The new memory improves the efficiency of the data center

     Advantech announced the release of the SQRAM CXL 2.0 Type 3 memory module. Compute Express Link (CXL) 2.0 is the next evolution in memory technology, providing memory expansion through high-speed, low-latency interconnections designed to meet the needs of large AI training and HPC clusters. CXL 2.0 builds on the original CXL specification and introduces advanced features such as memory sharing and scaling, allowing for more efficient use of resources across heterogeneous computing environments.

     Fixed allocations often limit traditional memory architectures, leading to bottlenecks of underutilized resources and data-intensive workloads. E3. The CXL 2.0 memory module based on the EDSFF standard overcomes these limitations and supports dynamic resource management. This not only improves performance, but also reduces costs by maximizing available resources.

     The CXL memory module works through PCIe 5.0 ports. This high-speed connection ensures that data transfer remains fast and efficient, even as memory expands across different systems. The PCIe 5.0 interface provides up to 32GT/s bandwidth per channel, allowing the module to provide the bandwidth required for data-intensive applications. Adding more capacity will lead to better performance and memory bandwidth without requiring more servers and capital expenditures.

     CXL 2.0 allows multiple hosts to access the shared memory pool, optimizing resource allocation and improving overall system efficiency. The CXL memory pool technology enables computing components, such as cpus and accelerators, of multiple servers in a chassis to share memory resources, reducing resource redundancy and resolving low memory utilization.

     CXL memory modules can be added or removed from the system without shutting down the server, allowing for dynamic memory expansion. For data centers, this means being able to scale memory resources as needed, ensuring optimal performance without disruption.