[Beowulf] Re: Finally, a solution for the 64 core 4TB RAM market

Joe Landman landman at scalableinformatics.com
Thu May 28 11:11:37 EDT 2009


Jason Riedy wrote:

> code, and the sequential/slow parts are faster.  The main
> difference will be in I/O, but bumping the memory way up will make
> I/O less important for many uses.

Er ... ah ... no.  It will only delay the inevitable.  It won't 
eliminate it.

Blue Gene units have massive main memories.  And they aim for very fast 
IO.  Same with most of the other big iron.

If you have to do IO, you should do it as fast as you can do it 
(safely), so you can minimize that fraction spent in serial IO 
operations.  Yeah, MPI-IO ... parallel operations across a collection of 
inherently serial resources ...


-- 
Joseph Landman, Ph.D
Founder and CEO
Scalable Informatics,
email: landman at scalableinformatics.com
web  : http://scalableinformatics.com
        http://scalableinformatics.com/jackrabbit
phone: +1 734 786 8423 x121
fax  : +1 866 888 3112
cell : +1 734 612 4615
_______________________________________________
Beowulf mailing list, Beowulf at beowulf.org sponsored by Penguin Computing
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf



More information about the Beowulf mailing list