- Related Stories
-
IBM regains supercomputer bragging rights
June 20, 2004 -
Apple developers to catch Tiger's tale
May 4, 2004 -
Apple delivers Xserve G5
March 23, 2004 -
Garage sale-ing for a Mac supercomputer
February 12, 2004 -
Apple continues modest move into clusters
January 7, 2004
In November, a machine called System X with 1,100 dual-processor Power Mac G5 workstations climbed to third place on the Top500 list of the most powerful supercomputers. On Monday, Huntsville, Ala.-based Colsa announced it's buying a larger system called MACH 5 to run Army simulations of the aerodynamics of flight much faster than the speed of sound.
System X, which vanished from the most recent list for upgrades, had sustained performance of 10.3 trillion calculations per second, or "teraflops." The Colsa system, made of dual-processor Xserve G5 machines, is expected to reach about 15 teraflops when it's up and running this fall, said project manager Mike Whitlock.
By comparison, the fastest system on a new version of the Top500 list, NEC's Earth Simulator, runs at a speed of 35.8 teraflops, and only one other system exceeded 15 teraflops.
Hewlett-Packard and IBM dominate the market for high-performance technical computing, with sales of $1.79 billion and $1.62 billion, respectively, in 2003, according to researcher IDC. But Apple is angling for its own share. It has released management software to control large groups of servers, and it sells models geared for supercomputing cluster use with unneeded components stripped out.
Much of the credit to Apple's successes thus far is due to the processor it uses--IBM's PowerPC 970--Illuminata analyst Gordon Haff said.
"The Macintosh software and the nice management features of OS X are factors here, but certainly performance of the processor is an enormous factor," Haff said. "PowerPC is a fast processor."
Indeed, the chip's abilities to perform a type of mathematical calculation called "floating-point operations" were compelling. "The floating-point units in that processor were particularly attractive," Whitlock said.
Colsa will use Mac OS X as the primary operating system, though it will evaluate other options including Red Hat Linux and Yellow Dog Linux, Whitlock added.
MACH 5, which stands for Multiple Advanced Computers for Hypersonic, G5, will occupy 42 racks and 600 square feet of floor space, said Anthony DiRienzo, a Colsa executive vice president. Apple was the winning bidder among six companies, DiRienzo said, declining to name the competitors.
One difference between MACH 5 and System X is the networking system that connects the individual servers. Where System X used the high-speed InfiniBand technology, MACH 5 will use the more conventional 1 gigabit-per-second Ethernet, DiRienzo said.
The fluid dynamics simulations Colsa will run require more processor power than top networking speed, he said.
System X's successor at the Virginia Polytechnic Institute and State University is expected to be running in time to produce a score for the coming November version of the Top500 list, said Alex Grossman, director of server and storage hardware at Apple. The upgraded Virginia Tech system is being built using Xserve G5 machines.
Apples cluster option would never be taken seriously by
any "Real" IT department. I guess this puts paid to that type
of thinking.
clusters should not be taken seriously.
http://news.com.com/
Commentary%3A+A+grain+of+salt+with+your+Apple/
2452-7339_3-5137856.html
I suspect the author would like a little salt to eat his words
with!
year, they also rarely reported that OS X is very secure, and then
when it had its first and only major security flaw to date, they
were all over it, saying Apple was handling the situation poorly.
They tend to regard Macs as toys, and consistently suggest that
to get real work done requires a Windows or Linux PC.
I also hate the fact that they refer to Mac users as "Mac fans" or
"Mac lovers."
"... Apple Computer, a real-world cousin to an academic system
that briefly appeared high on a list of the most powerful
machines."
Umm, do you KNOW why the Apple cluster was not on the latest
ranking list? I'm sure you do. But do you tell the reader the
reason or do you make it seem that the Mac cluster simply
wasn't powerful enough?
The Apple cluster was down for maintenance at the time of the
rankings and could not be tested. Look for it to either reclaim
the 3rd place in the top 500 or slip to 4th when the rankings are
done again in November.
"System X, which vanished from the most recent list for upgrades, had sustained performance of 10.3 trillion calculations per second, or "teraflops."
appeared high on a list of the most powerful machines..."
