Basically, emit them in Verilog like you'd emit UInt<n> -- don't flatten them, not even on ports.
It's not inconceivable that we could specialize Vec[Bool] in Chisel instead, but someone's got to do this to provide decent simulation performance. This is probably the single most profitable simulation performance optimization beyond DCE.
Basically, emit them in Verilog like you'd emit UInt<n> -- don't flatten them, not even on ports.
It's not inconceivable that we could specialize Vec[Bool] in Chisel instead, but someone's got to do this to provide decent simulation performance. This is probably the single most profitable simulation performance optimization beyond DCE.