High Performance GPU/CPU Server Systems and Control Workstation

Specifications High-Performance GPU/CPU Server Systems and Control Workstation The Laboratory for Autonomous Systems Research (LASR), Code 1700, located at the U.S. Naval Research Laboratory (NRL) in Washington, D.C., requires the procurement, installation, configuration, and support of high-performance computing platforms and an integrated control workstation. The purpose of this procurement is to provide: 1. A 4-server GPU/CPU rack capable of multi-threaded processing, GPU-accelerated workloads, AI and simulation applications, and integration with a system equivalent to Omniverse-based digital twin and simulation environments 2. A control workstation for remote monitoring, asset management, and system integration with the server rack. All systems and services must be fully compatible, operational, and support integrated management workflows. This document specifies the minimum technical and performance requirements for the hardware, software, and services to be provided by the vendor. Scope The vendor shall provide all hardware, software, and services necessary to meet the technical specifications outlined in this document. All deliverables, including server systems, GPUs and CPUs, storage, memory, networking, and supporting infrastructure, shall be fully assembled, configured, tested, and operational. All systems and services must be compatible with one another and suitable for high-performance computing, AI and simulation workloads, rendering, and digital twin applications. Installation Site -- 1 of 9 -- All installation, configuration, and integration work will at the Naval Research Laboratory, Washington, DC 20375. Technical Specifications: The vendor must provide / system must include: 1. High-performance GPU/CPU server system including 1.1. Complete 5U GPU direct-attached servers, fully configured to be installed together in a single rack including all of the following components and capabilities: 1.1.1. CSE-528G-R000NDP 5U rackmount chassis. 1.1.2. H14DSG-O-CPU motherboard. 1.1.3. Server processor supporting high-performance multi-threaded workloads. 1.1.4. Direct-attached GPU architecture supporting PCIe Gen 4 or higher. 1.1.5. Maximum supported DDR4 or DDR5 system memory. 1.1.6. NVMe and SATA storage compatibility. 1.1.7. PCIe expansion slots for GPUs and additional I/O cards. 1.1.8. Integrated high-speed networking ports (10GbE/25GbE or higher). 1.1.9. Redundant hot-swappable power supplies with optimized airflow. 1.1.10. Integrated remote management capability (IPMI or equivalent). 1.1.11. Total quantity: 4 1.2. Complete dual CPU processors, each including all of the following components and capabilities: 1.2.1. Turin 9335 DP/UP processor with 32 cores, 64 threads, 3.0 GHz base frequency, and ~128 MB cache. 1.2.2. Maximum supported DDR5 memory. 1.2.3. NVMe and SATA storage compatibility. 1.2.4. PCIe expansion capability for high-speed I/O cards. 1.2.5. Integrated networking ports supporting 10GbE, 25GbE, or higher. 1.2.6. Redundant hot-swappable power supplies with optimized cooling. 1.2.7. Integrated remote management capability. 1.2.8. Total quantity: 8 (2 per server) 1.3. Complete DDR5 server memory modules, including all the following: 1.3.1. 96 GB DDR5-6400 memory modules. 1.3.2. 2Rx4 architecture with low-profile form factor. 1.3.3. ECC Registered DIMM (RDIMM) type. 1.3.4. Full support for DDR5-6400 data rates. 1.3.5. Compatibility with supported server motherboards. 1.3.6. Total quantity: 96 (24 per server) -- 2 of 9 -- 1.4. Boot / System Complete NVMe PCIe SSD modules including all the following: 1.4.1. 7450 PRO series. 1.4.2. Capacity of 960 GB. 1.4.3. M.2 2280 form factor. 1.4.4. PCIe 4.0 NVMe interface. 1.4.5. Enterprise-grade 3D TLC NAND technology. 1.4.6. Full PCIe 4.0 NVMe performance. 1.4.7. Compatibility with server platform storage interfaces. 1.4.8. Total Quantity: 4 (1 per server) 1.5. High-Capacity / High-Performance Complete NVMe PCIe SSD modules, including all of the following: 1.5.1. 9550 PRO series. 1.5.2. Capacity of 7.68 TB. 1.5.3. U.2 15mm form factor. 1.5.4. PCIe 5.0 NVMe interface. 1.5.5. Enterprise TLC NAND rated for ~1 DWPD. 1.5.6. Full PCIe 5.0 NVMe performance. 1.5.7. Compatibility with enterprise server platforms. 1.5.8. Total quantity: 8 (2 per server) 1.6. Complete GPU accelerator modules, including all of the following: 1.6.1. RTX PRO 6000 Server Edition. 1.6.2. Approximately 96 GB GDDR7 GPU memory. 1.6.3. PCIe 5.0 GPU interface. 1.6.4. Support for GPU compute, AI, rendering, and virtualization workloads. 1.6.5. Compatibility with supported GPU server platforms. 1.6.6. Total quantity: 8 (2 per server) 1.7. Complete system management software licenses, including all of the following: 1.7.1. Single node license for server management software suite. 1.7.2. Hardware monitoring, alerting, reporting, and remote system management functionality. 1.7.3. Compatibility with all server platforms included in the procurement. 1.7.4. Total quantity: 4 (1 per server). 1.8. Complete SMC workstation integration including all the following: 1.8.1. Establish secure network connectivity between the SMC workstation and all servers. 1.8.2. Configure the workstation to remotely access, monitor, and manage all servers. 1.8.3. Ensure proper firewall, VLAN, or network segmentation configurations to allow uninterrupted remote management while maintaining network security. 1.8.4. Verify 4 servers and hardware are visible, manageable, to the SMC workstation. 1.8.5. Provide documentation and training for proper use of the SMC workstation for system monitoring, alerting, and remote administration. -- 3 of 9 -- 1.8.6. Quantity: 1 1.9. Complete Omniverse Enterprise subscriptions, including all of the following: 1.9.1. Per GPU subscription. 1.9.2. One- to three-year subscription term. 1.9.3. Support for collaborative 3D design, simulation, rendering, and AI workflows. 1.9.4. Compatibility with GPU server p…