AmpereOne Family AmpereOneM Processor

Ampere’s AmpereOne® M processor is part of the AmpereOne® family of high-performance server-class processors. It is engineered to deliver exceptional performance for AI Compute: from inference to the broad spectrum of mainstream data center workloads. With 12 DDR5 memory channels, AmpereOne® M is ideal for computationally intensive generative AI (GenAI) applications such as large language model (LLM) AI inference and memory-intensive object caching. Its high memory bandwidth drives superior performance and maximum model density, empowering clouds and enterprises to handle more AI inference tasks per processor in secure virtual machines or containers with ease.

AmpereOne® M processors feature a new platform architecture with a higher pin count and additional memory channels, distinguishing it from other AmpereOne platforms while continuing to maintain Ampere’s Cloud Native processing principles including:

• Performance: Delivers optimal tokens-per-second performance and maximum instance, session or agent density for advanced AI-enabled digital services
• Efficiency: Minimizes energy consumption for AI Compute workloads to enable sustainable and high capacity data center operations
• Scalability: Seamlessly adapts to evolving AI Compute requirements by supporting the full range of compute workloads within virtualized or containerized environments on a single easy-to-deploy platform

AmpereOne® M implements an advanced and innovative architecture that delivers power-efficient AI Compute at lower cost.