Graphics card NVIDIA H200 NVL 141GB HBM3e PCIe 5.0

AMD RDNA 4.0 Navi 48, Processor with 256-bit Memory Interface 4,096 Universal Shader Processors for Geometry and Pixel Rendering Graphics Memory Bandwidth: 645 GB/s (GDDR6 ECC RAM) Single Precision Floating Point Performance (Peak): 47.8 TeraFlops Half Precision Floating Point Performance: 95.7 TeraFlops Passmark: ? points High-Level Shading Language:OpenGL: 4.6 / OpenCL: 3.0 / DirectX: 12.2 / Vulkan: 1.3 / Shader Model: 6.8 • Memory: 32-GB • Shader Units / Compute-Einheiten / Ray / AI: 4.096 / 64 / 64 / 128 • Power Consumption: 300 watts (TDP) • External Power Supply: 1x PCIe CEM 5.0 16-Pin • Interface: PCIe 5.0 x16 • Form Factor: Full-Size ATX / Dual-Slot • Cooling: 1x Radial Fan (65mm), Reference Design • Connections: 4x DisplayPort 2.1a • Special Features: AMD Software Adrenalin Edition, AMD FidelityFX™ Super Resolution 4, AMD Fluid Motion Frames, AMD Radeon™ Super Resolution, AMD Smart Access Memory, AMD Smart Access Video, AMD Radeon™ Boost, AMD Radeon™ Anti-Lag, AMD Radeon™ Image Sharpening 2, AMD Enhanced Sync Technology, AMD FreeSync™ Technology, AMD Radeon™ Chill, AMD Virtual Super Resolution • Supported Video Formats: H264 Encode and Decode, H265/HEVC Encode and Decode, AV1 Encode & Decode 8, 10, 16-bit RGB High Dynamic Range Output (HDR) 3-Year Manufacturer Warranty

AMD RDNA 4.0 Navi 48, Processor with 256-bit Memory Interface 4,096 Universal Shader Processors for Geometry and Pixel Rendering Graphics Memory Bandwidth: 645 GB/s (GDDR6 ECC RAM) Single Precision Floating Point Performance (Peak): 47.8 TeraFlops Half Precision Floating Point Performance: 95.7 TeraFlops Passmark: ? points High-Level Shading Language:OpenGL: 4.6 / OpenCL: 3.0 / DirectX: 12.2 / Vulkan: 1.3 / Shader Model: 6.8 • Memory: 32-GB • Shader Units / Compute-Einheiten / Ray / AI: 4.096 / 64 / 64 / 128 • Power Consumption: 300 watts (TDP) • External Power Supply: 1x PCIe CEM 5.0 16-Pin • Interface: PCIe 5.0 x16 • Form Factor: Full-Size ATX / Dual-Slot • Cooling: 1x Radial Fan (65mm), Reference Design • Connections: 4x DisplayPort 2.1a • Special Features: AMD Software Adrenalin Edition, AMD FidelityFX™ Super Resolution 4, AMD Fluid Motion Frames, AMD Radeon™ Super Resolution, AMD Smart Access Memory, AMD Smart Access Video, AMD Radeon™ Boost, AMD Radeon™ Anti-Lag, AMD Radeon™ Image Sharpening 2, AMD Enhanced Sync Technology, AMD FreeSync™ Technology, AMD Radeon™ Chill, AMD Virtual Super Resolution • Supported Video Formats: H264 Encode and Decode, H265/HEVC Encode and Decode, AV1 Encode & Decode 8, 10, 16-bit RGB High Dynamic Range Output (HDR) 3-Year Manufacturer Warranty

NVIDIA Hopper GH100, processor with 5,120-bit memory connection 16,896 universal shader processors for geometry and pixel rendering Graphics memory bandwidth 3.36 TB/s (HBM3e ECC RAM) Single precision floating point performance (peak) = 248.3 teraflops Half precision floating point performance = 62.08 teraflops High-level shading language: OpenGL: 4.6 / OpenCL: 3.0 / DirectX: 12.2 / Vulkan: 1.3 / CUDA 12.6 Shader Model 6.7 • Memory: 96 GB • Shader units / TMUs / Tensor / ROPs: 16,896 / 528 / 528 / 24 • Power consumption: 700 watts (TDP) • External power supply: 1x PCIe CEM 5.0 16-pin • Interface: PCIe 5.0 x16 • Form factor: Full-size ATX / dual slot • Cooling: Passive • Connectors: None • Special features: 7 x NVENC | 7 x JPEG decoders NVLink, NVIDIA VR-Ready, NVIDIA VR SLI 10-bit color channel depth, HDCP 2.2 3-year manufacturer's warranty

