Okay, just to be very very explicit, do you believe that my user project should be driving the

la_data_in

pins or the

la_data_out

pins? At first I assumed I should be driving the

la_data_in

pins, because I assumed they were the inputs to the LA, but now I'm thinking I should be driving the

la_data_out

pins (as in, they are the outputs from my user project). Do you know which way around it is?

From looking at the

user_proj_example.v

it sure seems like this new interpretation is correct (i.e. I should be driving the

la_data_out

pins, not the

la_data_in

pins).

Sorry, I know I'm asking this a ton of different ways, because I'm trying to be super duper certain I'm getting this the right way around. In the Verilog I see:

input  [127:0] la_data_in,
    output [127:0] la_data_out,
    input  [127:0] la_oenb,

In the .lef I see:

la_data_in[0] is DIRECTION INPUT
  la_data_out[0] is DIRECTION OUTPUT TRISTATE
  la_oenb[0] is DIRECTION INPUT

From this I am inferring that the

la_oenb

pins are being driven by the Caravel harness itself (and should not be driven by my user project), and I should be driving the

la_data_out

pins (and not the

la_data_in

pins).

Does the above summary agree with your understanding?

The example project uses la so you always refer to that. And your testbench will fail if you have the pins connected badly.

And yes you can ignore the enable signal

Also, better place to discuss this is in the caravel channel, it's definitely come up before