In cloud service environments, memory and storage performance requirements have been driven by growth in high-speed computing and data transmission needs, especially for the computing-intensive applications in terms of Artificial Intelligence or Machine Learning work load. The increasing system DRAM capacity demand induced the higher TCO of IT equipment became pain point of many companies. Moreover, in the traditional structure, system memory is directly linked to a CPU or GPU and can’t be shared by others, which causes inefficient DRAM usage in most of the time.
Inventec X680m CXL Memory Accelerator offers new breakthrough high-speed CPU to device interconnection, memory scalability and performance optimization within single width PCIe-Gen5-x16 CEM form factor. CXL protocol enables memory expansion, pooling and sharing solutions to enhance system memory utilization. X680m uses CXL-compliant Altera Agilex™ 7 I-series FPGA controller which provides lower latency and high-performance CXL interface. Combining memory tiering technology with on-board storage and Networking interfaces, X680m facilitates lower cost NVME SSD operating as virtualized system DRAM to replace up to 50% system memory capacity and bring significant TCO reduction and best cost performance application to enterprises and Cloud Service Providers.
