Paradigm + Hanbo | HAMi supports the scheduling of Hanbo accelerators, promoting the implementation of domestic AI computing power

Recently, the heterogeneous computing power management project HAMi, led and contributed to as open-source by $PHANCY (06682.HK)$ , has reached a new milestone -officially achieving full support for accelerator cards under Hantbo SemiconductorThis means that when users deploy the Hanvo AI accelerator card in a Kubernetes cluster,they can uniformly schedule and manage computing power resources through HAMi,significantly improving the efficiency and flexibility of domestic heterogeneous chips.

This time, HAMi's support for the Hanvo accelerator card has achieved deep integration between domestic computing power chips and mainstream cloud-native open-source tools. By deploying the HAMi Vastai Device Plugin in a Kubernetes environment,users can easily report, allocate, and manage Hanvo accelerator card resources just like managing GPUs.。

About HAMi vGPU

Previously, if companies wanted to use domestic AI accelerator cards, they needed specialized personnel to configure and manage them. Now with HAMi vGPU, it’s like having an additional 'smart butler' that can:

- Automatically manage: how many cards are being used and by whom is clear at a glance

- Use on demand: Just like renting a shared bike, declare it, and it’s ready to use

- Flexible allocation: One machine can run multiple tasks simultaneously without interference

Schedule Vastai accelerators like GPUs: After completing the deployment of HAMi and related plugins, users only need to declare resources simply in a Pod to utilize Vastai accelerators. Below is a simple example of usage:

This update not only enriches HAMi’s hardware ecosystem but also provides developers and enterprise users with more options for domestically produced computing power. In the future,HAMi will continue optimizing support for different vendors’ accelerator cards, driving heterogeneous computing in cloud-native environments toward standardization and generalization.。

详见：https://github.com/Project-HAMi/HAMi/blob/master/docs/vastai-support_cn.md

About Phancy

Phancy Group is a global leader in general artificial intelligence technology, with the mission of 'AI for everyone.' As an open-source project for heterogeneous computing power management led and contributed to by Phancy, HAMi has become a CNCF sandbox project, committed to virtualizing, partitioning, and sharing heterogeneous computing power such as GPU/NPU.

About Vastai

Vastai Technologies, founded in 2018 and headquartered in Pudong, Shanghai, is a pioneer in the field of core AI computing power. Vastai focuses on the independent R&D of high-end GPU chips, with its business covering cutting-edge areas like AI inference acceleration, graphics rendering, and AIGC content generation. Its core products, including the Zaitian VA1 and Zaitian VA10 AI inference accelerators and smart all-in-one machines, have achieved scaled deployments across multiple industries.

36K Views