My China Dream | Rainer Spurzem

Professor Rainer Spurzem, 54, had two China dreams.

The first came about 25 years ago in 1987, when he was still a college student in Germany and had followed a woman he liked very much to China. She went to Guangzhou for a Chinese-language exchange program. And Spurzem spent a couple of weeks visiting her before traveling from Shenzhen to Beijing.

In Beijing, the "sea" of bicycles during rush hours made a huge impression on the young man, as did the long lines at the post office, all where people waited patiently for one or two hours just to make a phone call.

His budding romance went nowhere, so Spurzem soon forgot his first dream about China.

But he didn't know then that a second China dream would come true and help him build a professional connection with Chinese scientists.

By 2003, Spurzem had already obtained an honorary professorship from the University of Heidelberg, Germany. He became a senior astrophysicist and designed a supercomputer in Germany. Five years later, he found his desire for supercomputing research was stagnating in Europe.

At about this time, at an international astronomical conference, Spurzem met a chief scientist of the National Astronomical Observatories under the Chinese Academy of Sciences. The scientist persuaded him to apply for funding from China's Ministry of Science and Technology.

In April 2010, with the financial help from the ministry and numerous research institutes under the CAS, a supercomputer designed by Spurzem's team was put into operation and named "Tiger".

"One important aspect, which is often forgotten as our supercomputing and the Tiger computer capture public attention, is that we are actually doing astrophysical research on fascinating problems like black holes in galaxies and the gravitational waves which were predicted by Einstein," Spurzem says.

Astrophysicists use supercomputers for astrophysical simulations - much like how car manufacturers simulate car crashes digitally - where they simulate collisions of galaxies and black holes, Spurzem says.

"Car companies use computer simulations to save money, because crashing real cars is more expensive. We use computers because galaxy and black hole collisions in nature take incredibly long time, and we can simulate this lengthy period with ultra-high speed on the computer," he says.

Spurzem's major contribution was to revolutionize the design of supercomputers and bring them to China.

A supercomputer is at the frontline of current processing capacity, particularly in speeding up calculations often used for numbers-intensive tasks, such as problems in quantum physics, weather forecasting and climate research.

In a supercomputer, a large number of processors are used in clusters to enhance calculations speed.

Spurzem's innovation is that he applied the theories of the graphics processing unit, a core part of the video driver of a personal computer, to supercomputers.

"With the installation of GPU cards, the computing power of a supercomputer is improved, becoming 50 times faster than traditional supercomputers. This is a technological revolution," Spurzem says.

Although GPU card technology is proven on home computers, it had not been applied to supercomputers before China's Tianhe 1A came out. "In the case of applying GPU cards on supercomputer, China is the first," Spurzem says.

Tianhe 1A, China's milestone supercomputer, can perform 2.57 quadrillion computing operations per second and was first among the Top 500 supercomputers' list in November 2010.

Spurzem's Tiger computer uses the same technology as Tianhe.

"When I came to China in 2009, the Tianhe computer was still under design, and it became the fastest in 2010. It is perhaps because China is a late beginner in some key technologies that it could leapfrog forward."

In Europe and the United States, supercomputers had been around since the 1960s, and hardware revolutions meant many program applications had to be changed.

"That was why it was hard for me to get support in Europe when I came up with the idea of the GPU cards. In China, there are better research possibilities," Spurzem says.

Maxwell Tsai, 24, from Guangzhou, cherishes the opportunity to be Spurzem's student.

"In our team, even the most senior scientist shows respect for another person's idea. Professor Spurzem, for example, is very supportive of my new discoveries, even though they may not be completely to his research interest," Tsai says.

"Instead of telling me what to do, Spurzem would brainstorm ideas so I can figure out what to do and how to do it under his guidance."

In 2011, Spurzem's work received recognition by Europe - and his team was awarded the Partnership for Advanced Computing in Europe prize, which is usually granted to major science achievements with high-performance computing resources in Europe.

On Dec 5, China's top political leader Xi Jinping met with 20 foreign experts working in China. Spurzem was among them.

"This is new for Chinese leaders," Spurzem says. "In the US, about 50 percent of scientists are foreign immigrants. Maybe China, in the same way, will have a bigger lead in attracting world talents."

Currently, Spurzem leads a 10-person team in a program named Silk Road at the National Astronomical Observatories, and team members come from five different countries. Spurzem decided to quit his job at the University of Heidelberg to work full time in China.

"In China there is good technical support, excellent students and great research opportunities. So, why not?"

chengyingqi@chinadaily.com.cn

(China Daily 01/22/2013 page18)