16 bit 5 tick torchless ALU with one forwarder for 6 tick CPUs [Diagonal]

[EEVV]

[Divine intellect warning]

I made a 16 bit torchless ALU that uses a modified version of Koy's BCS for the better output->input route.
Screenshot:

But really, you should `/warp 16bitalu` for the full immersive experience™.

[Update: fully functional with 4 tick Bout as of 2017/8/8]
Capabilities (if you have the inverters):
Addition, subtraction,
All logic gates,
Disable ALU out,
Right shift w/ sign (on forwarding side and non forwarding side),
Left shift (on forwarding side only)

[estp]

look nice

[Koyarno]

So, if the shift is only on the forwardable side, shouldnt it be wise to memory map the accumulator to a register too?