equivalent of systemC wait( time ) or SystemVerilog #time

[AdamRose66]

I notice that the simulator documentation for registerAction says this:

/// Registers an abritrary [action] to be executed at [timestamp] time.
///
/// The [action], if it returns a [Future], will be awaited.
static void registerAction(int timestamp, dynamic Function() action) { ...

So I wondering if it is possible to write a wait( int delay ) function which awaits a Future to provide wait( time ) like semantics ? Or has this already been done somewhere ?

Thanks,
Adam.

[mkorbel1]

Thanks for the question!

I think you could achieve this with something like this:

Future<void> wait(int delay) async {
  final toComplete = Completer();
  Simulator.registerAction(Simulator.time + delay, toComplete.complete);
  await toComplete.future;
}

Now you can just await wait(10) or wait(10).then(..., for example.

This is a useful utility, and I think something like it should be added to ROHD's Simulator by default.

I filed an issue tracking it here: #450

[AdamRose66]

minor naming comment : if we want to expose the await syntax, then maybe await delay(10) or delay(10).then(... ?

optionally we could also have

wait( int n ) async
{
  await delay( 10 );
}

[mkorbel1]

By the way, ROHD-VF already has a waitCycles function, which is useful if you wish to wait on edges rather than simulator time
https://intel.github.io/rohd-vf/rohd_vf/LogicWaiter/waitCycles.html

[AdamRose66]

Yes, I think I saw that in my search through the docs.Sometimes we don’t have an explicit clock. But my main concern is actually performance. I would hope that waiting for the time equivalent of 10,000 clock cycles should be much efficient than actually waiting on 10,000 individual clock edges.On 3 Jan 2024, at 17:36, Max Korbel ***@***.***> wrote: By the way, ROHD-VF already has a waitCycles function, which is useful if you wish to wait on edges rather than simulator time https://intel.github.io/rohd-vf/rohd_vf/LogicWaiter/waitCycles.html —Reply to this email directly, view it on GitHub, or unsubscribe.You are receiving this because you authored the thread.Message ID: ***@***.***>

[mkorbel1]

Yes, you're correct that waiting for the equivalent time is much faster than waiting for that many clock edges.