TOKYO, August 23, 2024 – A paper on Preferred Networks (PFN)’s second-generation AI processor MN-Core™ 2 has been accepted to Hot Chips 2024, an international conference on high-performance chips. Hosted by IEEE (Institute of Electrical and Electronics Engineers), the hybrid event will be held between Sunday, August 25 and Tuesday, August 27 at the Stanford University campus and online. PFN’s CTO of Computer Architecture and Kobe University professor Junichiro Makino will present the study on August 27.

MN-Core 2 server blade

PFN at Hot Chips 2024

Time and date: 4:15-6:16 pm on Tuesday, August 27 during the High-Performance Processors Part 3 session in the Advance Program
Place: Memorial Auditorium, Stanford University and online
Speaker::Junichiro Makino, CTO of Computer Architecture, Preferred Networks
Title: MN-Core 2: Second-generation processor of MN-Core architecture for AI and general-purpose HPC applications
Outline: PFN and Kobe University are co-developing the MN-Core series of processors with a unique architecture optimized for deep learning. The first generation of the series demonstrated high energy efficiency, powering the supercomputer that topped the Green500 ranking three times. The second-generation processor MN-Core 2 maintained the energy efficiency with a chip size reduced to one fifth of the first generation, resulting in greater areal efficiency and less manufacturing costs. Pilot tests showed MN-Core 2 showed high performance in general-purpose HPC as well as machine learning. The next generation of MN-Core processors is under development at PFN.
Registration: Attendance to Hot Chips requires paid registration at: https://hotchipssymposium.regfox.com/hot-chips-2024

About MN-Core Series™

Jointly developed by PFN and Kobe University, the MN-Core™ series of processors are optimized for matrix operations that are essential for deep learning. To maximize the number of arithmetic units on the MN-Core chip, other functions such as network control circuits, cache controllers and command schedulers, are incorporated in the compiler software rather than the hardware, achieving highly efficient deep learning operations while keeping the costs down. MN-3, PFN’s supercomputer powered by MN-Core, has topped the Green500 list of the world’s most energy-efficient supercomputers three times between June 2020 and November 2021. The second-generation MN-Core 2 started operating in 2023. PFN plans to provide MN-Core 2’s computing power to external parties in 2024.