Japan's Fugaku loses title as world's fastest supercomputer

What next, Japan will spend more tax payer money to keep with that?

It’s more like a prestige object, isn’t it? Maybe, that handful of researchers are now happy like kindergarten children that they’ve got the toy of their dreams, but they could also do their job with a normal personal computer. Those worldwide rivaling supercomputing hasn’t even solved any single one of the problems on this planet so far, and probability is therefore very high , they also won’t in the future. In fact, one can easily get the same or similar useless or limited useful results by own thinking and then using an old programmable pocket calculator, those also already for decades could and can handle regressions and transition matrices. Might then be a slower processing and take 30 minutes or an hour, that admitted, but even that doesn’t matter so much. lol

I guess the iPhone won out?

However it's still the world's fastest supercomputer with a fax add on

I have a female friend who suddenly brought this up a year and a half ago or so after it was on domestic news, saying, "I'm so proud of Fugaku and Japan. Does the US have a supercomputer like ours?" I said, "I'm not American, for starters. Second, I don't know, and I really am not that interested. I didn't help create it, and I stink when it comes to computer technology. Why would I be proud of something I didn't help create?"

She didn't catch the nuance at all. Neither, it seems, have a number of people since. I am glad there is fierce competition if it leads to innovation and our advancement as a species, but aside from people taking part in the work, I really don't get the idea of pride here aside from it covering a total sense of insecurity.

Competition is good.

Well one less thing Riken can boast about, I would be much more interested in reports where practical applications of Fugaku are examined, research or industrial projects where it has usefulness for example.

Competition? In just only heating up the planet more and unnecessarily? Those biggest supercomputers can’t do more than a few older and former supercomputers combined couldn’t do also, and the global merits or innovations from them are , if ever existent, quite rare and limited. But maybe you could name a few to convince me, let’s say only three big breakthroughs that have been calculated and then solved a big global problem. Only 3, I don’t want the task solving to become impossible, lol Good luck!

The HPE Frontier system can perform over 1 quintillion calculations per second.

1,000,000,000,000,000,000 operations per second.

Just because it was used for analyzing some data doesn't mean a less powerful computer couldn't have done the same...

Maybe pushing the boundaries of supercomputing but hasnot even entered the race for quantum computing yet

There is nothing special about supercomputer anymore, they are nothing more than X86+nVIDIA GPU server farms, with the only engineering being case design, network switch, and cooling.

Once upon a time supercomputer used to be custom engineered hardware, but not anymore.

The computing equivalent of 'who has the biggest you-know-what'. A game played by insecure nationalist regimes with tax-payers' money.

quote: 'artificial intelligence capability'.

LOL. Yeah. Absolutely. The term is meaningless. You want something that functions like a human being? Take a bus ride. Listen to two people talking with each other - more likely talking at each other. You could emulate that with Eliza in 16K on a Z80 in the 1980s.

'AI', really 'machine learning', has no generalised form. It has to be applied in a specific implementation, as it is in nature. If what you really mean is running more algorithms per second, then that could be deemed a basic speed, but not ML.

As for the capability of a system, that depends upon what you are doing with it, the language and the implementation more than the processor speed.

You want to have some fun, produce software that self-optimises dynamically (at both the program and language level) on a processor that can also self-optimise dynamically at a microcode level for whatever you are running. Hardware, firmware and software dynamically re-writing itself to work as fast as it can on any single problem.

If you want to build you own supercomputer (on a budget) wikipedia 'Beowulf Cluster'. This is a cool computer club activity.

quote: Fugaku's predecessor, the K supercomputer, was decommissioned in 2019.

Hinagiku, Hinagiku...

The US controls the semiconductors core technologies, so they can easily have many supercomputers for the nationalist contests as much as they want (as long as they have enough materials and resources). China will likely acquire and master the technologies in semiconductors within a few years, so the Western advantages in semiconductors will be gone as with solar panels and solar cells.

Supercomputers play an important role in the field of computational science, and are used for a wide range of computationally intensive tasks in various fields, including quantum mechanics, weather forecasting, climate research, oil and gas exploration, molecular modeling (computing the structures and properties of chemical compounds, biological macromolecules, polymers, and crystals), and physical simulations (such as simulations of the early moments of the universe, airplane and spacecraft aerodynamics, the detonation of nuclear weapons, and nuclear fusion). They have been essential in the field of cryptanalysis. (Wikipedia)

https://en.wikipedia.org/wiki/Supercomputer