Grafikkarten für diverse Anforderungen

Grafikkarte, NVIDIA GeForce RTX™ 3070 Ti GPU (Boost Clock: 1.815MHz), 8GB GDDR6X SDRAM (256bit, 19Gbps), 6.144 CUDA-Kerne, PCI Express 4.0 x16, 3x DisplayPort 1.4a und 2x HDMI 2.1 (max. 4 unterstützte Bildschirme), DirectX 12 Ultimate, OpenGL 4.6, NVIDIA G-Sync-fähig, AUTO-EXTREME-Technologie, MaxContact technology, RGB Lighting, GPU Tweak II, Real-Time Ray Tracing, 2,7-Slot Kühllüfter, Axial-tech Fan Design, Nvidia DLSS, Nvidia Studio, NVIDIA Ampere GPU-Technologie, NVIDIA Reflex, zwei kugelgelagerte Lüfter, 2nd Gen Ray Tracing, 3rd gen Tensor Cores, HDCP, L/H/B: 299.9 x 126.9 x 51.7mm (2.7 Slots), 36 Monate Bring-In Herstellergarantie

GPU-Rechenprozessor mit NVIDIA A40 GPU, ohne Lüfter (passive), Ampere Architektur, PCI Express 4.0 x16, 48GB GDDR6 ECC RAM, 10.752 Cuda Kerne, 84 RT Cores Gen 2, 336 Tensor Cores Gen 3, NVLink: 2-Way (112GB/s Bidirectional), 2-Slot Design, 384-bit Memory Interface, 696 GB/Sek. Memory Bandwidth, 3x DisplayPort 1.4 (deaktiviert by default), NVIDIA Virtual PC, NVIDIA RTX Virtual Workstation (vWS), NVIDIA Virtual Compute Server (vCS, no MIG), Stromanschluss: 8-pin CPU, 36 Monate limitierte Herstellergarantie

GPU-Rechenprozessor mit NVIDIA A2 GPU, ohne Lüfter (passive), Ampere Architektur, PCI Express 4.0 x8, 16GB GDDR6 ECC RAM, 1.280 Cuda Kerne, 10 RT Cores Gen 2, 40 Tensor Cores Gen 3, 1-Slot Design, 128-bit Memory Interface, 200GB/Sek. Memory Bandwidth, 1x Video Encoder, 2x Video Decoder (H.265, H.265 VP9, AV1 Decode), NVIDIA vApps, vPC, vWS, vCS, NVIDIA AI Enterprise, 36 Monate limitierte Herstellergarantie

GPU-Rechenprozessor inkl. passiver Kühlung, NVIDIA A16 GPU, 4x 16GB GDDR6 RAM inkl. ECC (4x 232GB/s), 5.120 CUDA Kerne (4x 1.280), 160 Tensor Cores (4x 40), 40 RT Cores (4x 10), PCI Express 4.0 x16, NVIDIA Ampere GPU-Technologie, NVIDIA Virtual PC (vPC), NVIDIA Virtual Applications (vApps), NVIDIA RTX Workstation (vWS), NVIDIA Virtual Compute Server (vCS), max. 250W Stromverbrauch (8-pin CPU Poweranschluss), 36 Monate limitierte Herstellergarantie

GPU-Rechenprozessor mit NVIDIA A100 GPU, ohne Lüfter (passive), Ampere Architektur, PCI Express 4.0 x16, 80GB HBM2e RAM, 6.912 Cuda Kerne, 108 Streaming Multiprocessors, 432 Tensor Cores Gen 3, NVLink: 2-Way (600GB/s Bidirectional), 2-Slot Design, 5.120-bit Memory Interface, 1.555GB/Sek. Memory Bandwidth, NVIDIA Virtual Compute Server mit MIG (Multi-Instance GPU) Support (max. 7 GPU Instanzen), max. 300W Stromverbrauch, Stromanschluss: 8-pin CPU, 36 Monate limitierte Herstellergarantie

Workstation Grafikkarte, NVIDIA T1000 GPU (Turing™), 896 CUDA-Kerne, 8GB GDDR6 SDRAM (128-bit, 160GBps), PCIe 3.0 x16 Low-Profile, 4x Mini DisplayPort, OpenCL, DirectCompute 5.0, DirectX 12, OpenGL 4.5, Shader Model 5.1, Nvidia Scalable Geometry Engine, NVIDIA Mosaic, Full-Scene Anti-Aliasing (FSAA), GPU Tessellation, NVIDIA GPU Boost, OpenGL Quad Buffered Stereo Support, Dynamic Parallelism, 1 Steckplatz für Lüfter, 16 x unabhängige anisotrope Texturfilter, NVIDIA GigaThread engine, Transparentes Multisampling, HDR Ready, interne 12-Bit Anzeige-Pipeline, 32K Texture and Render Processing, NVIDIA Turing GPU-Architektur, HDCP 2.2, 36 Monate Bring-In Herstellergarantie

Grafikkarte mit NVIDIA GeForce RTX 4090 GPU, 16.348 CUDA-Kerne, 24GB GDDR6X SDRAM, PCI Express 4.0 x16, 2.535MHz Core Clock, Dual Bios, GIGABYTE WINDFORCE-Kühlungsdesign, RGB-Fusion, NVIDIA Tensor Core, Nvidia RT Core, 3D Active Fan, Metall-Rückplatte, DirectX 12, OpenGL 4.6, 3x DisplayPort 1.4, 1x HDMI 2.1, L/B/H: 340x150.2x75.2mm (3.7 Slot), 36 Monate Herstellergarantie