On 3/9/06, Eric Lowe <[EMAIL PROTECTED]> wrote:
> >> The performance
> >> analysis was, as I recall, done mostly using US-III+.
> >
> > Did this include the concept that dwarfpages (8k) are no longer
> > available to both kernel and user land applications?
>
> I wasn't directly involved in the 64K prototype but only 64K and larger
> were used for user applications, and the page_t was 64K in span
> (PAGESIZE=65536). There may have been some 8K mappings in the kernel due
> to OBP handing off translation lists with holes -- I don't remember the
> details there.

Who was involved in this project?

> > Can Sun release the code? I'd be more than happy to write a project
> > proposal then. I assume we will receive some help from the HPC
> > community. The next generation of vector supercomputers may use
> > similar large page sizes (>= 64k) as minimum configurable value and
> > Opensolaris may be a good testbed implementation - assuming we can
> > completely eradicate dwarfpages from kernel and user land.
>
> I'll ask around.

Did you receive any answers yet? I'd like to write a project proposal
for a 64k kernel project - assuming Sun is willing to release the
sources of their original prototype.

> Keep in mind though that just flipping page_t's to be 64K isn't a magic
> pill; most applications in the HPC space, etc. still need huge pages to
> fit their working set in the TLB, and spend most or all of their TLB
> misses on the working data set, so 64K is not likely to be a win at all in
> that space.

It depends on the application. For the majority of smaller and medium
sized applications such as FEM 64k pages have a serious advantage. I/O
operations are limited by dwarfpages, too. MPSS buys nothing in this
case.
--
Holger
_______________________________________________
opensolaris-discuss mailing list
[email protected]

Reply via email to