> On Dec 15, 2018, at 1:53 PM, Jaromír Doleček <jaromir.dole...@gmail.com>
> wrote:
>
> Bigger physical pages also reduce size of page tables (big deal if it
> e.g. allows to reduce e.g. 4-level to 3-level mapping, but even "just"
> factor or two or four is very good) and hence speed up page faults,
> and improve TLB utilization. It's really good thing overall and
> something which is very desirable on monster memory machines.
This is only the case if a real "larger physical page size" is available, of
course, as opposed to simply clustering e.g. 4 4KiB physical pages into a 16KiB
VM page.
Another thing that would be interesting would be to bucket memory so that the
allocator could easily request e.g. a large contiguous region, which would
enable you to, for example, map all of libc's text with a single L2 PTE.
-- thorpej
