[Beowulf] Feedback on large pages in Linux

[Joe Landman]

Toon Moene wrote:

> Fascinating.
> 
> How would you envision this done ?
> 
> Most of our "large memory usage" is of the form:
> 
> main program:
> 
>       READ * L, M, N
>       CALL MAIN(L, M, N)
>       ...
>       END
> ....
> separate file:
>       SUBROUTINE MAIN(L, M, N)
>       REAL U(L, M, N), V(L, M, N), T(L, M, N), Q(L, M, N)
>       ....
> etc.
> 
> I.e., the memory used (automatic arrays) is based on the stack (at 
> least, that's how most Fortran compilers would implement it.

Yes, that is how it was implemented in IRIX as I remember.  You simply 
ran the code with a helper environment/application that let you use 
large pages.

With Linux (today) you have a hugepagefs that you need to use by 
mmap'ing files into, or by creating shm segments.  See 
http://developer.osdl.org/dev/robustmutexes/src/fusyn.hg/Documentation/vm/hugetlbpage.txt 
for more info.

> 
> The "application" doesn't allocate these pages, the operating system does.

Of course.  But the usage model is the same one you have been using.  My 
point (perhaps unclear) was that the usage model for large pages is a 
bit different under Linux.  Not necessarily what we might want.

> 
> Cheers,
> 


-- 

Joseph Landman, Ph.D
Founder and CEO
Scalable Informatics LLC,
email: landman at scalableinformatics.com
web  : http://www.scalableinformatics.com
phone: +1 734 786 8423
fax  : +1 734 786 8452 or +1 866 888 3112
cell : +1 734 612 4615