Both functions and tasks are subroutines that are reusable RTL or testbench codes.
Return Value and Argument Directions#
Function#
- must return a single value, whose function you declare in the header
- arguments are all inputs, cannot declare output or inout ports
- called from expressions
y = my_func(a,b);
Task#
- does not return value directly, to get data out, must use output or inout arguments
- can have input, output, and inout ports
- called as a standalone statement
my_task(a,b,c,), not inside an expression
Timing Control and Simulation time#
Function#
- cannot include any timing controls (
@, wait, #delay) or non-zero simulation-time constructs - must execute in zero simulation time - pure combinational logic
Task#
- can include timing controls, delays, event controls, and even
fork … join - useful for driving testbenches or modeling multi-cycle behaviors.
Usage in RTL vs. Testbench#
- Functions are used in small, combinational operations such as checksum, bit-field packing, and arithmetic helpers.
- Tasks are used when you need a sequence of events such as stimulus generation, bus transactions, and waiting for handshakes
Calling Rules and Synthesis#
Functions#
- can be called anywhere an expression is allowed (assign statements, inside procedural blocks, and in parameter calculations)
- synthesizable if they are zero-time and returns single value.
Tasks#
- cannot be used in expressions
- generally not synthesizable and used most often in testbenches
Example#
Function#
function int add (input a, input int b);
add = a + b;
endfunction
Task#
task wait_for_flag(
input logic clk,
input logic flag,
output int count);
count = 0;
while (!flag) begin
@(posedge clk);
count++;
end
endtask