Using Verilog Case Statement With Continuous Assignment

Question

I'm trying to perform continuous translation in Verilog from some greycode values to some binary values. That is, I'm trying to take a greycode value coming in on one bus and continually translate it to a binary value on another bus. I'm trying to do this in Verilog (cannot use SystemVerilog for other reasons).

I wanted to do something like this:

wire [DATA_SIZE - 1:0] binary_snap;
always @(greycode_snap) begin
case (greycode_snap)
    8'b00000000 : binary_snap = 0;
    8'b00000001 : binary_snap = 1;
    8'b00000011 : binary_snap = 2;
    8'b00000111 : binary_snap = 3;
    8'b00001111 : binary_snap = 4;
    8'b00011111 : binary_snap = 5;
    8'b00111111 : binary_snap = 6;
    8'b01111111 : binary_snap = 7;
    8'b11111111 : binary_snap = 8;
    8'b11111110 : binary_snap = 9;
    8'b11111100 : binary_snap = 10;
    8'b11111000 : binary_snap = 11;
    8'b11110000 : binary_snap = 12;
    8'b11100000 : binary_snap = 13;
    8'b11000000 : binary_snap = 14;
    8'b10000000 : binary_snap = 15;
    default     : binary_snap = 0;
endcase
end

Where greycode_snap is a wire.

As you may have guessed, this doesn't synthesize and produces the error:

Procedural assignment to a non-register binary_snap is not permitted, left-hand side should be reg/integer/time/genvar

I don't want to change binary_snap to a reg so next I tried this:

wire [DATA_SIZE - 1:0] binary_snap;
assign binary_snap = (greycode_snap == 8'b00000001) ? 1 :
                     (greycode_snap == 8'b00000011) ? 2 :
                     (greycode_snap == 8'b00000111) ? 3 :
                     (greycode_snap == 8'b00001111) ? 4 :
                     (greycode_snap == 8'b00011111) ? 5 :
                     (greycode_snap == 8'b00111111) ? 6 :
                     (greycode_snap == 8'b01111111) ? 7 :
                     (greycode_snap == 8'b11111111) ? 8 :
                     (greycode_snap == 8'b11111110) ? 9 :
                     (greycode_snap == 8'b11111100) ? 10 :
                     (greycode_snap == 8'b11111000) ? 11 :
                     (greycode_snap == 8'b11110000) ? 12 :
                     (greycode_snap == 8'b11100000) ? 13 :
                     (greycode_snap == 8'b11000000) ? 14 :
                     (greycode_snap == 8'b10000000) ? 15 : 0;

Again, greycode_snap is a wire.

Which the compiler accepts, but in my (limited) experience I believe this will synthesize into something gross (depending on the synthesis tool used). So I tried a little optimization:

wire [DATA_SIZE - 1:0] binary_snap;
assign binary_snap = (greycode_snap[0] == 0) ?
                            (greycode_snap[4] == 0) ?
                                (greycode_snap[2] == 0) ?
                                    (greycode_snap[1] == 0) ? 1 : 2
                                : //else
                                    (greycode_snap[3] == 0) ? 3 : 4
                            : // else
                                (greycode_snap[6] == 0) ?
                                    (greycode_snap[5] == 0) ? 5 : 6
                                : // else
                                    (greycode_snap[7] == 0) ? 7 : 8
                        : // else
                            (greycode_snap[4] == 1) ?
                                (greycode_snap[2] == 1) ?
                                    (greycode_snap[1] == 1) ? 9 : 10
                                : //else
                                    (greycode_snap[3] == 1) ? 11 : 12
                            : // else
                                (greycode_snap[6] == 1) ?
                                    (greycode_snap[5] == 1) ? 13 : 14
                                : // else
                                    (greycode_snap[7] == 1) ? 15 : 0;

Again, greycode_snap is a wire.

But this code is already extremely convoluted, rigid, and not maintainable.

How can I do this in a clean way? It may also be important in the future that this code is expandable. If you have any suggestions related to fast greycode counting/translation schemes, those would also be appreciated.

Answer 1

Your code works if you change the declaration of binary_snap from wire to reg. But you can keep it as a wire and use your case statement if you put it into a function.

wire [DATA_SIZE - 1:0] binary_snap;
assign binary_snap = f(greycode);
function [DATA_SIZE - 1:0] f(input [7:0] reg code);
case (code)
    8'b00000000 : f = 0;
    8'b00000001 : f = 1;
    8'b00000011 : f = 2;
    8'b00000111 : f = 3;
    8'b00001111 : f = 4;
    8'b00011111 : f = 5;
    8'b00111111 : f = 6;
    8'b01111111 : f = 7;
    8'b11111111 : f = 8;
    8'b11111110 : f = 9;
    8'b11111100 : f = 10;
    8'b11111000 : f = 11;
    8'b11110000 : f = 12;
    8'b11100000 : f = 13;
    8'b11000000 : f = 14;
    8'b10000000 : f = 15;
    default     : f = 0;
endcase
endfunction