 | |
| Active | From 2021 |
|---|---|
| Sponsors | MEXT |
| Operators | Riken |
| Location | Riken Center for Computational Science (R-CCS) |
| Architecture |
|
| Operating system | CustomLinux-based kernel |
| Memory | HBM2 32 GiB/node |
| Storage | |
| Speed | 442 PFLOPS (perTOP500 Rmax), after upgrade; higher 2.0 EFLOPS on a different mixed-precision benchmark |
| Cost | US$1 billion (total programme cost)[2][3] |
| Ranking | TOP500: No. 4,June 2024 |
| Purpose | Scientific research |
| Legacy | TOP500 No.1, June 2020 – June 2022 |
| Website | www |
| Sources | Fugaku System Configuration |
Fugaku(Japanese:富岳) is apetascalesupercomputer at theRiken Center for Computational Science inKobe,Japan. It started development in 2014 as the successor to theK computer[4] and made its debut in 2020. It is named after an alternative name forMount Fuji.[5]
It became the fastest supercomputer in the world in the June 2020TOP500 list[6] as well as becoming the firstARM architecture-based computer to achieve this.[7] At this time it also achieved 1.42 exaFLOPS using the mixed fp16/fp64 precision HPL-AI benchmark. It started regular operations in 2021.[8]
Fugaku was superseded as the fastest supercomputer in the world byFrontier in May 2022.[9]
The supercomputer is built with theFujitsu A64FX microprocessor. This CPU is based on theARMversion 8.2A processor architecture, and adopts theScalable Vector Extensions for supercomputers.[10] Fugaku was aimed to be about 100 times more powerful than theK computer (i.e. a performance target of1 exaFLOPS).[11][12]
The initial (June 2020) configuration of Fugaku used 158,976 A64FX CPUs joined using Fujitsu's proprietarytorus fusion interconnect.[7] An upgrade in November 2020 increased the number of processors.[13]
Fugaku uses a "light-weight multi-kernel operating system" namedIHK/McKernel. The operating system uses bothLinux and the McKernel light-weightkernel operating simultaneously and side by side. The infrastructure that both kernels run on is termed theInterface for Heterogeneous Kernels (IHK). The high-performance simulations are run on McKernel, with Linux available for all otherPOSIX-compatible services.[14][15][16]
Fugaku uses a three-tiered storage system to provide parallel storage to the compute nodes. The first-level LLIO storage is an NVM-based file I/O accelerator co-developed by Fujitsu and RIKEN that is allocated on per-job storage basis to the compute nodes for storing temporary data with low latency. The LLIO system stages data in and out of a second-level Fujitsu Exascale File System (FEFS), which uses disk-based storage based onLustre software to provide a large persistent high-performance filesystem,[17][18] and a tape-based archive to store a large volume of infrequently accessed data.
Besides the system software, the supercomputer has run many kinds of applications, including several benchmarks. Running the mainstreamHPL benchmark, used byTOP500, Fugaku is at petascale and almost halfway to exascale. Additionally, Fugaku has set world records on at least three other benchmarks, includingHPL-AI; at 2.0 exaflops, the system has exceeded the exascale threshold for the benchmark.[19] A description of that benchmark is as follows:
The solver method of choice is a combination ofLU factorization and iterative refinement performed afterwards to bring the solution back to 64-bit accuracy. The innovation of HPL-AI lies in dropping the requirement of 64-bit computation throughout the entire solution process and instead opting for low-precision (likely 16-bit) accuracy for LU, and a sophisticated iteration to recover the accuracy lost in factorization.[20]
The reported initial performance of Fugaku was aRmax of 416 petaFLOPS in theFP64high performance LINPACK benchmark used by the TOP500.[7] After the November 2020 upgrade in the number of processors, Fugaku's performance increased to a Rmax of 442 petaFLOPS.[13]
In 2020, Fugaku also attained top spots in other rankings that test computers on different workloads, includingGraph500, HPL-AI, andHPCG benchmark. No previous supercomputer has ever led all four rankings at once.[21]
After a hardware upgrade, as of November 2020, "Fugaku increased its performance on the new mixed precisionHPC-AI benchmark to 2.0 exaflops, besting its 1.4 exaflops mark recorded six months ago. These represent the first benchmark measurements above one exaflop for any precision on any type of hardware." (a 42% increase)[22] Interestingly, the Arm A64FX core-count was only increased by 4.5%, to 7,630,848, but the measured performance rose much more on that benchmark (and the system does not use other compute capabilities, such asGPUs), and a little more on TOP500, or by 6.4%, to 442 petaflops, a new world record[23] and widening the gap to the next computer by that much. For the HPCG benchmark, it is 5.4 times faster, at 16.0 HPCG-petaflops, than the number two system,Summit,[24] which happens to also be second on TOP500.
As of November 2020, Fugaku's performance surpassed the combined performance of the next 4 supercomputers on the TOP500 list, and surpassed the remaining top-10 computers on the HPCG benchmark by a margin of 45%.[25]
On May 23, 2019, Riken announced that the supercomputer was to be named Fugaku.[26] In August 2019, the logo for Fugaku was unveiled; it depictsMount Fuji, symbolising "Fugaku's high performance" and "the wide range of its users".[4][27] In November 2019, the prototype of Fugaku won first place in theGreen500 list.[28][29] Shipment of the equipment racks to the Riken facility began on December 2, 2019,[30] and was completed on May 13, 2020.[31] In June 2020, Fugaku became the fastest supercomputer in the world in the TOP500 list, displacing the IBM Summit.[7]
Fugaku has been used for research on masks related to theCOVID-19 pandemic.[32][33]
In 2023, Fugaku has been used to develop Japanese language Generative AI Models by theTokyo Institute of Technology, Riken Research Institute,Fujitsu Ltd andTohoku University.[34]
In 2018, Nikkei reported the programme would cost¥130 billion (c. US$1 billion).[3][8]
| Name | Start year | End year | Performance (PFLOPS)[note 1] | Cost (million USD) (not inflation adjusted) | TOP500 ranking | CPU/GPU vendor | CPU | OS |
|---|---|---|---|---|---|---|---|---|
| Fugaku | 2020 | — | 442[35] | 1213[3][note 2] | June 2020 to November 2021 1st[35] | Fujitsu | A64FX | Linux (RedHat 8) and McKernel |
| Summit | 2018 | — | 148 | 300[36] | June 2018 to November 2019 1st | IBM,Nvidia | POWER9,Tesla | Linux (RedHat) |
| Sierra | 2018 | — | 94 | November 2018 to November 2019 2nd | ||||
| Sunway TaihuLight | 2016 | — | 93 | 280[37] | June 2016 to November 2017 1st | NRCPC | Sunway SW26010 | Linux (Raise) |
| K | 2011 | 2019 | 10 | 1045[38] | June 2011 – November 2011 1st | Fujitsu | SPARC64 VIIIfx | Linux |
(in English)| Records | ||
|---|---|---|
| Preceded by IBM Summit 148.6 petaFLOPS | World's most powerful supercomputer June 2020 – May 2022 0.54 exaFLOPS | Succeeded by HPE Frontier 1.1 exaFLOPS |
| Divisions and subsidiaries |
|  | ||||
|---|---|---|---|---|---|---|
| Joint ventures and shareholdings |
| |||||
| Products, services and standards |
| |||||
| People | ||||||
| Places | ||||||
| Other | ||||||
