Writing tagged parallelism

An interactive walk through tiled matrix multiplication, memory locality, outer products, and thread-owned output tiles.

How owner-computes scheduling splits a large result matrix into disjoint C tiles so threads can use outer-product updates without atomics or competing writes.