For years, x86 and Linux processors have ruled supercomputing. Linux still runs 500 of the world’s TOP500 supercomputers. For almost the same time, x86 CPUs have dominated supercomputers, until now. On June 22, Japan’s Fugaku supercomputer, powered by Fujitsu’s 48-core A64FX SoC and powered by Red Hat Enterprise Linux (RHEL), became the first ARM-powered supercomputer to be named the world’s fastest computer.
By winning over others, he wasn’t even close. Fugaku had a High Performance Linpack (HPL) result of 415.5 petaflops, beating the IBM Summit system second by a factor of 2.8x.
Fujitsu has been working on creating ultra-high-speed silicon since moving from the obsolete SPARC architecture to the ARM-based A64FX. This was the first CPU to adopt the ARMv8-A Scalable Vector Extension (SVE). This is an ARM extension to your instruction set specifically for supercomputers. Fujitsu worked with ARM to develop the A64FX.
As fast as it is, in simple or reduced precision, used in machine learning and artificial intelligence applications, Fugaku’s peak performance made it the first supercomputer to break the 1,000-petaflops barrier, or 1 exaflop. Fugaku is installed at the RIKEN Center for Computational Sciences (R-CCS) in Kobe, Japan.
AMD and Cray are working with the US Department of Energy. USA Along with the Oak Ridge National Laboratory to build a supercomputer, Frontier, which uses EPYC CPU and Radeon Instinct GPU to officially destroy the exaflop wall in 2021. At this point, it looks like it will be a run between Fugaku and Frontier to run faster than an exaflop.
For now, Summit, with its Power9 CPUs and NVIDIA Tesla V100 GPUs, is the second fastest supercomputer. It offers 148.8 petaflops in HPL.
At No. 3, we found the Lawrence Livermore National Laboratory (LLNL) Sierra with 94.6 petaflops in HPL. Like Summit, it uses Power9 CPU and NVIDIA Tesla V100 GPU. Sunway TaihuLight of China is number 4, with an HPL mark of 93 petaflops. Fully powered by Sunway 260-core SW26010 processors. In the N. # 5, we find China’s Tianhe-2A (Milky Way-2A), with an HPL performance of 61.4 petaflops from a hybrid architecture combining Intel Xeon CPUs and custom Matrix-2000 coprocessors.
Intel’s x86 architecture is the heart of just one of the top five supercomputers. But it is still the top processor architecture. You’ll find it at 481 of the TOP500 systems. Of these, Intel claims 469, with AMD installed at 11 and Hygon at the rest. ARM processors are present in just four TOP500 systems, three of which use the new Fujitsu A64FX processor, and the other works with Marvell’s ThunderX2 processor. But being number 1 counts a lot.
China continues to dominate the TOP500 in systems, as 226 Chinese supercomputers are on the list. The United States is number 2 with 114 systems; Japan is third with 30; France is 18; and Germany claims 16. In terms of raw computing power, the United States outperforms China in aggregate list performance, with 644 petaflops at 565 petaflops. Small but powerful Japan, despite its much smaller system count, offers a total of 530 petaflops.
Unsurprisingly from that breakdown, Chinese manufacturers dominate the list: Lenovo (180), Sugon (68), and Inspur (64) account for 312 of the 500 installed systems. Of the US providers, HPE claims 37 systems, while Cray / HPE has 35 systems. Fujitsu is represented by just 13 systems, but thanks to Fugaku’s victory, the company tops the list in aggregate performance with 478 petaflops. Lenovo ranked second in performance with 355 petaflops.
Finally, as has been true since 2017, the world’s 500 fastest supercomputers run Linux. When it comes to high performance, it’s a Linux world.