Huawei has unveiled its Kunpeng Developer Kit at Huawei Connect in Shanghai, China.
Kunpeng is a server chip designed using the ARM architecture and is utilised in Huawei’s servers.
The developer kit is fully deployed on the Huawei DevCloud and offers a variety of tools, including:
- Dependency Advisor
- Porting Advisor
- Compiler Collection
- Tuning Kit
The improved code porting of the Developer Kit’s Dependency and Porting Advisors will reduce partners’ labour costs by 50%, Huawei said.
The company added that iterative compilation can improve performance by 15%, while performance fine-tuning can improve efficiency by up to five times.
Huawei’s Developer Programme 2.0 has also been launched, and is now available in four different versions for universities, individuals, startups, and partners.
Huawei will invest $1.5 billion over the next 5 years into the programme.
Huawei also launched ModelArts v2.0, which includes a variety of impressive new AI features.
Notably, the platform allows those with minimal coding and AI experience to develop accurate AI models by simply following these three steps:
- Upload and label data
- Click to start AI training
- Verify the model and roll out your services
Huawei has also included a new feature that allows customers to use their existing labeled data to automate the labeling of other data.
“We want to significantly lower the threshold for developers,” said Jia Yongli, GM of the EI Product Department of Huawei Cloud.
ModelArts 2.0 also allows customers to develop new operators, create domain-specific AI frameworks, and create activation functions and neural networks.
Bradley Prior is a guest of Huawei at Huawei Connect in China.