World’s Fastest Supercomputer Now in China

What is the world’s fastest supercomputer? It can no longer be found in American soil as that title has been bagged by the Chinese with their newest supercomputer, the Tianhe-1A aka “Milky Way.”

The behemoth of computing power has more than 7,000 graphics processors and 14,000 Intel chips. It can perform more than 2.5 thousand trillion calculations a second or 2.5 Petaflops per second.

In contrast to the former world number one supercomputer, Oak Ridge National Laboratory’s XT5 Jaguar which can only perform at 1.75 per second, Tianhe-1A is 47% faster according to Prof Jack Dongarra from the University of Tennessee.

He told the BBC News:

This is all true. I was in China last week and talked with the designers, saw the system, and verified the results.

I would say it’s 47% faster than the Oak Ridge National Laboratory’s machine, 1.7 Pflops (ORNL system) to 2.5 Pflops (Chinese system).

Tianhe-1A is kind of unique for it draws on the power of graphics cards made by nVidia and processors by Intel, the two now becoming rivals in the chip-making industry.

The nVidia chips are used for simple calculations while the task of crunching complicated mathematical operations are given to the Intel chips.

The supercomputer is now being used to serve the Chinese weather service and the National Offshore Oil Corporation.

Looking at the Top500 list of supercomputers, the top 10 as of June 2010 are 7 US-made supercomputers, 2 from China and 1 from Germany.

I wonder when will a supercomputer made by Filipinos will be included in the list?

3 Comments

  1. I wonder if the comparison of supercomputers in terms of instructions per second remains fair, given the difference in the types of operations that GPUs (the nVidia graphics processing units) and CPUs (the Intel central processing units) can handle. GPUs can handle so much more instructions per second than CPUs, even though they're usually much cheaper; adding GPUs therefore is a cheap way to increase a computer's FLOPS rating. And yet, because computer programs designed to run on the much-simpler computational capabilities of the GPU are much harder to code than traditional programs that run on the CPU, the usefulness of the GPUs on such computers are limited. On PCs, for instance, GPUs are often only useful when playing games.

    Reply

Leave a Reply