Fugaku Remakes Exascale Computing In Its Own Image – The Next Platform
When originally conceived, Japans Post-K supercomputer was supposed to be the countrys first exascale system. Developed by Fujitsu and the RIKEN Center for Computational Science, the system, now known as Fugaku, is designed to be two orders of magnitude faster than its predecessor, the 11.3-petaflops (peak) K computer. But a funny thing happened on the way to exascale. By the time the silicon dust had settled on the A64FX, the chip that will power Fugaku, it had morphed into a pre-exascale system.
The current estimate is that the RIKEN-bound supercomputer will top out at about 400 peak petaflops at double precision. Given that the system has to fit in a 30 MW to 40 MW power envelope, thats about all you can squeeze out of the 150,000 single-socket nodes that will make up the machine. Which is actually rather impressive. The A64FX prototype machine, aka micro-Fugaku, is currently the most energy-efficient supercomputer in the world, delivering 16.9 gigaflops per watt. However, extrapolating that out to an exaflop machine with those same (or very similar) processors would require something approaching 60 MW to 80 megawatts.
But according to Satoshi Matsuoka, director of the RIKEN lab, the performance goal of achieving two orders of magnitude improvement over the K computer will be achieved from an application performance perspective. That was the plan from the beginning, Matsuoka tells The Next Platform.
To imply that 100-fold application boost amounts to exascale capability is a bit of stretch, but if Fugaku effectively performs at that level relative to the performance of applications on the K machine, that is probably more important to RIKEN users. It should be pointed out that not all applications are going to enjoy that magnitude of speedup. The table below illustrates the expected performance boost for nine target applications relative to the K computer.
Even though Fugaku has only 20 times the raw performance and energy efficiency of its predecessor, the 100X performance improvement is the defining metric, says Matsuoka. That kind of overachievement (again, on some codes) is the result of certain capabilities baked into the A64FX silicon, in particular the use of Arms Scalable Vector Extension (SVE), which provides something akin to an integrated 512-bit-wide vector processor on-chip, delivering about three teraflops of peak oomph.
Perhaps even more significant is the 32 GB of HBM2 stacked memory glued onto the A64FX package, which delivers 29X the bandwidth of the memory system on the K computer. The choice to dispense with conventional memory and go entirely with HBM2 was the result of the recognition that many HPC applications these days are memory-bound rather than compute bound. In fact, achieving better balance between flops and memory bandwidth was a key design point for Fugaku. The compromise here is that 32 GB is not much capacity, especially for applications that need to work with really large datasets.
The other aspect of Fugaku that could earn it exascale street cred is in the realm of lower precision floating point. Although the system will deliver 400 peak petaflops at double precision (FP64), it will provide 800 petaflops at single precision (FP32) and 1.6 exaflops at half precision (FP16). The half precision support alludes to AI applications that can make extensive use of 16-bit floating point arithmetic to build artificial neural networks. Fugaku may even manage to hit an exaflop or better on the HPL-AI benchmark, which makes extensive use of FP16 to run for High Performance Linpack (HPL).
When run on the 200 petaflops Summit machine at Oak Ridge National Laboratory, HPL-AI delivered 445 petaflops on Linpack, which was three times faster than the result performed solely with FP64. More to the point, if the same iterative refinement techniques using FP16 can be used on real applications, its possible that actual HPC codes can be accelerated to exascale levels.
The more straightforward use of reduced precision math, employing both FP16 and FP32, is for training AI models. Again, work on Summit proved that lower precision math could attain exascale-level computing on these machines. In this particular case, developers employed the Tensor Cores on the systems V100 GPUs to use a neural network to classify extreme weather patterns, achieving peak performance of 1.13 exaops and sustained performance of 0.999 exaops.
Whether reduced precision exaflops or exaops qualifies as exascale computing is a semantic exercise more than anything else. Of course, thats not going to be very satisfying for computer historians or even for analysts and journalists attempting to track HPC capability in real-time.
But perhaps thats as it should be as. The attainment of a particular peak performance or Linpack performance numbers does little to inform the state of supercomputing. And given the increasing importance of AI workloads, which are not based on 64-bit computing, its not surprising that HPC is moving away from these simplistic measures. The expected emergence of neuromorphic and quantum computing in the coming decade will further muddy the waters.
