Microchip Releases Free VectorBlox 3.0 SDK
Article excerpt
Highlighted: the sentence this signal was extracted from
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...
Keep reading with a free account
The rest of this article, and every signal for Microchip Technology, is in your free account.
