Neurxcore, a French AI processor startup, has introduced a new line of Neural Processor Units (NPUs) for AI inference applications. The company claims that their product line, which is based on an enhanced and extended version of NVIDIA's Deep Learning Accelerator (Open NVDLA) technology, offers significant improvements in energy efficiency, performance, and feature set compared to the original NVIDIA version.
Additionally, Neurxcore's NPUs benefit from NVIDIA's development and incorporate patented in-house architectures. The SNVDLA IP from Neurxcore is primarily focused on image processing, including classification and object detection, but it also offers versatility for generative AI applications. The SNVDLA has already been silicon-proven and operates on a 22nm TSMC platform. It has been showcased on a demonstration board running a variety of applications.
The IP package contains the Heracium SDK, developed by Neurxcore using the open-source Apache TVM (Tensor-Virtual Machine) framework. This SDK is used to configure, optimize, and compile neural network applications on SNVDLA products. Neurxcore offers a range of products for various applications, including sensors and IoT, wearables, smartphones, smart homes, surveillance, Set-Top Box and Digital TV (STB/DTV), smart TV, robotics, edge computing, AR/VR, ADAS, servers, and more.
Português
USD $