That caught me right away too. You see this type of word
shaping a lot these days in journalism, and when you point it out
their answer is always, "What bias?"
CNET authors/editors: The VT supercomputer did not "BRIEFLY"
appear on the list. It RANKED 3rd in the world at that time...end
of story. If you go back to that survey and resulting list, you will
STILL find the VT Apple cluster right in the 3rd position. It didn't
go anywhere.
It does not appear on THIS list simply because it was being
improved with new hardware. It was NOT surpassed, as implied,
it simply was not compared for the purpose of the latest
rankings.
Is it really SO difficult to write a positive story about Apple
without undermining the achievement? (especially with specious
reporting?)
computer.
system to use on a supercomputer!
Can you imagine that? It should turn a 10 TerraBit machine into
a one GigaBit pile of crap. Imagine rebooting every ten seconds.
Reinstalling every other minute of operation.
This has to be the best joke of the century.
not everyone. I'm not an Apple loyalist or anything like that but
since I started paying attention to computer trends you can see
what's going on. By the way, are these cnet guys supposed to be
journalists? I mean, since they do not do any of that or don't
appear to be a lot people have to wonder.
on this cluster, while the apple sites are reporting 25 teraflops.
This is a pretty huge discrepancy. Which of the two is shoddy
reporting?
performance capability of greater than 25 TFlop/s" so that is
probably its theoretical maximum. This guy Whitlock (who we
assume was interviewed by CNET, as he is not quoted in the
press release) states that it will reach 15, which is presumably
the sustained workload they expect it to achieve. Also, with
regard to the Top500 list, it says "only one other system [beside
the Earth Simulator] exceeded 15 teraflops" so that must be their
target to make them #2 or 3 on the list.
Are you surprised that anyone would try to make their product
sound great in a press release? When I bought my last sports-car
I read all the road tests in all the magazines. One managed to
achieve a 0-60 time a full second quicker than any of the others.
Guess which 0-60 time is quoted by the car maker; the one
most testers achieved, or the stand-out?
PS which "apple sites" are you referrring to? The only reference
on Apple.com is just a link back to this CNET story. Do you
mean the Mac boosters like Mac Daily News and MaCNN? That's
about as objective reporting as Fox News on the GOP.
have to give CNET some slack with an obvious opinion piece.
Unfortunately the author of the opinion is not identified and that
is inexcusable.
theoretical speed of the system, whereas the 15 teraflops is the
estimated actual throughput.
Also, just as a note, I can't find anything from Apple directly
claiming 25 teraflops. The press release from COLSA states "It is
expected to deliver a peak performance capability of more than
25 TFlops/second", but that's from them, not Apple, and "peak"
to me implies theoretical, not sustained actual throughput,
which is what the "Top 500" list measures.
Here's the link to the PR:
http://www.prnewswire.com/cgi-bin/stories.pl?
ACCT=104&STORY=/www/story/06-21-2004/
0002197322&EDATE=
The CNET article (which is the only one Apple has referenced on
their Hot News link) gives performance as 15 teraflops, as an
estimate.
Second, People use WINDOZE because they HAVE to, while Mac
users use Macs because the WANT to.
Third, PC users swear AT their PC's & Mac users swear BY their
Macs.
Fourth: future of computing is in multi-processor super
computing grid clusters. While PC WINTEL maintains the fact that
dual processor PowerMacs are useless & 4 GHZ single processor
Pentiums are much faster & better for desktop towers.
Fifth: FOLLOW THE MONEY: Editors are bias towards who butters
their bread, notice all the WINTEL ads flashing around during
your reading the article or the posts?....mmmmm....coincidence
that CNET typists are biased towards Wintel Cartel Monopoly?
-Eyes wide open in Seattle -
- 24 Tera Flops NOT 15
- by Thomas, David October 28, 2004 2:20 AM PDT
- Check your facts, prove me wrong. But I believe the COLSA
- Like this Reply to this comment
-
(21 Comments)super cluster, MACH5, is running at 24 tera-flops
By the way, since OSX, IS essentially a Linux/Unix OS with a REAL
user interface, WHY ON EARTH would anyone replace it??!!!
Don't be stupid, please.