Alibaba's T-Head Shows Off an Android 12 RISC-V Port with TensorFlow Lite Capabilities

New roadmap shows a route to upstreaming and future support for Android 13 and Android 14, along with third-party driver enhancements.

Staff at Alibaba's semiconductor division, T-Head, have shown off a functional port of Android 12 to free and open source RISC-V cores — and have even showcased it running TensorFlow Lite for edge-AI work.

"Last year Alibaba T-Head reported that the basic features of Android had been ported onto RISC-V-based XuanTie cores," says Alibaba Cloud's Mao Han of the division's work. "Since then more effort has been spent on Android 12 and enabling third-party modules to support video, camera, and Wi-Fi/Bluetooth features based on RISC-V. Specifically, we have focused on incorporating high-performance RISC-V processors onto Android 12."

Android 12 is now running on RISC-V cores, and has brought TensorFlow Lite along for the ride. (📷: T-Head)

Work to get Android up and running on RISC-V devices reached its first major milestone two years ago when the PLCT Lab announced its first boot of a minimal Android Open Source Project (AOSP) environment. Shortly after, T-Head released a more functional Android 10 port — while having apparently skipped Android 11 to concentrate on Android 12 for its latest release.

Building on that Android 10 base, T-Head's Android 12 port includes a selection of new features including support for the Bazel build system, the new Android Runtime (ART) and Bionic modules, and a range of third-party drivers for everything from HDMI displays to Wi-Fi, USB, and camera devices.

Alibaba's T-Head has outlined a roadmap to upstreaming and Android 14 support. (📷: T-Head)

"In addition to the changes in the Android Open Source Project, we also improved and configured the demo app of TensorFlow Lite to execute on RISC-V-based cores," Han explains. "First, we add RISC-V build support to the TensorFlow Lite project. We then compile the installation package with Android Studio. Then, we integrate the hardware accelerator and Hardware Abstraction Layer (HAL) of the neural network so that user application calls of Neural Networks API (NNAPI) can be supported."

At present, however, the project remains largely in-house. T-Head's next milestone, Han says, will be to merge the resulting code upstream, upgrade to Android 13, and then Android 14. "By the release of Android 13, we plan to rebase the support for RISC-V with minimum changes," Han writes. "The refined patch set of the core components for RISC-V would be ready for submission by the development window of Android 14."

More details are available on the RISC-V International blog.

ghalfacree

Freelance journalist, technical author, hacker, tinkerer, erstwhile sysadmin. For hire: freelance@halfacree.co.uk.

Latest Articles