

It's also not clear, if the warning is generated during processing of the MegaFunction or other parts of your design.

Altera_Forum · ‎04-15-2012

Thank you for your reply.

The section of my testbench for stimulating the convertor:

--- Quote Start ---

//Convert signals

logic convert_rst;

logic convert_en;

logic[9:0] in;

logic [31:0] out;

//Module

fpconvert converter(convert_rst,CLOCK_50,convert_en,in,out);

//Convert Signals

initial

begin

convert_rst = 1'b0;

convert_en = 1'b0;

in = 10'b0000000001;

# 50ns;

convert_rst = 1'b0;

convert_en = 1'b1;

in = 10'b0000100001;

# 50ns;

convert_rst = 1'b0;

convert_en = 1'b1;

in = 10'b0000100001;

# 50ns;

convert_rst = 1'b0;

convert_en = 1'b1;

in = 10'b0000100001;

# 50ns;

convert_rst = 1'b0;

convert_en = 1'b1;

in = 10'b0000100001;

# 50ns;

convert_rst = 1'b0;

convert_en = 1'b1;

in = 10'b0000100001;

# 50ns;

convert_rst = 1'b0;

convert_en = 1'b1;

in = 10'b0000100001;

# 50ns;

convert_rst = 1'b0;

convert_en = 1'b1;

in = 10'b0000100001;

end

--- Quote End ---

As you can see, it is the megafunction output port which is high Z.

I tried simulating without the 'mains' module (which is my entire design) which removed the 'Non-positive replication multiplier inside concat' error. So from what I can tell, it's a either an issue with the Megafunction or Modelsim.

Altera_Forum · ‎04-15-2012

Problem solved. I made a new copy of ALTFP_CONVERT with a different name and that seemed to fix it. Surprising how changing the name of a module seems to fix things... I have a feeling I may have made an incorrect port assignment in another module, which messed up the synthesis of ALTFP_CONVERT when I tried testing it by itself in modelsim.

Thank you for your help :)