That said, users will continue to rely primarily on 64-bit flops to run HPC simulations, which will continue to be heavily used by the scientists and engineers for the foreseeable future.
With that in mind, RIKEN is already planning for its post-Fugaku system, which Matsuoka says is tentatively scheduled to make its appearance in 2028. According to him, RIKEN is planning to do an analysis on how it can build something 20X more powerful than Fugaku. He says the challenge is that current technologies wont extrapolate to such a system in any practical manner. Which once again means they will have to innovate at the architectural level, but this time without the benefit of Moores Law.
Featuring highlights, analysis, and stories from the week directly from us to your inbox with nothing in between.
Subscribe now
View original post here:
Fugaku Remakes Exascale Computing In Its Own Image - The Next Platform
- What Is Fault-Tolerant Quantum Computing, and When Will Industry Get It? - DirectIndustry e-Magazine - November 19th, 2026 [November 19th, 2026]
- From Lab to Live: Inside the Practical Realities of Deploying Quantum Computing at Data Center Scale - The Quantum Insider - November 19th, 2026 [November 19th, 2026]
- Securities Regulators Warn That Slow Progress on Quantum Security is a Mistake - The Quantum Insider - November 19th, 2026 [November 19th, 2026]
- FBI Leads Federal Team Protecting Quantum Research From Foreign Threats - ExecutiveGov - November 19th, 2026 [November 19th, 2026]
- Quandela and Hermes-Epitek Join Forces to Advance Photonic Quantum Technologies - The Quantum Insider - November 19th, 2026 [November 19th, 2026]
- memQ and Infleqtion Partner to Advance Interoperability in Networked Quantum Computing - citybiz - November 19th, 2026 [November 19th, 2026]
- Can Quantum Computers Improve Delivery? Our Journey from the NEDO Challenge to an Excellent Research Award - - November 19th, 2026 [November 19th, 2026]
- IQM and AMD Demonstrate Reproducible Hybrid QPU-GPU Chemistry Workflow Using ROCm - Quantum Computing Report - November 19th, 2026 [November 19th, 2026]
- BMO Achieves Quantum Utility In Machine Learning With IBM Computers - Quantum Zeitgeist - November 19th, 2026 [November 19th, 2026]
- To the ramparts over quantum threat: IOSCO - Investment Executive - November 19th, 2026 [November 19th, 2026]
- Quantinuum Says Early Quantum Work Builds Skills For The Future - Quantum Zeitgeist - November 19th, 2026 [November 19th, 2026]
- Senate Bill Would Expand Federal Quantum Research and Security - Legis1 - November 19th, 2026 [November 19th, 2026]
- Quantum tunnelling: Diamond physics could lead to more reliable quantum computers and medical sensors - Open Access Government - November 19th, 2026 [November 19th, 2026]
- Bank of Montreal achieves quantum utility amid race to deploy the technology - The Globe and Mail - November 19th, 2026 [November 19th, 2026]
- Omniva Wins QEst Hack 2026 with Quantum Parcel Routing Project - The Quantum Insider - November 19th, 2026 [November 19th, 2026]
- Toyota Is Betting $1 Million a Year on Quantum Computing to Build Better Cars - Inc.com - November 19th, 2026 [November 19th, 2026]
- Honoring the 2026 HPC Legends at SC26 - HPCwire - November 19th, 2026 [November 19th, 2026]
- Scalable photonic quantum computer provides new insights into the fundamentals that shape its performance - Phys.org - November 19th, 2026 [November 19th, 2026]
- A game you cannot win unless you have a quantum computer - Google Research - November 19th, 2026 [November 19th, 2026]
- Light-based Quantum Computer May Reveal Which Resources Boost Its Power. - Quantum Zeitgeist - November 19th, 2026 [November 19th, 2026]
- Mid-sized businesses can't leave quantum readiness to their suppliers - Computer Weekly - November 15th, 2026 [November 15th, 2026]
- The Quantum Threat is Already Here. The Quantum Computer Isn't. - WashingtonExec - November 13th, 2026 [November 13th, 2026]
