CHIP DESIGNER ARM has set its sights on rivals IBM, Intel and Nvidia in the supercomputer market, having announced Scalable Vector Extensions (SVE) technology being developed for the ARMv8-A architecture.
The SVE architecture is already set to be used by Fujitsu for the Post-K supercomputer being built for Japan's RIKEN Advanced Institute for Computational Science. The machine should be world's most powerful when it goes live in 2020.
SVE was detailed at the Hot Chips conference in Cupertino, and is capable of handling vectors from 128 to 2,048 bits in length. It is intended for supercomputer makers like Fujitsu to help them adopt ARM products and deploy them in the world's biggest and most powerful computers.
The technology is a flexible extension to the ARM instruction set that can shift vector calculations from software to hardware. The scheduler arranges calculations depending on the available hardware.
ARM engineers will shortly submit patches to the GNU Compiler Collection and the LLVM compiler infrastructure project to support SVE auto-vectorisation. This will enable developers to take advantage of the technology without having to customise applications.
Compiled programs will also be portable and will be able to run on any microprocessor with SVE without needing to recompile the source code.
Fujitsu chose ARM in July for the Post-K supercomputer after deciding to shift from the 2GHz Sun Sparc64 cores used in the K supercomputer. The current K supercomputer ranks fifth in the latest Top 500 list of the world's most powerful supercomputers.
If everything goes to plan, Fujitsu's Post-K machine will become the world's most powerful supercomputer, capable of 1,000 petaflops.
ARM continues to develop parts that may enable it to take on Intel in servers and high-end computing. The company explained that cloud computing workloads, for example, require highly scalable configurations where the key is the interconnect, rather than the raw power of the microprocessors.
The firm's roadmap foresees ARM partners using the company's products first in storage and web-serving workloads, before shifting up to data analytics, networking, collaborative platforms and systems management.
"A vast number of the workloads being deployed on servers are things like storage, web applications and networking," Lakshmi Mandyam, director of server systems and ecosystems at ARM, told the INQUIRER.
"Those kinds of things make up a significant proportion of the server market. They are also some of the fastest growing applications in terms of the workloads being deployed.
"These workloads are 'scale out' workloads that lend themselves to running across many different nodes. Each one of those nodes doesn't have to be very compute intensive.
"Those workloads are actually more about the input-output and memory performance than they are about the compute performance [of the microprocessor]."
ARM, therefore, sees the cloud computing model becoming more like the high-performance computer market, where the CPUs of the world's most powerful supercomputers don't go for out-and-out raw speed.
Instead, the machines in the top 10 typically weigh in at between 1.45GHz and 2.6GHz, but achieve their computational power via the number of cores and the speed and scalability of the interconnects.
The Sunway TaihuLight, the world's most powerful supercomputer, has 10,649,600 cores, three times that of its predecessor. Typical machines in the top 10 have anything from 115,984 to 1,572,864. µ
Block to the future
My bay or the highway
Firm's first high-end speaker gets the thumbs up from us
Those who escape fines will be very Jammy Dodgers