IBM and the DoE launch the world’s fastest Supercomputer
The supercomputer — which fills a server room the size
of two tennis courts — can spit out answers to 200 quadrillion (or 200 with 15
zeros) calculations per second, or 200 petaflops, according to Oak Ridge
National Laboratory, where the supercomputer resides.
"If every person on Earth completed one calculation per second, it would take the world population 305 days to do what Summit can do in 1 second," according to an ORNL statement.
Put another way, if one person were to run the calculations,
hypothetically, it would take 2.3 trillion days, or 6.35 billion years. The
former "world's fastest supercomputer," called Sunway TaihuLight, can
perform 93 quadrillion calculations a second (93 petaflops), while humming away
inside China's National Supercomputing Center in Wuxi.
The supercomputer is
an IBM AC922 system that's made up of 4,608 computer servers — each comprising
processors (the brains of the computer). But what's actually going on inside
these processors is what makes the difference.
"Summit's computer architecture is quite different from
what we have had before," Daniel Jacobson, a computational biologist at
ORNL, who is working on Summit, told Live Science. For one thing, the computer
uses the new Tensor Core feature in its graphics cards (made by Nvidia), which
is designed specifically for applications focusing on machine learning and artificial
intelligence (AI), and to be fast.
Basically, unlike older computer chips, these chips are
optimized for a special type of mathematical operation on matrices — or
rectangles filled with numbers with rules for adding, subtracting and
multiplying the different rows and columns. Computers equipped with AI programs
often learn using so-called neural networks, which have several layers in which
lower calculations feed into higher ones. And this process requires the heavy
use of matrices.
Image Credits 📷: Oak Ridge National Laboratory
"This is a brand-new feature that has allowed us to break
the exascale barrier," Jacobson said, referring to a processing speed
that's over a billion billion calculations per second.
In addition, Summit has loads of superfast memory (RAM)
available on each of its nodes, where localized calculations can take place.
"Each node on Summit has 512 Gb [gigabytes] of RAM and the
network that communicates between nodes uses adaptive routing, and is thus
incredibly fast, which helps us scale the calculation across all the nodes very
efficiently," Jacobson said. So-called adaptive routing means Summit has
some flexibility in how it runs calculations — sort of like networks of brain
cells connected to synapses.
Trending Articles
Comments
Post a Comment