Interesting. I wonder if the slowdown of small pages is a result of excessive pointer chasing on the kernel side, which intuitively does a better job of trashin

Interesting. I wonder if the slowdown of small pages is a result of excessive pointer chasing on the kernel side, which intuitively does a better job of trashing the TLB than sequential accesses would.