ASRock Industrial Integrates Axelera AI Metis Accelerators into Edge AI Platforms
ASRock Industrial has partnered with Axelera AI to integrate Metis AIPU accelerators into its edge AI platforms. These accelerators, available in PCIe AIC or M.2 form factors, deliver up to 214 TOPS of INT8 inference performance to edge devices. This integration brings high-performance, low-power digital in-memory computing to edge devices, making it easier for enterprises to deploy computer vision and AI workloads at scale. It offers a cost-effective and energy-efficient alternative to traditional GPU-based edge architectures. The Metis AIPU is manufactured using Samsung's 5nm process technology and operates within a low power envelope of 8 to 15 watts. It utilizes Digital In-Memory Computing (D-IMC) to minimize data transfer overhead and maximize computational efficiency.
## BACKGROUND
Edge AI refers to running artificial intelligence algorithms locally on hardware devices rather than in centralized cloud data centers. Traditional architectures often suffer from the 'memory wall,' where moving data between the processor and memory consumes significant energy and time. In-memory computing (IMC) addresses this bottleneck by performing calculations directly within the memory array, with digital IMC (D-IMC) offering high computational density for AI workloads.