Optimizing NN inference performance on Arm NEON and Vulkan
This article is an overview of a presentation given by ax Inc. at arm developer AI virtual tech talks on September 21 2021.
Overview
This talk starts with an overview of ailia SDK, then introduces optimization techniques for inferring neural networks at high speed in Arm environments. Based on our research for developing ailia SDK, we introduce the optimization for Arm CPU using NEON SIMD instructions and various optimal compute shader implementations for Arm Mali using Vulkan. In addition, we demonstrate how various machine learning models actually operate at high speed in Arm environments.
Links to the video and slides
You can also download slides here.
Video timestamps
ax Inc. has developed ailia SDK, which enables cross-platform, GPU-based rapid inference.
ax Inc. provides a wide range of services from consulting and model creation, to the development of AI-based applications and SDKs. Feel free to contact us for any inquiry.