Published in


Huawei 7nm Kirin 810 Beats Snapdragon 855 and Kirin 980 on AI Benchmark Test

At Huawei’s Nova 5 Series smartphone launch event yesterday the Chinese tech giant unveiled its new 7nm SoC, the Kirin 810. The chipset features a new AI accelerator built on Huawei’s Da Vinci architecture that delivers better AI performance than SoCs such as Qualcomm’s Snapdragon 855 processor and even Huawei’s top-end Kirin 980 chip.

The Kirin 810 is Huawei’s second 7nm SoC, following the Kirin 980 introduced last year. Its transistor density is 64 percent higher than a 10nm processor, and power efficiency is 28 percent higher.

Huawei has packed different chipset series into its smartphones: the mid-range Kirin 710 is a cost-effective 64-bit chipset; while the high-end Kirin 980 is used in Huawei’s flagship smartphones like the P30 and Mate 20 Pro. The Kirin 810 is positioned between these two, and competes with other high-performance mobile SoCs like the Qualcomm Snapdragon 730.

The Kirin 810 features two Arm Cortex A76 CPUs running up to 2.23GHz, six Arm Cortex A55 cores at 1.88GHz, and a Mali-G52 GPU running at 820MHz. The chipset incorporates the “HiAI” computing architecture Huawei introduced in 2017 for AI acceleration.

A feature of the Kirin 810 that will be of interest to the machine learning community and users alike is the new embedded neural processing unit (NPU). Based on Huawei’s homegrown architecture, Da Vinci, the NPU delivers tensor computation, high efficiency, and high performance under half precision FP16 and INT8 data types. In the Kirin 970 and 980 Huawei used an NPU designed by Chinese AI chip unicorn Cambricon.

In the AI Benchmark test — a highly recognized evaluation metric that tests real-world performance of mobile SoCs — the Kirin 810-powered smartphone Nova 5 scored 32280, outperforming the Snapdragon 855-powered Asus Zenfone 6 (25428). In the AnTutu benchmark test the Nova 5 scored about 237k, surpassing the Kirin 970, Snapdragon 730, and Snapdragon 835.

Powered by the Kirin 810, Huawei’s Nova 5 can run deep learning algorithms to enable features such as a first-ever night mode for front camera selfies and high-quality backlit pictures. AI algorithms process raw image data to enhance human subjects while deposing dark backgrounds.

The US recently placed Huawei on its Bureau of Industry and Security “Entity List.” To comply with sanctions, SoftBank-owned chip IP company Arm suspended its business with Huawei, blocking access to micro-processors such as the Cortex-A77 CPU, Mali-G77 GPU and the Arm ML (machine learning) processor announced at Computex this year. The Kirin 810 was however already in production when the sanctions were announced in May. Huawei can still develop chipsets based on previously licensed Arm IP.

The US sanctions have also left Huawei at odds with Google regarding the Android ecosystem and services. In a statement issued yesterday Huawei said that its current smartphone models like the P30 and Mate 20 will update to the new Android Q operating system when it is released later this year. The company is also preparing its own operating system, Hongmeng OS, which it plans to release early next year.

The Nova 5 and Nova 5 Pro smartphones are now on presale in China.

Journalist: Tony Peng & Fangyu Cai | Editor: Michael Sarazen

2018 Fortune Global 500 Public Company AI Adaptivity Report is out!
Purchase a Kindle-formatted report on Amazon.
Apply for Insight Partner Program to get a complimentary full PDF report.

Follow us on Twitter @Synced_Global for daily AI news!

We know you don’t want to miss any stories. Subscribe to our popular Synced Global AI Weekly to get weekly AI updates.




We produce professional, authoritative, and thought-provoking content relating to artificial intelligence, machine intelligence, emerging technologies and industrial insights.

Recommended from Medium

WAIC 2021 Ushers in a New Era for Chinese Tech

OpenAI Puts CV Models Under Their Microscope

Eddie: A Knowledge Backed Question Answering Agent — Part 1

[Dream Report] The Final Form of Humanity

New DeepMind Approach ‘Bootstraps’ Self-Supervised Learning of Image Representations

Artificial Intelligence

Artificial Intelligence simulation image

Automation is Not Intelligence

Caedmon: Creating Creativity

Get the Medium app

A button that says 'Download on the App Store', and if clicked it will lead you to the iOS App store
A button that says 'Get it on, Google Play', and if clicked it will lead you to the Google Play store


AI Technology & Industry Review — | Newsletter: | Share My Research | Twitter: @Synced_Global

More from Medium

Cash App Labs Modifies the Very Deep VAE to Achieve a 2.6x Speedup and 20x Memory Reduction

How Microsoft & OpenAI Are Squeezing the Best Out of GPT-3

Four Architectures that Showcase Meta AI’s Progress in Multimodal Deep Learning

Yoshua Bengio Team Challenges the Task-Diversity Paradigm in Meta-Learning