Atlas 1.8SP5-D
21" and 3 OpenU platform built to support up to 8 GPUs and leveraging single-phase immersion cooling. This system is engineered for optimal energy efficiency, delivering sustainable performance while reducing data center operational costs.
| Cooling system: Single-phase immersion cooling |
| CPU brand: AMD |
| Number of sockets: 2 |
| CPU socket: SP5 (LGA 6096) |
| Memory slots: 24 slots |
| GPU/FPGA: 8 slots |
| Form factor: 21" 3OU |
Atlas Immersion Platform
High-Density AI & HPC Performance
Up to 8 GPUs in a compact 3 OpenU platform, optimized for 21" high-density infrastructures and immersion-cooled data centers.
Powered by AMD EPYC™ 9005 series processors, the platform delivers high-performance compute for modern AI and HPC workloads.
Combined with next-generation GPUs and high-bandwidth interconnect technologies, it ensures efficient acceleration for demanding applications.
The system supports up to 8 dual-slot GPUs and NVIDIA® NVLink® technology, enabling fast GPU-to-GPU communication and scalable performance for AI training, inference, and scientific computing.
Accelerated by the latest NVIDIA GPU architectures
The Atlas platform is designed to support the latest generation of high-performance GPU accelerators for AI, HPC, and advanced computing workloads.
The system is fully compatible with leading NVIDIA architectures including NVIDIA RTX 6000 Pro and NVIDIA H200 NVLink4, while also supporting a broader range of GPU configurations depending on workload and deployment requirements.
NVIDIA H200 NVL4
- LLM
- Generative AI
- HPC simulation large scale training
NVIDIA RTX 6000 PRO
- AI development
- Computer vision
- Digital twin
- Visualization
- Inference
AI Cluster Ready Infrastructure
The Atlas platform is designed to deliver maximum compute density for modern AI clusters. By integrating up to 8 GPUs within a compact 3 OpenU chassis and a 21" wide architecture, the system enables higher rack utilization, improved space efficiency, and optimized energy performance.
This high-density design makes Atlas an ideal building block for large-scale AI training clusters, HPC environments, and enterprise accelerated computing platforms, allowing more compute power to be deployed per rack.
Combined with immersion cooling, the platform supports sustained high GPU power levels while maintaining reliable and efficient thermal performance.
Higher GPU density per rack
Deploy more GPUs within a smaller footprint to accelerate AI training infrastructure.
Optimized for single-phase immersion cooling
Immersion cooling enables sustained high-power GPU operation while improving overall infrastructure efficiency.
Cluster-ready architecture
Combined with 6 PCIe NIC slots for high-speed networking, the Atlantis enables efficient interconnection of multiple GPU nodes.
Designed for Immersion Cooling
The Atlas platform is purpose-built for single-phase immersion cooling, enabling stable operation under sustained high GPU power while simplifying infrastructure management.
To provide complete confidence for data center operators, 2CRSi supports immersion deployments with a dedicated manufacturer warranty, ensuring that the system remains protected when operated in qualified immersion environments.
Front I/O accessibility
All networking and system connections are accessible from the top of the server, simplifying installation and maintenance while avoiding complex cabling at the bottom of the immersion system.
Simplified maintenance
The server layout allows operators to quickly access compute nodes and networking components, reducing intervention time and improving serviceability.
Optimized for high-density deployments
The immersion-ready architecture allows higher compute density per rack while maintaining reliable thermal performance for GPU-intensive workloads.