MSI’s display of the CD182-S6091-X2 (DLC) server node at Computex signals a shift toward high-density, liquid-cooled infrastructure designed for AMD’s upcoming 6th generation “Venice” EPYC platform. The system utilizes direct liquid cooling to support four high-performance processors and 16-channel DDR5 memory configurations within a single Open Compute Project (OCP) v3 rack unit, according to MSI’s product documentation.
Engineering High-Density Server Nodes
The CD182-S6091-X2 (DLC) achieves its density by packing two dual-socket server nodes into a single 1OU chassis. According to MSI, this configuration allows for four AMD processors to operate within the footprint of one OCP v3 rack unit. The move to direct liquid cooling (DLC) addresses the thermal limitations of air cooling as processor thermal design power (TDP) continues to rise. With 5th generation EPYC chips reaching 500 Watts, liquid cooling is becoming a standard requirement for maintaining performance in high-density environments.
Did you know?
The CD182-S6091-X2 (DLC) uses a serial liquid-cooling setup where CPUs are on one loop and the 32 RDIMMs are on a separate, dedicated loop to manage the heat generated by high-capacity memory.
Infrastructure Requirements for PCIe Gen6
The transition to AMD’s Venice platform introduces PCIe Gen6 connectivity, which MSI has integrated into the node’s expansion architecture. Each node features an OCP 3.0 slot for network interface cards, alongside both Full-Height Half-Length (FHHL) and Half-Height Half-Length (HHHL) PCIe Gen6 x16 slots. Local storage is handled via four front-facing E1.S drive bays, also utilizing the Gen6 interface, ensuring compatibility with next-generation storage hardware like Micron’s 9650 SSDs.
Scalability with ORv3 Liquid-Cooled Racks
MSI demonstrated the scalability of these nodes using an ORv3 liquid-cooled rack capable of housing 28 servers. This 44OU rack supports a total power load of 100kW, accommodating 112 Venice processors and 1,792 RDIMMs. The rack design removes traditional power supplies from the nodes, opting instead for a 48VDC busbar system. Power is provided by two 55kW Chicony power shelves, while a 100kW Auras Coolant Distribution Unit (CDU) manages the liquid-to-liquid heat exchange for the entire rack.
Technical Specifications Comparison
| Feature | Specification |
|---|---|
| Form Factor | 1OU2N (2 nodes per 1OU) |
| Memory | 16-channel DDR5 per server |
| Storage | 4x E1.S PCIe Gen6 slots |
| Management | ASPEED AST2700 BMC |
Frequently Asked Questions
Why is liquid cooling necessary for the Venice platform?
As processor TDPs increase—with predecessors already hitting 500 Watts—air cooling struggles to maintain the necessary thermal headroom. Liquid cooling allows for higher compute density and more efficient heat dissipation for both CPUs and high-capacity memory.
What is the benefit of the 48VDC busbar system?
By using a busbar, the system eliminates the need for individual power supplies in each node. This saves physical space, simplifies node installation and removal, and streamlines the power delivery architecture for high-density racks.
Does the CD182-S6091-X2 (DLC) support current generation drives?
The system uses PCIe Gen6 interfaces for its E1.S drive bays. While it is designed for next-generation hardware, the sources do not state if these slots are backward compatible with existing PCIe-based NVMe storage solutions.
Are you preparing your data center for the next generation of server hardware? Subscribe to our newsletter for the latest updates on high-density computing trends and hardware releases.
Keep reading