My question is about the fork...join_none construct in SV.
According to the LRM section 9.3.2 Parallel blocks, Table 9-1—fork-join control options:
The parent process continues to execute concurrently with all the processes spawned by the fork. The spawned processes do not start executing until the parent thread executes a blocking statement.
Using the join_none doesn't wait for any of the forked process to complete and proceeds executing the non-blocking statements. I am trying to think of a scenario where one would use join_none during design/verification.
2 Answers
The most common use of fork/join_none in SystemVerilog is when you want to spawn a dynamic number of parallel threads. In a testbench, each thread could represent a single device on a bus.
initial begin N = $urandom_range(1,10); for( int j = 1; j <= N; ++j ) fork automatic int k = j; // local copy, k, for each value of j start_a_thread(k); join_none endjoin_none is widely used in verification methodologies, such as UVM. It is used to spawn off multiple threads in parallel in the testbench. For example, it can be used to start an interrupt monitor in parallel with a data bus driver.