GPU name: AD104 Architecture: Ada Lovelace Memory size: 24 GB Memory type: GDDR6 with ECC Memory bus: 192 bit Memory bandwidth: 300 GB/s FP32 (Float) Power: 31.3 TFLOPS Shader units: 7.680 Third generation Tensor cores: 40 Supported APIs: OpenCL / DirectCompute / OpenACC / CUDA Power consumption: 40-60 Watt Thermal solution: Passive

The NVIDIA A10 Tensor Core GPU delivers a versatile platform for mainstream enterprise workloads, like AI inference, training, and HPC. With TF32 and FP64 Tensor Core support, as well as an end-to-end software and hardware solution stack, A10 ensures that mainstream AI training and HPC applications can be rapidly addressed. Multi-instance GPU (MIG) ensures quality of service (QoS) with secure, hardware-partitioned, right-sized GPUs across all of these workloads for diverse users, optimally utilizing GPU compute resources. Scope of delivery: • NVIDIA A10

The NVIDIA A100 Tensor Core GPU delivers unprecedented acceleration at every scale for AI, data analytics, and HPC to tackle the world’s toughest computing challenges. As the engine of the NVIDIA data center platform, A100 can efficiently scale up to thousands of GPUs or, using new Multi-Instance GPU (MIG) technology, can be partitioned into seven isolated GPU instances to accelerate workloads of all sizes. A100’s third-generation Tensor Core technology now accelerates more levels of precision for diverse workloads, speeding time to insight as well as time to market. Scope of delivery: • NVIDIA A100 • 8-pin male (graphics card) to 2x 8-pin female (power supply)

The NVIDIA A2 Tensor Core GPU provides entry-level inference with low power, a small footprint, and high performance for NVIDIA AI at the edge. Featuring a low-profile PCIe Gen4 card and a low 40-60 watt (W) configurable thermal design power (TDP) capability, the A2 brings adaptable inference acceleration to any server. Scope of delivery: • NVIDIA A2 • Low-Profile (SFF) Bracket mounted • Additional Full-Height (ATX) Bracket

NVIDIA® A40 delivers the data center-based solution designers, engineers, artists, and scientists need to meet today’s challenges. Built on the NVIDIA Ampere architecture, the A40 combines the latest generation RT Cores, Tensor Cores, and CUDA® Cores with 48GB of graphics memory for unprecedented graphics, rendering, compute, and AI performance. From powerful virtual workstations accessible from anywhere, to dedicated render nodes, the A40 is built to tackle the most demanding visual computing workloads from the data center. Scope of delivery: • NVIDIA A40 • 8-pin male (graphics card) to 2x 8-pin female (power supply)

GPU name: GH100 Architecture: Hopper Memory size: 80 GB Memory type: HBM2e Memory bus: 5120 bit Memory bandwidth: 1280 GB/s Supported APIs: OpenCL / DirectCompute / OpenACC / CUDA Shader units: 7,296 Tensor cores: 456 FP32 (Float) Power: 24.08 TFLOPS FP64 (double) Power: 12.04 TFLOPS Power consumption: 350 Watt Thermal solution: Passive

Der NVIDIA L40 liefert eine noch nie dagewesene visuelle Rechenleistung für das Rechenzentrum und bietet Grafik-, Rechen- und KI-Funktionen der nächsten Generation für GPU-beschleunigte Anwendungen. Basierend auf der Ada Lovelace GPU-Architektur verfügt der L40 über RT Cores der dritten Generation, die die Echtzeit-Raytracing-Fähigkeiten verbessern, und Tensor Cores der vierten Generation mit Unterstützung für das FP8-Datenformat, die eine Inferenzleistung von über einem Petaflop liefern. Diese neuen Funktionen werden mit CUDA Cores der neuesten Generation und 48 GB Grafikspeicher kombiniert, um Visual Computing-Workloads von hochleistungsfähigen virtuellen Workstation-Instanzen bis hin zu großen digitalen Zwillingen in NVIDIA Omniverse zu beschleunigen. Scope of delivery: • NVIDIA L40

Der NVIDIA L40S liefert eine noch nie dagewesene visuelle Rechenleistung für das Rechenzentrum und bietet Grafik-, Rechen- und KI-Funktionen der nächsten Generation für GPU-beschleunigte Anwendungen. Basierend auf der Ada Lovelace GPU-Architektur verfügt der L40S über RT Cores der dritten Generation, die die Echtzeit-Raytracing-Fähigkeiten verbessern, und Tensor Cores der vierten Generation mit Unterstützung für das FP8-Datenformat, die eine Inferenzleistung von über einem Petaflop liefern. Diese neuen Funktionen werden mit CUDA Cores der neuesten Generation und 48 GB Grafikspeicher kombiniert, um Visual Computing-Workloads von hochleistungsfähigen virtuellen Workstation-Instanzen bis hin zu großen digitalen Zwillingen in NVIDIA Omniverse zu beschleunigen. Scope of delivery: • NVIDIA L40S