Comment by dang

If by the cap you mean the heap size passed in as the --dynamic-space-size argument, it didn't hit the cap. It was using about 2/3 of that.

> My understanding is the new parallel GC compacts the heap rather infrequently

Can you explain more?