vtr-verilog-to-routing
vtr-verilog-to-routing copied to clipboard
Placement `channel_width_fac`
Throughout the placement code, we need to frequently use min
and max
functions in time-sensitive parts of the code before accessing channel_width_fac
. This data structure should be updated to cover the entire grid (from 0 to width-1
, and so on). During the pre-computation of the inverse of the average channel width, we can assume that the channel width at width-1
is the same as the channel width at width-2
. This change would simplify the code, remove an asymmetry that is no longer necessary, and might slightly speed up the code.