Huawei open sources MindSpore: claims to provide 'all-scenario AI computing framework'

Mar 30, 2020

[Shenzhen, China, March 30, 2020] Huawei made a series of important announcements at the Huawei Developer Conference 2020 (Cloud) – HDC.Cloud, on March 28, notably that MindSpore, the all-scenario AI computing framework, goes open source on Gitee, and that ModelArts Pro, the first-ever AI app development suite for enterprises, goes live on HUAWEI CLOUD. Huawei also showcased some of the significant applications for Huawei's Atlas AI computing platform, on the cloud, edge, and devices. In doing so, Huawei has delivered the full-stack, all-scenario AI solutions for developers that it had first unveiled at HUAWEI CONNECT 2018.

MindSpore to Officially Go Open Source

Professor Chen Lei, chief scientist at Huawei MindSpore and an IEEE Fellow, announced that Huawei's all-scenario AI computing framework MindSpore would go open source on Gitee. Huawei remains committed to building a global open source community, to serve as the backbone for a vibrant AI software and hardware ecosystem.

MindSpore now open source

MindSpore has focused on lowering entry barriers for AI developers, with Professor Chen noting: "MindSpore natively adapts to all scenarios across the device, edge, and cloud. We implement AI Algorithms As Code through on-demand collaboration for easier model development, and provide cutting-edge technologies, and co-optimization with Huawei Ascend AI processors to improve runtime efficiency and computing performance. We also support other processors such as GPU and CPU."

ModelArts Pro, First-ever Enterprise-Grade AI App Development Suite

Yuan Jing, General Manager of HUAWEI CLOUD General AI Services, unveiled ModelArts Pro, a first-of-its-kind AI app development suite for enterprises. ModelArts Pro is positioned as an enterprise AI productivity tool on the cloud, and provides an innovative model for putting industrial AI into practice, by integrating the know-how of experts in the field into corresponding suites and workflows that empower AI app developers. This novel approach boosts overall AI development efficiency and implementation results. ModelArts Pro provides enterprises with the skills to build AI apps, and makes the boundless potential of AI broadly accessible.

Best Partner Practices for the Huawei Atlas AI Computing Platform

Zhang Dixuan, General Manager of Huawei Atlas Data Center Business, shared some of the best device-edge-cloud development practices by partners who had made use of the Huawei Atlas AI computing platform. The presentation included a detailed demonstration of the performance optimization process behind the high-density cloud AI inference solution, as well as the benefits of efficient edge inference and ultra-low-consumption device-side inference, which are based on cloud-edge collaboration.

Huawei has worked with dozens of partners to promote the use of the Atlas series accelerator module, card, edge station, and server powered by Huawei Ascend AI processors, across numerous fields, including smart transportation, smart power, smart finance, smart city, and smart manufacturing.

Mr. Zhang also unveiled the Tensor Boost Engine (TBE) for open beta, which aims to support developers in operator customization, and enrich the broader AI ecosystem. Huawei plans to award more than 100 universities and partners as excellent operator contributors. This initiative will further bolster the computing power on Ascend AI processors.

Professor Li Tao from Nankai University showcased fundus screening, an AI-enabled application based on the Huawei Atlas 200 DK Developer Kit.

Dr. Li Jin, co-founder and CTO of Apulis Technology, shared development practices related to Huawei Atlas, and revealed the industry's first open source heterogeneous AI platform powered by the Huawei Ascend 910 processor.

At HUAWEI CONNECT 2018, Huawei released the first-ever full-stack, all-scenario AI solution, which encompasses chips, chip enablement, training and inference framework, and application enablement, and can be deployed in diverse scenarios, including public cloud, private cloud, edge computing, IoT industry terminals, and consumer terminals.

Two years later at HDC.Cloud, Huawei has shown how much progress it has made in its ambitious goal, especially on the developer side. The full-stack, all-scenario solutions that have been introduced, allow for global developers to develop AI applications with greater ease, incorporate the benefits of AI across numerous industries, and add value for enterprises and society at large, by contributing to a robust global AI ecosystem.