Back

Zoomed in with the 100x objective we can see that the purple areas are not actually interlayer dielectric, it's structural coloration from fine features that are too small for the 20x objective to resolve.

The column muxing logic looks to have 16 way symmetry, then another 4 ways inside that.

But each of those 128 regions likely isn't one sense amp, it's probably several bits high (16 bits if my hypothesis about the tiles being 2048 bits wide is correct).

All of the solid colored pink regions down the center spine are actually big parallel buses of wires that are too small to resolve. This objective is a 100x/0.90 with a pixel size of 42.5 nm on this camera, but the limiting optical resolution is around around 370 nm for green light (550 nm) and 270 nm for blue light (400 nm).

I don't know exactly what node this chip is made on, or what the metal pitch on this layer is, but it's clearly well below my optical resolution.