Leave it to Huawei, the Chinese conglomerate that has its fingers in pies from wearables to telecommunications infrastructure, to kick off the 2019 Consumer Electronics Show with a new record. The company today unveiled what it claims is the industry’s highest-performance Arm-based processor, dubbed the Kunpeng 920.
It’s purpose-built for applications like big data processing and distributed storage, William Xu, director of the company’s board and chief strategy marketing officer, said during a press conference in Shenzhen, China.
“Huawei has continuously innovated in the computing domain in order to create customer value,” he added. “We believe that with the advent of the intelligent society the computing market will see continuous growth in the future … We will work with global partners in the spirit of openness, collaboration, and shared success to drive the development of the Arm ecosystem and expand the computing space and embrace a diversified computing era.”
The Kunpeng 920, much like Huawei’s recently launched HiSilicon Kirin 980, is manufactured on a 7-nanometer process and was designed in-house. Much of its performance gains come from optimized branch prediction algorithms and an increased number of OP units, along with an improved memory subsystem architecture, according to Huawei.
June 5th: The AI Audit in NYC
Join us next week in NYC to engage with top executive leaders, delving into strategies for auditing AI models to ensure fairness, optimal performance, and ethical compliance across diverse organizations. Secure your attendance for this exclusive invite-only event.
The Arm processor’s 64 cores are clocked at 2.6GHz and paired with 8-channel DDR4 memory at a bandwidth that’s 46 percent faster than some competing systems. Other onboard accouterments include two 100G RoCE ports and support for PCIe Gen4 and CCIX.
In SPECint — a benchmark suite of 12 programs designed to test integer performance — the Kunpeng 920 scored over 930, or almost 25 percent higher than the industry benchmark, while drawing 30 percent less power than “that offered by industry incumbents.”
The previous record holder was Fujitsu’s 7-nanometer A64X, which can achieve up to 2.7 teraflops performance per chip.
During Sunday’s briefing, Huawei also took the wraps off new servers in its flagship series — TaiShan — powered by the Kunpeng 920. (For point of reference, TaiShan servers provide the hardware backbone for Huawei’s elastic cloud, bare metal, and cloud phone services.) They’re specially designed for high-performance, high-efficiency scenarios, and to that end have a resource scheduler that delivers up to a 20 percent computing performance boost.
Three new TaiShan server models — one with a focus on storage, another on high density, and a third on both — will be made available sometime in 2019.
“With Kirin 980, Huawei has taken smartphones to a new level of intelligence. With products and services (e.g., Huawei Cloud) designed based on Ascend 310, Huawei enables inclusive AI for industries,” Xu added. “Today, with Kunpeng 920, we are entering an era of diversified computing embodied by multiple cores and heterogeneity. Huawei has invested patiently and intensively in computing innovation to continuously make breakthroughs. We will work with our customers and partners to build a fully connected, intelligent world.”
