Kinara, Inc. announced the Kinara Ara-2 Edge AI processor, a new product designed to enhance Edge servers and notebooks.
The Ara-2 processor delivers high performance, cost and power efficiency, making it ideal for running applications such as video analytics, large language models (LLMs) and other generative AI models. It is also well suited for Edge applications utilizing both traditional AI models and modern ones with transformer-based architectures.
Ara-2 has an advanced feature set and delivers 5-8 times the performance of Ara-1. It combines real-time responsiveness with high throughput. Demand for LLM and generative AI is growing, but most applications currently rely on GPUs in data centers, resulting in high latency, cost and privacy concerns. Kinara's Ara-2 aims to solve these problems by moving computing power to the user Edge. It supports billions of parameters used in generative AI models and eases the transition from GPUs with computational engines and a software development kit (SDK) that provides highly accurate quantization and direct FP32 support.
Ara-2 also emphasizes security with features such as secure boot, encrypted memory access, and a secure host interface. Ara-2 is available in a variety of forms, including a standalone device, USB module, M.2 module, and PCIe card with multiple Ara-2s.
Kinara will demonstrate the Ara-2 at CES and invites interested parties to schedule a meeting in their suite at the Venetian Hotel on January 9, 10 or 11, 2024.