IBM and GENCI team up to create a new breed of ultra-fast supercomputer
IBM has announced that it is to collaborate with leading French computing agency GENCI (the grand équipement national de calcul intensif) to work on massively increasing the speed of supercomputers.
The aim of the partnership is to take a step towards computers that can work on the exascale level – that’s the ability for a system to perform a billion billion calculations every second, or one exaflop. The current fastest systems in the world can perform between ten and 33 million billion calculations per second, or 10 to 33 petaflops.
“Put into context, if exascale computing is the equivalent of an automobile reaching 1,000mph, today’s fastest systems are running within a range between ten and 33 miles per hour,” IBM said in a statement.
The collaboration is planned to initially run for 18 months, and is predicted to end up with computers at speeds of more than 100 petaflops. That marks a sizeable leap from today’s fastest systems and a big step towards exascale computing.
Looking forward to getting a 100-petaflop system in your iPhone? You might be waiting for some time. IBM and GENCI say the focus of the collaboration is on complex scientific applications for the new ultra-fast systems, which will use computing technology from IBM’s OpenPOWER ecosystem.
(Above: GENCI’s CURIE supercomputer)
GENCI is owned by a mixture of bodies, including the French state, the Commissariat à l’énergie atomique and the French National Centre for Scientific Research. Access to the OpenPOWER ecosystem will allow GENCI to tap into a wide network of high-performance computing innovations based on IBM’s POWER processor technology. As an example, IBM says this will include the connection of Nvidia GPUs accelerators to POWER processors through high-speed Nvidia NVLink interconnects.
IBM seems confident that this partnership with a leading French computer agency will end up in the next stage of evolution for supercomputers: “We fully expect our collaborative efforts will produce innovations capable of moving the supercomputing industry that much closer to exascale,” said Michel Teyssedre, CTO of IBM France.
If we do reach an exascale level of computing, the repercussions could extend much further than faster phones and laptops. One exaflop is thought to be the processing speed of the human brain at neural level, which means exascale computers could help power simulations of the human brain, such as through the Human Brain Project.