- Correcting quantum computer errors may be more complicated than expected - New Scientist - November 13th, 2026 [November 13th, 2026]
- Encryption That Protects Your Bank Account Could Fall to a Quantum Computer in 3 Days, Study Suggests - ScienceAlert - November 5th, 2026 [November 5th, 2026]
- Quantum computer errors slashed by more than 50% with single check - Interesting Engineering - November 5th, 2026 [November 5th, 2026]
- Japan Just Powered Up a Quantum Computer Built From Single Atomsand It May Not Need a Giant Freezer to Run - Popular Mechanics - November 3rd, 2026 [November 3rd, 2026]
- This Brilliant Quantum Computing Stock Could Double in a Year - Yahoo Finance - November 1st, 2026 [November 1st, 2026]
- Our estimate for factoring RSA-2048 with cat qubits: pushing the frontier between size and speed - Alice & Bob - November 1st, 2026 [November 1st, 2026]
- Prediction: This Is the Quantum Computing Stock Most Likely to Still Be Standing in 2030 - The Motley Fool - November 1st, 2026 [November 1st, 2026]
- This Brilliant Quantum Computing Stock Could Double in a Year - The Motley Fool - November 1st, 2026 [November 1st, 2026]
- Quantum: 12.9% of technology signals are already at operational level - EU Science Hub - November 1st, 2026 [November 1st, 2026]
- 3 Quantum Computing Stocks to Buy Before They Soar in 2027 - The Motley Fool - November 1st, 2026 [November 1st, 2026]
- Are These 3 Quantum Computing Stocks Worth Buying Before 2027? - Yahoo Finance - November 1st, 2026 [November 1st, 2026]
- Are These 3 Quantum Computing Stocks Worth Buying Before 2027? - The Motley Fool - November 1st, 2026 [November 1st, 2026]
- This Brilliant Quantum Computing Stock Could Double in a Year - AOL.com - November 1st, 2026 [November 1st, 2026]
- The Quantum Threat Is Closer Than Finance Thinks - Crowdfund Insider - November 1st, 2026 [November 1st, 2026]
- The Continuum+: Who will capture the quantum advantage? - Simmons & Simmons - November 1st, 2026 [November 1st, 2026]
- Coinbase Is Racing to Quantum-Proof Bitcoin Before It Has To - Startup Fortune - November 1st, 2026 [November 1st, 2026]
- 3 Quantum Computing Stocks to Buy Before They Soar in 2027 - The Globe and Mail - November 1st, 2026 [November 1st, 2026]
- Coinbase CEO Brian Armstrong Says Quantum Computing Is a Bigger Threat to Crypto Than AI: When, Not If - Benzinga - November 1st, 2026 [November 1st, 2026]
- intelligence bets on quantum with deal for Israeli software startup Classiq - The Jerusalem Post - October 28th, 2026 [October 28th, 2026]
- Quantum Computings Energy Bill Is Coming, and We Havent Priced It In - EE Times - October 28th, 2026 [October 28th, 2026]
- The Genesis Mission and the Future of the Future of Computing - Issues in Science and Technology - October 28th, 2026 [October 28th, 2026]
- Intels $2.3 Billion Bet on a Quantum Future is a Decade in the Making - The Next Platform - October 28th, 2026 [October 28th, 2026]
- Giving quantum computers a way to pick up where they left off - Florida International University - October 28th, 2026 [October 28th, 2026]
- Guest Post: The Next Threat to Encryption May Not Need a Quantum Computer - The Quantum Insider - October 28th, 2026 [October 28th, 2026]
- How EPB and IonQ Use Quantum to Optimize the Power Grid - IonQ - October 28th, 2026 [October 28th, 2026]
- We're a Step Closer to Quantum Computers That Can Correct Their Own Mistakes - ScienceAlert - October 28th, 2026 [October 28th, 2026]
- Fairfax Schools to Bring Quantum Computing to K-12 - GovTech - October 28th, 2026 [October 28th, 2026]
- Quantum computing and the burden of proof - Computing UK - October 28th, 2026 [October 28th, 2026]
- Can Classiq Make Intel the Cadence of Quantum Computing? - The Futurum Group - October 28th, 2026 [October 28th, 2026]
- Uncovering the Hidden Disorder in Silicon Quantum Computers - Department of Energy (.gov) - October 28th, 2026 [October 28th, 2026]
