Japan is set to embark on an ambitious project to develop a ZetaFLOPS-scale supercomputer capable of performing one sextillion floating-point operations per second. The Ministry of Education, Culture, Sports, Science and Technology has announced plans to build a successor to the country’s Fugaku supercomputer, which was once the world’s most powerful high-performance computing machine.
According to TomsHardware, The new supercomputer aims to achieve an unprecedented 50 ExaFLOPS of AI performance, with RIKEN and Fujitsu leading the development process starting next year. The project is expected to receive significant government funding, with an initial allocation of ¥4.2 billion ($29.06 million) in the first year and a total investment exceeding ¥110 billion ($761 million). The supercomputer will feature advanced storage systems, special-purpose accelerators, and CPUs designed by Fujitsu, including its next-generation Monaka processor featuring up to 150 enhanced Armv9 cores.
Japan’s Ambitious Plan for a ZettaFLOPS-Scale Supercomputer
Japan’s Ministry of Education, Culture, Sports, Science and Technology (MEXT) has announced plans to develop a successor to the country’s Fugaku supercomputer, which was once the world’s most powerful High-Performance Computing (HPC) machine. The new supercomputer aims to achieve an unprecedented 50 ExaFLOPS of AI performance with Zetta-scale peak performance in mind, utilizing Artificial Intelligence (AI) for scientific purposes.
The Zetta-class designation indicates a system capable of performing one sextillion floating-point operations per second, making it 1,000 times faster than ExaFLOPS. If Japan manages to build such a system by 2030, as planned, it will likely again have the world’s fastest supercomputer. The development of this supercomputer is expected to be led by RIKEN, one of Japan’s most prominent research institutions, in collaboration with Fujitsu.
The Ministry of Education, Culture, Sports, Science, and Technology plans to allocate ¥4.2 billion ($29.06 million) in the first year of development, with total government funding expected to exceed ¥110 billion ($761 million). The new supercomputer is expected to use Japanese technology, with a focus on maintaining software compatibility with Fugaku.
Computational Node Performance
Each computational node of the Fugaku Next supercomputer is expected to have peak performance of several hundred FP64 TFLOPS for double-precision computations, around 50 FP16 PFLOPS for AI-oriented half-precision calculations, and approximately 100 PFLOPS for AI-oriented 8-bit precision calculations. The memory bandwidths are expected to reach several hundred TB/s using HBM-type memory.
To put these numbers into context, the peak performance of a Fugaku computational node is 3.4 TFLOPS for double-precision calculations, 13.5 TFLOPS for half-precision calculations, and 1.0 TB/s for memory bandwidth. The significant increase in performance is expected to enable the supercomputer to handle complex AI workloads and large-scale scientific simulations.
Architecture and Hardware
The Ministry of Education, Culture, Sports, Science, and Technology does not envision any particular architecture for the Fugaku Next supercomputer. Still, its documents suggest that it should use a CPU with special-purpose accelerators or a combination of CPU and GPU. The supercomputer is also expected to feature an advanced storage system capable of handling both traditional I/O workloads for data science and large-scale checkpointing and new I/O requirements for AI workloads.
Fujitsu’s next-generation CPUs, featuring up to 150 enhanced Armv9 cores, are likely to be used in developing the supercomputer. The CPU will adopt a multi-chipset configuration, with these cores distributed across multiple-core dies, which will be paired with SRAM dies, and I/O dies. The I/O dies will facilitate support for DDR5 memory, PCIe 6.0 connectivity, and CXL 3.0 for linking various accelerators and extenders.
Timeline and Future Prospects
Japan’s next-generation supercomputer is expected to come online in 2030. It is expected to use processors with even more cores and more advanced connectivity than Monaka. The supercomputer’s development could also involve using a 1nm-class process technology or perhaps an even more advanced node.
The successful development of this supercomputer will not only cement Japan’s position as a leader in HPC and have significant implications for various fields, including scientific research, data science, and AI. The potential applications of such a powerful system are vast, ranging from simulating complex systems to analyzing large datasets and developing more sophisticated AI models.
External Link: Click Here For More
