How does awk handle collisions in a hashmap?

Does awk use separate chaining, open addressing or does it have its own way of handling collisions in a hashmap? Do gawk and nawk implement the same algorithm? Thank you.