SK Hynix 96GB DDR5 RDIMM 6400MT/s (1cnamo, 24Gb)

The SK Hynix 96GB DDR5 RDIMM (1cnamo) leverages next-generation 24Gb monolithic die architecture to deliver unprecedented capacity per DIMM slot, enabling servers to achieve terabyte-scale memory configurations without increasing slot count. Operating at 6400 MT/s with DDR5 on-die ECC and registered buffering, this module provides the bandwidth and reliability demanded by latency-sensitive enterprise workloads. The 1cnamo designation reflects SK Hynix's advanced 1c DRAM process node, the company's most refined fabrication geometry as of 2025, yielding improved power efficiency and signal integrity compared to prior 1a and 1b nodes.

As a Registered DIMM, the module incorporates a Register Clock Driver (RCD) that re-drives command and address signals, enabling high-rank counts and large memory subsystems to operate at full speed without electrical loading penalties. The DDR5 architecture natively includes a power management integrated circuit (PMIC) per DIMM, moving voltage regulation on-die and improving per-slot power delivery accuracy. On-die ECC provides a first layer of error scrubbing before errors can escalate, complementing the server platform's existing SDDC or ADDDC RAS features.

Launched in Q2 2025, this RDIMM targets memory-dense AI inference servers, large-scale in-memory database deployments, and high-performance computing environments where per-socket memory capacity is the primary constraint. Platforms based on AMD EPYC Genoa/Turin and Intel Xeon Scalable (Sapphire Rapids/Granite Rapids) with DDR5 memory channels benefit directly from the 96GB capacity point, allowing a 12-slot dual-socket server to reach 2.304TB of total system memory — a configuration previously requiring 128Gb or multi-rank stacking solutions.

Ideal for

• Large language model (LLM) inference servers requiring multi-terabyte memory pools to hold full model weights in-DRAM without NVMe offload
• SAP HANA and Oracle In-Memory Database deployments where entire datasets must reside in DRAM for sub-millisecond query response
• High-frequency trading and financial risk analytics platforms demanding both high capacity and deterministic low-latency memory access
• Virtualization hosts running high VM densities where each virtual machine is allocated multi-gigabyte working sets and memory overcommit must be minimized
• Genomics and bioinformatics workloads that assemble or analyze terabyte-scale reference genomes entirely in-memory to eliminate I/O bottlenecks
• In-memory caching tiers (Redis, Memcached, Apache Ignite) serving distributed microservices architectures at scale with maximum node memory footprint

Technical specifications

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