I don't really have anything off hand.

But code primarily slowed down by pointer chasing down big maps, which is MillenialMan's complaint and fits my own experience, will absolutely be sped up linearly.

A bunch of cores sitting around waiting for the next HAMT node to come in will not interfere with each other in the slightest.