pub fn ba_gates_output(group: u32, local_out: u32) -> u32Expand description
The BA output a thread owns.
PARENT: n = blockIdx.x * BA_GATES_OUTS + local_out.
TWIN: n = (g0 + g) * BA_GATES_OUTS + local_out, where g0 + g walks the
same 0..ceil(N/4) range in tiles of BA_GATES_GROUPS.
The two must be the same function of (output group, local_out).