CHINA TOPIX

11/02/2024 07:21:58 am

Make CT Your Homepage

Move Over China; IBM, Nvidia Building World's Fastest Supercomputers with U.S. Money

IBM's Blue Gene/P supercomputer

(Photo : Wikipedia) The IBM Blue Gene/P supercomputer at Argonne National Lab runs over 250,000 processors.

A $325 million contract was granted IBM and Nvidia by the U.S. Department of Energy to develop two supercomputers that will enable GPUs and CPUs to transmit data 12 times faster than ever before.

The two supercomputers called Summit and Sierra will be available for civilian and scientific use. Summit, which has more powerful processing power than Sierra, will be located at the Oak Ridge National Laboratory in Tennessee while Sierra will be at the Lawrence Livermore National Laboratory in California.

Like Us on Facebook

During peak performance, Summit and Sierra will deliver raw computing power of at least 100 petaflops, which will make them the most powerful computers ever built. To date, China's Tianhe-2 or Milky Way 2 is the most powerful computer in the world, delivering 55 petaflops of processing power.

According to IBM, the supercomputer systems have the power to transmit data from their processors at 17 petabytes per second, which is comparable to moving over 100 billion Facebook photos in one second.

IBM reveals all information generated globally is estimated at 2.5 million gigabytes every day. If published in book form, this data will occupy 250 million football fields.

IBM's data-centric strategy involves placing computing power where data is also stored. This move will save energy. It also makes the computing process smoother when it comes to moving and passing data between data storage and CPUs and GPUs.

With IBM's partnership with Nvidia and Mellanox, IBM's computing power will be enhanced by its CPU and GPU communications courtesy of Nvidia and Mellanox that will provide graphics and networking capabilities at a maximum level.

IBM also partnered with Mellanox for its built-in intelligence since IBM also uses Nvidia's NVLink technology in its IBM Power CPUs and Nvdia's next generation graphic cards to further improve data handling.

With this project, IBM also aims to focus more on data centers as it responds to the growing demand of large scale computing and open platforms from different organizations and industries.

Real Time Analytics