Fujitsu’s follow-up to mighty supercomputer chip will power data centers

Fujitsu has announced the next chip to follow the A64FX, the ARM processor that is used in one of the most powerful supercomputers in the world, the Fugaku.

The new chip from the Japanese tech giant was announced in a presentation as part of the Fujitsu ActivateNow: Technology Summit, held at the Computer History Museum in Mountain View, California. Fujitsu CTO Vivek Mahajan said that it will be an “Arm-based CPU for Next-gen DC” set for release sometime in 2028.

With the working title of ‘MONAKA’, Fujitsu told The Register that its focus was on making a high performance chip with better energy efficiency, saying that it wanted to contribute to “the realization of a carbon-neutral and sustainable society.”

Energy savings

The new processor is said to be able to increase HPC workloads, as well as deliver optimal performance for AI and data analysis applications, all whilst providing “overwhelming energy efficiency” over coterminous rivals. Fujitsu claims it will have 1.7 times the application performance with double the performance per watt. 

MONAKA is part of the program run by the New Energy and Industrial Technology Development Organization (NEDO) research agency in Japan. It’s goal is to increase energy savings by 40% within the nation’s data centers come 2030.

Fujitsu is also contributing to this with other developments besides MONAKA, such as low power consumption accelerators, smart network interface cards (NIC) that use photonics, and disaggregation technology.

It is speculated that the new chip will therefore be more akin to those used in servers catering to cloud storage, cloud hosting and colocation providers. However, these chips are less concerned with saving energy as such rather than achieving full utilization. 

“The next-generation DC CPU (MONAKA) that we’re developing will have a wider range of features and will prove more energy efficient,” a Fujitsu spokesperson told The Register. “The range of potential applications is wider than that of the A64FX, which has special characteristics (e.g., interconnects) specific to Fugaku.”

Details are scarce at this point, it seems likely that MONAKA will have the same standout features as the A64FX, such as 28Gbps Tofu-D interconnect, high-speed HBM2 stacked memory, and 512-bit Scalable Vector Extensions.

Go to Source