ABC9 "&mfs" step gives incorrect result on big endian for objects with >5 variables about yosys HOT 8 CLOSED

Could you file this against ABC instead of Yosys? This isn't in our code.

from yosys.

Yosys uses a fork of ABC, which does not have issue tracking enabled, which is why I figured it was better to report it here. But I can take it upstreams if you like.

from yosys.

It's true that we have a fork of ABC, but we follow upstream very closely.

And, as the Yosys ABC9 maintainer, let me tell you that I find the ABC code equally difficult to work with; but the effort that would be required to replace it with something cleaner and more maintainable would be immense, because ABC has lots of little tricks to improve QoR that are not documented.

from yosys.

Right, because who cares that we're outputting an incorrect netlist as long as it toggles really fast? 😇

from yosys.

Who knows, perhaps something might become of toymap.

from yosys.

We could also compile abc to wasm, run it thru wasm2c, and it will work fine on any platform, LE or BE.

from yosys.

And the code will be about as readable as before. Sounds good to me!

from yosys.

Perfect! Looking forward to this solution being merged then.

from yosys.