Monday Aug 06, 2007
Many news sources now covering UltraSPARC T2, the new high-performance chip from Sun.
This new UltraSPARC T2 chip leads in many ways. I'll cover the performance numbers tomorrow.
For now:
http://www.computerworld.com.au/index.php/id;898889798
http://www.reuters.com/article/technologyNews/idUSN0625780420070806
http://www.channelweb.co.uk/vnunet/news/2195718/sun-lifts-lid-niagara-processor
etc..
For some of my previous comments:
http://blogs.sun.com/bmseer/entry/news_trickles_out_on_niagara2
Please remember that the previous generation chip, the UltraSPARC T1,
just set an application-tier world record (all details at link). How many times has the "old" chip with half as many threads set a world record weeks before the new one is announced?
A final note. I venture that this chip is going to lead for database, application tier, and of course web tier, oh and don't forget HPC, yes it is that versatile.
Friday Aug 03, 2007
News is starting to trickle out on Niagara2. Sun continues to do revolutionary
things that will have huge effect on how we compute. For what has leaked so
far see: http://www.informationweek.com/blog/main/archives/2007/08/suns_niagara_2.html
At OpenSPARC.net you will
find a in-depth presentation of the upcoming Niagara 2 processor: "An 8-core, 64-thread, 64-bit, power efficient SPARC SoC (Niagara2)".
The presentation was held at the International Solid State Circuits Conference 2007.
So the question becomes how do we now judge performance of chips with different
architectures? One thing that hasn't changed is system performance and system cost. Look at what
a system costs compare it against other systems in the same price range.
The computer industry can no longer determine performance based on old dinosaur habits. Remember when vendors determined all performance by GHz. Then remember back
in '04 when Intel quite naming processors by GHz... they realized that
GHz is only part of the overall equation for processor performance.
What does that equation look like in very rough terms?
performance = GHz * number-of-cores * number-of-threads-per-core * efficiency-of-design * other * software efficiency
...so if you compare performance between different designs by any of these factors
alone (example: perf/core, perf/Ghz, perf/thread) you are being very misleading!
Case in point, IBM loves to talk about perf/core, but they avoid telling everyone
that they lead the industry, by several factors in price/core!
The IT jungle says: "340,000. It costs $17,700 to activate a core, so pushing it to eight cores costs another $141,600." So on a per core basis that means POWER6 costs
$60,000/core. Actually I've estimated that in a system with memory an IBM power6
system costs $65K/core to $130K/core! Any customer needs to do their own pricing.
So comparing systems of dramatically different cost-per-core on a per-core basis
NO LONGER MAKES ANY SENSE! No wonder IBM wants to stilt comparisons and avoid
pricing! The only other clues on list price on power6 is
here.
Wednesday Feb 07, 2007
IT News has an article about Elliptic Curve Cryptography (ECC) and its
interoperabililty across multiple platforms, including Solaris, Windows
and Linux. Also a mentioned of Niagara 2.
For more info see:
http://www.itnewsonline.com/showstory.php?storyid=8249&scatid=8&contid=3
Wednesday Feb 07, 2007
CNET news has some press about Sun's next CMT technology.
"We have silicon back (from manufacturer Texas Instruments) and
Victoria Falls systems running in the labs at Sun at full speed,
full performance," John Fowler, executive vice president of systems,
said via Webcast Tuesday. "In the first half of calendar 2008, we
will introduce products based on Victoria Falls."
You can read more at:
http://news.com.com/Sun+dual-Niagara+servers+due+in+2008/2100-1010_3-6156883.html
...all I can say is fast to production, and fast in the lab!
|
|
|
|
From your...
When the Sun UltraSPARC T1 came out it set all kin...