Karl H

Sorry. This fell on my backburner. I will need to test everything again. From my recollection, I was deliberately trying to force an email to be queued, even if there...

@rachitnigam, I just tried this change on the unsigned-dot-product example in `interp/`, and It seems to have simulated fine?

I just checked on my checkout of main (which is on commit f4e80ec0). It looks like Icarus executes correctly and produces a waveform. So no, it looks like Icarus was...

I have tried my hardest to make sure Firesim does not have to change anything, but fabric seems to cause the majority of issues I have hit so far, so...

You need to pull `platforms/xilinx_vcu118/garnet-firesim` past commit 5bf5b9a2de8c6670024f01334a17da58feb3bdad. That resolves the naming issue that shows up in synthesis.

> constant selects in always_* My immediate guess from the error being thrown is that Icarus does not like the `1'b0` or `WIDTH`. Because `always_ff` "desugars" to something like Verilog's...

Ok, after some tinkering, I have narrowed down the problematic expression in the `always_comb`. It is the bit slicing of `acc_next`. I did some pruning just to make the module...

We may not even need to go that far, depending on how Calyx uses `always_comb`. Verilog 2001 introduced syntax that mirrors SystemVerilog's `always_comb` that we should be able to use....

I saw that, which means the blast radius of this kind of change will shrink. If I make this change to `std_fp_div_pipe`, is there a way to run a set...