<Wanda[cis]>
so at any given grid cell you have 8 primary clocks and 8 secondary clocks
<Wanda[cis]>
the primary clocks are normal; they are assigned per-quadrant and driven from the clock center in the middle of the device
<Wanda[cis]>
the secondary clocks have finer granularity though
<Wanda[cis]>
each column is used to distribute two secondary clocks, which reach one column left and two columns right (or the other way around, I don't remember), and you can mux the clocks per-column
<Wanda[cis]>
so effectively the clock regions for secondary clocks 04 15 26 and 37 are all misaligned with each other
<Wanda[cis]>
there's also a bunch of ugly special casing close to the edge of the device, where the regions are smaller / larger then usual
<Wanda[cis]>
and that special casing is done differently between ECP2 and ECP3
<Wanda[cis]>
anyway, yeah, Lattice. I was bored on a train and decided to resurrect a 3-year-old stale branch where I worked on Lattice geometry extraction
<Wanda[cis]>
and by "resurrect" I mean "rewrite from scratch using the old one as reference" because the approach I took back then was kinda garbage
<Wanda[cis]>
so now we have a full routing database for ECP, XP, and MachXO. which is 3 subtargets out of 13 or maybe 14.