[Beowulf] HPC Market Question

Kilian CAVALOTTI kilian.cavalotti.work at gmail.com
Mon Jan 19 09:17:04 EST 2009


On Friday 16 January 2009 14:25:02 Jon Aquilina wrote:
> i havent done much digging in the transistor count but from people i have
> talked to in regards to the i7s the i7s are true quad cores im not sure
> what that means for the core 2 quads that came before the i7.

It means that Core2 Quad were architected as two dual-core chips sitting on 
the same silicon die. Each pair of cores shared a L2 cache, but nothing was 
shared (except the FSB, yikes!) amongst the four cores. In that respect, they 
were often considered as "fake" quad-cores, and more as "dual-dual-cores".

On the contrary, Nehalem quad-cores feature four independent cores with no 
intermediate hierarchical relationship, each with its own L2 cache. The four 
of them also share a large L3 cache, which make it a "real" quad-core, as 
Barcelona was.

The following summarizes the architectural differences between Core2 Quad 
(Harpertown), Barcelona  and i7 (Nehalem):
http://www.realworldtech.com/includes/images/articles/Nehalem-1.gif

> > i7 is a pretty big discontinuity in intel's design.  it'll be interesting
> > to see how smoothly they make it.  but don't the first i7's actually
> > have _lower_ transistor count than current core2's?

Yes. Quad-core Nehalem is about 730M, whereas quad-core Penryn was over 800M. 
This is mainly due to the cache size: 12MB L2 for Hapertown, 1MB L2 + 8MB L3 
for Nehalem.


Cheers,
-- 
Kilian

A: Because it messes up the order in which people normally read text.
Q: Why is top-posting such a bad thing?
A: Top-posting.
Q: What is the most annoying thing in e-mail?
_______________________________________________
Beowulf mailing list, Beowulf at beowulf.org
To change your subscription (digest mode or unsubscribe) visit http://www.beowulf.org/mailman/listinfo/beowulf

-- 
This message has been scanned for viruses and
dangerous content by MailScanner, and is
believed to be clean.



More information about the Beowulf mailing list