------------------- Original query --------------------------
> I was recently asked by someone about SPARCS having problems with their
> MMU page mapping. The problem being the table thrashing if there was
> more than 16 or 32 MB of RAM, which in turn would decrease the
> performance by an order of magniture.
>
> I realize this is very vague, and I apologize for the lack of detail,
> but this is all I could get from the person.( The person just happens
> to be one of my many superiors. )
>
> So, my question to y'all is:
>
> Has anyone heard of the MMU page mapping causing thrashing if a SPARC
> has more than 16 or 32 MB of memory.
>
> Any responses to this cursed and vague question are greatly appreciated.
------------------------------------------------------------------------------
Thanks to:
David Fetrow <fetrow@biostat.washington.edu>
kevins@kuma3.Japan.Sun.COM (Kevin Sheehan {Consulting Poster Child})
"B.McCrone" <B.McCrone@daresbury.ac.uk>
The problem is an old one caused by the kernel not keeping enough
soft copies of MMU state around. In turn, it would have to throw real
physical pages out of memory to reclaim MMU resources.
This was known as the "pmeg bug", was found with the 4/60 machines and has
since been fixed by altering the kernel source.
Many thanks to those who replied.
Larry Chin {larry@cch.com} CCH Canadian Ltd.
System Administrator 6 Garamond Court
Research and Development Don Mills, Ontario.
(416) 441-4001 ext. 349 M3C 1Z5
The lion and the calf shall lie down together but the calf won't get
much sleep.
-- Woody Allen
This archive was generated by hypermail 2.1.2 : Fri Sep 28 2001 - 23:06:47 CDT