Microchip Releases Free VectorBlox 3.0 SDK

Microchip Technology Inc. has introduced version 3.0 of its VectorBlox Accelerator software development kit (SDK), adding new capabilities to improve the implementation of artificial intelligence inference on PolarFire FPGA and system-on-chip (SoC) devices. The updated development environment is designed to simplify the deployment of convolutional neural network (CNN) models while increasing performance and energy efficiency in edge AI applications operating under strict power and reliability constraints.

Available at no cost, the VectorBlox 3.0 SDK integrates with the CoreVectorBlox IP and provides a unified toolchain for optimizing, compiling, and deploying neural network models on PolarFire FPGA and SoC platforms. The software automates much of the development workflow, helping engineers accelerate FPGA-based AI implementation while reducing design complexity and shortening development time.

One of the main enhancements in the new release is support for sparse neural networks. By identifying and bypassing zero-value operations during inference, the accelerator reduces the computational and memory bandwidth required to execute convolutional neural networks. This sparsity-aware processing increases inference throughput while lowering power consumption, making it particularly suitable for always-on edge AI systems where energy efficiency is a primary design objective. Model compression techniques enabled by sparse neural networks also reduce hardware resource utilization without significantly affecting inference accuracy.

The VectorBlox architecture is designed to scale across different neural network sizes and to execute multiple AI workloads on a single FPGA. This allows developers to consolidate several vision- or sensor-based inference functions within a single low-power device, reducing overall system complexity and component count.

The software release also supports Spacecraft Pose Network v2 (SPNv2), a neural network designed to estimate the position and orientation of spacecraft from visual data. Running on PolarFire FPGA and SoC devices, the solution targets autonomous space operations including rendezvous and docking, satellite inspection, formation flying, and space debris removal. The underlying hardware platform incorporates features such as single-event upset (SEU) immunity, secure boot, and anti-tamper protection, making it suitable for aerospace, defense, and other mission-critical environments that require long-term operational reliability.

VectorBlox SDK 3.0 integrates with the Libero SoC Design Suite and is available free of charge together with the CoreVectorBlox IP, providing developers with a complete framework for implementing efficient AI inference on PolarFire-based embedded systems.

The post Microchip Releases Free VectorBlox 3.0 SDK appeared first on Embedded.