- Beyond Binary Quantum Computing: Exploring Higher-Dimensional Qudits - The Quantum Insider - October 28th, 2026 [October 28th, 2026]
- SoftBank teams up with five Japanese firms to build quantum-based platform - Light Reading - October 28th, 2026 [October 28th, 2026]
- Intel Just Bought the Quantum Toolbox, Not the Magic Box - Money Morning - October 28th, 2026 [October 28th, 2026]
- D-Wave partners with LG CNS to advance quantum use cases in Asia - Seeking Alpha - October 28th, 2026 [October 28th, 2026]
- Quantum Computing Inc. Announces Participation in Upcoming November Investor Conferences - PR Newswire - October 28th, 2026 [October 28th, 2026]
- Dell brings post-quantum cryptography to the root of trust - SiliconANGLE - October 28th, 2026 [October 28th, 2026]
- Google Just Solved a Major Quantum Computing Problem. Heres Why Investors Still Have a 10-Year Wait. - Yahoo Finance - October 28th, 2026 [October 28th, 2026]
- LG CNS and D-Wave Announce Strategic Relationship to Advance Quantum Optimization Adoption in South Korea - Yahoo Finance - October 28th, 2026 [October 28th, 2026]
- D-Wave Quantum delivers first annealing quantum computer to Italy as part of SQT deal (QBTS:NASDAQ) - Seeking Alpha - October 17th, 2026 [October 17th, 2026]
- IonQ tumbles as it buys cybersecurity firm Elisity in $1.3B all-stock deal (IONQ:NYSE) - Seeking Alpha - October 12th, 2026 [October 12th, 2026]
- IonQ to acquire cybersecurity firm Elisity in an all-stock deal for up to $1.3 billion - CNBC - October 12th, 2026 [October 12th, 2026]
- Tuning the Field: Notes from the NYC Quantum Summit - New York University - October 12th, 2026 [October 12th, 2026]
- Scientists make major step towards fixing the big problem with quantum computers - The Independent - October 12th, 2026 [October 12th, 2026]
- IonQ Agrees to Buy Elisity in Cybersecurity Deal Worth Up to $1.3 Billion - The Quantum Insider - October 12th, 2026 [October 12th, 2026]
- A Practical Path to Post-Quantum Readiness in Operational Technology - Foundation for Defense of Democracies - October 12th, 2026 [October 12th, 2026]
- DARPA selects IBM and IonQ for next phase in development of fault-tolerant quantum computer - Seeking Alpha - October 12th, 2026 [October 12th, 2026]
- Can Quantum Computers Break Bitcoin? An Ethereum Researcher Just Told Big Holders to Prepare for Bunker Mode - 24/7 Wall St. - October 12th, 2026 [October 12th, 2026]
- Why IonQ Is Betting Big on Quantum Cybersecurity With $1.3 Billion Elisity Acquisition - Barron's - October 12th, 2026 [October 12th, 2026]
- Quantum computing is already here - Fast Company - October 6th, 2026 [October 6th, 2026]
- U.S. GAO - Quantum Computing: Federal Actions Needed to Prepare for Emerging Cyber Threat - U.S. Government Accountability Office (.gov) - October 6th, 2026 [October 6th, 2026]
- Bank of America Put a $60 Price Target on IonQ. The Quantum Bet Is Picking Up Speed. - Barchart.com - October 6th, 2026 [October 6th, 2026]
- Quantum Computing Inc. to Attend the Maxim Growth Summit 2026 - PR Newswire - October 6th, 2026 [October 6th, 2026]
- Quantinuum partners with University of Western Australia to expand quantum computing access - Data Center Dynamics - October 6th, 2026 [October 6th, 2026]
- Infleqtion and Partners Build Optical Cavity Prototype to Shrink Quantum Sensors - HPCwire - October 6th, 2026 [October 6th, 2026]
- University of Arkansas, D-Wave to partner on quantum computing project - Talk Business & Politics - October 6th, 2026 [October 6th, 2026]
- planqc and LOGIQC Selected for Germanys Quantum Computing Competition - The Quantum Insider - October 6th, 2026 [October 6th, 2026]
- Xanadu Soars 16% on Multi-Year GlobalFoundries Deal to Deliver Quantum Computing at Scale - TIKR.com - October 6th, 2026 [October 6th, 2026]
- Rigetti Computing vs. D-Wave Quantum: Which Quantum Computing Stock Is the Better Buy? - The Motley Fool - October 6th, 2026 [October 6th, 2026]