Adrian Sampson

This is a nice idea. To expand on this a tiny bit: * The way the "simulation" wrapper would have to work is by re-instantiating the memories. That is, the...

@ayakayorihiro's efforts on a FIRRTL backend for Calyx (cf #1898, most recently) have brought up this issue again, and crystallized a bit more what we need to do here. Specifically,...

Thanks for the ping! I haven't implemented anything like priority in fud2 yet. Seems worth doing. Currently, the first thing that the BFS algorithm finds "wins" unless you use `--through`!...

I have actually always been a little confused about what `timeout` was supposed to do. Is it a cycle limit for the accelerator invocation, so "runaway" Calyx programs nonetheless complete...

Indeed; many thanks, @sgpthomas! That makes lots of sense to me. Seems like it wouldn't exactly hurt to have such a timeout, but given that this stuff is basically working...

The summary, @YJDoc2, is essentially just bit-rot—yes, we have disabled automatic updates, so it's not getting synced to the most recent version of the Calyx compiler. Getting things working again...

Awesome!!! Thank you so much for looking into this; truly amazing. As you suggest, it would be great to review a PR—your proposed changes all sound excellent, so we can...

:smiley: Always a fun kind of edge case to consider. My gut feeling says Option II (A terminates, B runs forever). The underlying semantic rules would be: * `control {}`...

Yeah, there's a strong argument for that as well. It would address the optimization problem: the `seq {}`-to-nothing optimization would remain allowed everywhere, because `control { seq {} }` is...

Awesome! This is very much a good plan. Just to iterate (i.e., bike-shed) on the details, here are a few assorted suggestions: * Re. `std_mem`: Instead of putting this in...