SK Hynix CXL 2.0 Type 3 256GB Memory Expander Module

The SK Hynix CXL 2.0 Type 3 256GB Memory Expander Module (HMCG96AGBRA115N) is a Compute Express Link 2.0-compliant memory expansion device built on SK Hynix's DDR5 DRAM technology. Designed as a Type 3 device, it operates as a memory expander without integrated compute, attaching to a host processor via the CXL.mem protocol over a PCIe Gen 5 x8 physical interface. This architecture allows servers to break through traditional DIMM slot constraints, dramatically expanding the total addressable memory pool available to CPU-centric and memory-intensive workloads.

The module delivers 256GB of usable capacity in a standard E3.S or compatible form factor, enabling rack-scale memory disaggregation in modern datacenter infrastructure. CXL 2.0 introduces persistent memory pooling capabilities and improved latency characteristics over the first generation, making this module a strong fit for platforms where main memory bandwidth and capacity are both critical. The interface is fully compliant with the CXL Consortium's 2.0 specification, ensuring interoperability with CXL 2.0-enabled processors and host platforms from major silicon vendors.

Targeted at enterprise datacenters, hyperscale cloud operators, and AI/ML infrastructure teams, this memory expander is particularly well suited to workloads that require far more working memory than conventional DIMM configurations can provide — including large language model inference, in-memory analytics, real-time databases, and high-performance virtualization environments. SK Hynix's manufacturing process for this module leverages advanced DDR5 DRAM dies, ensuring the reliability, endurance, and RAS (reliability, availability, serviceability) features expected in mission-critical deployments.

Ideal for

• Expanding host memory capacity in AI inference servers running large language models that exceed conventional DIMM slot limits
• In-memory database acceleration for real-time analytics platforms requiring hundreds of gigabytes of low-latency working memory per node
• Memory disaggregation in hyperscale and cloud datacenter environments to enable flexible, rack-scale memory pooling across multiple compute nodes
• High-density virtual machine consolidation on CXL 2.0-capable servers where per-VM memory allocation demands exceed onboard DIMM capacity
• HPC and scientific computing workloads with large working datasets that benefit from expanded memory address space beyond standard DRAM configurations
• Enterprise data warehouse and OLAP query acceleration where large result sets and intermediate aggregations require persistent in-memory staging

Technical specifications

Available from Omnixon Global. Submit an RFQ and our team will confirm configuration and availability for your order.