Was können Sie damit ausführen?
No cloud subscription. Your data stays on your hardware.
Image Generation
ML Training & Fine-tuning
Private AI / GDPR Compliance
Video AI
Software Development & Research
Technische Spezifikationen
Rechenleistung
GPU-Modell
NVIDIA Grace Blackwell
Architektur
Grace Blackwell
Tensor-Leistung
Up to 1 PFLOP FP4
Tensor-Kerne
5th Generation
RT-Kerne
4th Generation
CUDA-Kerne
Blackwell Generation
Speicher & Lagerung
Unified Memory
128 GB LPDDR5x
Speicherbandbreite
273 GB/s
Speicherinterface
256-bit
Speicher
4 TB NVMe M.2 (self-encrypting)
Konnektivität
Netzwerkkarte (NIC)
ConnectX-7 @ 200 Gbps
Ethernet
1x RJ-45, 10 GbE
Wi-Fi
Wi-Fi 7
Bluetooth
BT 5.4
USB-Anschlüsse
4x USB Type-C
System
CPU
20-core Arm (10x Cortex-X925 + 10x Cortex-A725)
Leistungsaufnahme
240W
Formfaktor
Desktop
Software-Stack
NVIDIA AI Enterprise (PyTorch, TensorFlow, JAX, RAPIDS, TensorRT)
Anwendungsfälle
Large Language Models, Image Generation, ML Training & Fine-tuning, Private AI / GDPR Compliance, Video AI, Software Development & Research
Frameworks
PyTorch, TensorFlow, JAX, RAPIDS, TensorRT, ONNX, Docker