With this patch parse nodes are allocated sequentially in chunks. This reduces fragmentation of the heap and prevents waste at the end of individually allocated parse nodes. Saves roughly 20% of RAM during parse stage.