US Retakes First Place From Japan on Top500 Supercomputer Ranking (engadget.com) 29
The United States is on top of the supercomputing world in the Top500 ranking of the most powerful systems. From a report: The Frontier system from Oak Ridge National Laboratory (ORNL) running on AMD EPYC CPUs took first place from last year's champ, Japan's ARM A64X Fugaku system. It's still in the integration and testing process at the ORNL in Tennessee, but will eventually be operated by the US Air Force and US Department of Energy. Frontier, powered by Hewlett Packard Enterprise's (HPE) Cray EX platform, was the top machine by a wide margin, too. It's the first (known) true exascale system, hitting a peak 1.1 exaflops on the Linmark benchmark. Fugaku, meanwhile, managed less than half that at 442 petaflops, which was still enough to keep it in first place for the previous two years. Frontier was also the most efficient supercomputer, too. Running at just 52.23 gigaflops per watt, it beat out Japan's MN-3 system to grab first place on the Green500 list. "The fact that the world's fastest machine is also the most energy efficient is just simply amazing," ORNL lab director Thomas Zacharia said at a press conference.
Fourteen Years Since The First Petaflop Computer (Score:5, Informative)
Just fourteen years ago, in 2008 Roadrunner became the first petaflop computer [extremetech.com] and was shut down as obsolete just five years later.
The exponential transistor curve is flattening, so it will surely be much more than fourteen more years before the first zetaflop computer starts operation.
Re: (Score:2)
There are other ways to compute than transistors at microwave frequencies, perhaps optical will be a big jump
Re: (Score:2)
But will they be ready to overtake transistors in fourteen years?
Re: (Score:2)
Working prototype devices exist, would they be on chips with billions in five years or ten is the question. There are other alternative tech out there too with working prototypes, such as logic gates in graphene sheets.
Point is some amazing jump in compute power density or frequency could happen.
Zettaflop (Score:2)
The exponential transistor curve is flattening, so it will surely be much more than fourteen more years before the first zetaflop
It's zettaflop. For some unknown reason, they went with double 't' for the two top prefixes despite 'peta' only having one. Another reason why this will take a long time though is that having a single, incredibly powerful machine is much less useful now. When I last submitted a big processing job it was split up and ran on several different clusters around the globe. There are still some things where you need a lot of computing power in one place but outside those specialized tasks high speed networks make
Re: Zettaflop (Score:1)
the system has 8,730,112 total cores (Score:3)
Re: (Score:2)
Thanks, that and interconnects is half the article we care about after the past 50!
Meaningless nationalism (Score:2)
I like the competition between nations because it increases technology overall. But I have to say it is stupid that it's required but even stupider is the fact that one country gets the credit for the supercomputer when the technology is possible thanks to multiple countries. For example, the AMD EPYC CPUs are made in Taiwan. Japan's supercomputer used ARM technology which is based in the UK.
Nationalism is stupid, and every time it is utilized it is to tug at your tribalistic instinct and fool you into supp
Re: (Score:2)
It was tribalism that allowed our puny brains to form civilization, and as tribalism becomes obsolete we won't be smart enough to maintain civilization.
We're already way past that, as evinced by the unsustainability of our resource consumption.
Re: (Score:1)
Re: (Score:2)
You're a brainwashed liberal slashdot enviro-wackjob.
yawn
There's nothing "unsustainable" about our resource consumption.
https://www.google.com/search?... [google.com]
Re: (Score:2)
Nationalism is stupid, and every time it is utilized it is to tug at your tribalistic instinct and fool you into supporting something.
Indeed. The problem is that nationalism is all that the mediocre have: I may be good for nothing, but at least I belong to a great tribe. Nationalism and mediocrity go hand in hand.
Re: (Score:1)
Nationalism is stupid, and every time it is utilized it is to tug at your tribalistic instinct and fool you into supporting something.
Your statement is STUPID. Nationalism (within reason) is a GOOD thing. We are tribal by nature and genetics.
Re: (Score:2)
Just because we are something by nature doesn't necessarily mean it's good. Also, "we" doesn't apply to everyone. Some of us are globalists "by nature and genetics".
Re: (Score:1)
Just because we are something by nature doesn't necessarily mean it's good. Also, "we" doesn't apply to everyone. Some of us are globalists "by nature and genetics".
You globalists can fuck right off.
Let it pay for itself (Score:3)
I wonder how long it would take to pay for itself by minting Bitcoins, or whichever
Re: (Score:2)
I don't think floating point operations are very helpful with that.
Imagine... (Score:4, Funny)
A beowulf cluster of these!
Linpack and the HPC Challenge benchmarks (Score:2)
http://www.netlib.org/benchmar... [netlib.org]
http://icl.cs.utk.edu/hpcc/ [utk.edu]
For those curious as to what is actually being measured.
Chip isotopic purity (Score:2)
The isotopic purity governs how much heat is generated and how localised it is. Pure si28 transfers heat better, so you get fewer hot spots.
Then there's the interconnect. It may be possible to use superconductors rather than metals. May be.
And there's the reliance on silicon, when purportedly newer GaAs techniques reduced the cost of the technology.
This is an incomplete ranking (Score:1)
For some reason, although patents will always be made in US, we must be clear that not all the best about computer's development happens only between the US, China and Japan.
I suspect Russians have their very big secrets on this stuff, and that is the interpretation I give when I read this post about amazing computers that already work in secret [slashdot.org]