- FUGAKUNEXT combines Fujitsu CPUs and Nvidia GPUs in Japan’s next national supercomputer
- The system is targeted at 600EFLOPS FP8 performance with a 100x application performance gain
- Riken, Fujitsu and Nvidia see the project as defining a new AI-HPC standard
Japan prepares its next national supercomputer, Fugakunext, through a collaboration between Fujitsu, Nvidia and Riken.
The system is planned for operation around 2030 and aims to mix simulation and artificial intelligence into a densely integrated platform.
For the first time in a Japanese flagship project, GPUs will be used as accelerators. Nvidia will (surprisingly) design the GPU infrastructure, Fujitsu will handle CPUs and system integration, and Riken will be involved in the software and algorithm work.
Feynman GPU
The result is expected to be an “AI-HPC platform” designed for science, industry and AI-driven discovery.
The benefits of the supercomputer are certainly ambitious. FUGAKUNEXT is designed to deliver more than 600flops of the FP8 AI Performance, which would make it the most powerful AI Super Computer yet announced.
The system is also expected to achieve up to one hundred times the increase in the application performance compared to Fugaku, while it remains within about the same 40 MW power budget.
Nvidia’s long-term roadmap points to the Feynman GPU architecture (named after theoretical physicist Richard Feynman), who arrives near 2028, so it could play a role in operating Fugakunxt.
Fujitsu develops a successor to his Monaka CPU for the project, so far named Monaka-X, with multiple cores, expanded SIMD capabilities and Arm’s Matrix Computation Engine for AI inferens.
Together with Nvidia’s accelerators, the system is expected to run large simulations along with demanding AI workload.
Hardware alone does not deliver target gains, so the project will also lean on innovations such as surrogate models, arithmetic with mixed precision and physics -information neural networks to speed up performance and at the same time maintain accuracy.
Makoto Gonokami, President of Riken, said: “It is a great honor for Riken to cooperate with Fujitsu and Nvidia in promoting the development of Fugakunext. Since ancient times, humanity has built up civilizations and advanced communities through the science of calculation. the computer, ”
Ian Buck, Vice President of Nvidia, added, “Fugakunext will deliver Zettascale —hydrobe with application speeds almost 100 times faster – within the same energy -footprint as its predecessor – speeds up research, increases industrial competitiveness and drives progress for people in Japan and around the world.”
