News on China's scientific and technological development.

t2contra said:

... to deliver real world benefits ...

Click to expand...

Jura said:

I don't mean to naysay as SamuraiBlue the Chinese achievements in supercomputing but if it had been my problem, I would've been interested in REAL benefits of ever-increasing output, in an analogy with increasing the caliber of main battery guns of Battleships in let's say 1910-40 I mean it was considered to be strategic investment (as supercomputing is), it worked (as supercomputers do, despite SamuraiBlue not liking the metrics but 'the ultimate Battleship', while tactically dubious, would've been just too expensive to build, and the analogy would be if an exascale computer could "forecast with certainty" (*) for example China's coastal area weather in real time?

Please, Log in or Register to view URLs content!

(*) some stuff can't be easily (because of "the chaotic nature of the problem")

just my thoughts after lunch, in the middle of Europe, anyway ...

Click to expand...

SamuraiBlue said:

There is a gimick in the fastest claim. The Sunway TaihuLight CPUs doesn't communicate with one another as fast as the Japan's K supercomputer. That is the reason why K is at top of

Please, Log in or Register to view URLs content!

ratings with less CPU cores while the Sunway TaihuLight that has 15 time more CPU can't out do K in Graph500.
It has to do as wiki explains,

Please, Log in or Register to view URLs content!

in which K is able to move more data at one time through parallel computing. This is required for real time weather prediction and/or process big data collected through all the sensors that is accumulating at an alarming rate.
At the end it all boils down to the architecture of the super computer's design.

Click to expand...

Jura said:

here's the problem: you know, I've been hearing about 'real world benefits to be delivered by a supercomputer' for a long time, for example twenty years ago
#1 had 2,048 cores and 368 GFlops

Please, Log in or Register to view URLs content!

while ten years after
#1 had 131,072 cores and 281 TFlops

Please, Log in or Register to view URLs content!

while ten years after
#1 has 10,649,600 cores and 93,015 TFlops

Please, Log in or Register to view URLs content!

and managers etc. have been talking about 'real world' all the time, so I just wonder, GENERALLY ABOUT ALL SUPERCOMPUTERS IRRESPECTIVE IF IN US, JAPAN, ..., if they'll ever deliver (I'm not interested in applications to beat benchmarks if you hear me)

Click to expand...

1. how far it's necessary to go to be able to predict California weather, Honshu Island weather, ... weather, in real time?
2. if people will go into
   Please, Log in or Register to view URLs content!
   computing?

SamuraiBlue said:

Yawn..

Click to expand...

Quickie said:

So much of distracting and irrelevant comment. You know that the K Computer in the Graph500 list didn't even make it into the Top500 list whereas the Taihulight is in the Graph500 2nd and Top500 1st place?

You have bragged so much about the K Computer being good at atmospheric/weather prediction application. Instead of just bragging, why not show the proof of exactly what it can actually do, like what the Taihulight has done by winning the 2016 ACM Gordon Bell prize by doing exactly the kind of application you claimed the K Computer was good at. i.e. atmospheric simulation/prediction application.

The fact of the matter is both the data communication speed and FLOPS performance are important in processing but we can tell from the above competition the latter is the more significant determinant in the above kind of application, and in probably most other application.

Please, Log in or Register to view URLs content!

Chinese supercomputer project wins top int'l prize
Source: Xinhua 2016-11-18 20:28:47

Please, Log in or Register to view URLs content!

NANJING, Nov. 18 (Xinhua) -- A Chinese team on Friday won the 2016 ACM Gordon Bell prize, a top honor in high-performance com

Please, Log in or Register to view URLs content!

g, for an application running on China's fastest supercomputer.

It is the first time a Chinese team has won the award.

The project, named "10M-Core Scalable Fully-Implicit Solver for Nonhydrostatic Atmospheric Dynamics," presents a method for calculating atmospheric dynamics, according to the Association for Computing Machinery, which presented the award at the International Supercomputing Conference in Salt Lake City in the

Please, Log in or Register to view URLs content!

.

"The application can help improve global climate simulation and weather prediction," said Yang Guangwen, director of the National Supercomputing Center in Wuxi. ....

Click to expand...

Quickie said:

That's a different K Computer. The K Computer we're talking about didn't make it to the Top500 list.

Click to expand...

Bltizo said:

I thought they were the same?

Click to expand...

Jura said:

...
if people will go into

Please, Log in or Register to view URLs content!

computing?
...

Click to expand...

taxiya said:

SB is doing exactly what he is good at "moving the goal post". Imagine the following:

After Ferrari won a formula 1 race, SB would come out and claim that Ferrari is nothing compared to Toyota prius which is the greenest car in the world. See how and where has he moved the goal post?!

Click to expand...

Jura said:

... and this early morning I took it ad absurdum (or not?!) by

(it's not clean but I needed to be quick taking the buildup from 2011; 2016 and assuming one exaFlops achieved in 2021:

the nitty-gritty:
a=2.2538;b=0.75104;c=2.3973;
solving y=a+b*exp(c*x) for 1000 (=one zetta):
z =
5.8824

in a graph:

the present time hardly distinguishable from zero (one exa expected only around 2020), but ... LOL!

EDIT

Please, Log in or Register to view URLs content!

Click to expand...