¿Qué puedes ejecutar en esto?
No cloud subscription. Your data stays on your hardware.
Image Generation
ML Training & Fine-tuning
Private AI / GDPR Compliance
Video AI
Software Development & Research
Especificaciones Técnicas
Cómputo
Modelo de GPU
NVIDIA Grace Blackwell
Arquitectura
Grace Blackwell
Rendimiento Tensor
Up to 1 PFLOP FP4
Núcleos Tensor
5th Generation
Núcleos RT
4th Generation
Núcleos CUDA
Blackwell Generation
Memoria y Almacenamiento
Memoria Unificada
128 GB LPDDR5x
Ancho de Banda de Memoria
273 GB/s
Interfaz de Memoria
256-bit
Almacenamiento
4 TB NVMe M.2 (self-encrypting)
Conectividad
Tarjeta de Red (NIC)
ConnectX-7 @ 200 Gbps
Ethernet
1x RJ-45, 10 GbE
Wi-Fi
Wi-Fi 7
Bluetooth
BT 5.4
Puertos USB
4x USB Type-C
Sistema
CPU
20-core Arm (10x Cortex-X925 + 10x Cortex-A725)
Consumo Energético
240W
Factor de Forma
Desktop
Pila de Software
NVIDIA AI Enterprise (PyTorch, TensorFlow, JAX, RAPIDS, TensorRT)
Casos de Uso
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