Supermicro GPU Super AI Station - NVIDIA GB300 Blackwell (ARS-511GD-NB-LCC-01-G2)
สอบถามสต็อกสินค้า ทักไลน์ @gearlab หรือ https://m.me/gearlabth
Supermicro Gold Series GPU Super AI Station บน NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip เหมาะสำหรับ Research Lab, AI Development, Data Science
- Key Features/Applications: Research Lab/National Lab, AI Development, AI Inference / Training, Data Science
- CPU: NVIDIA 72-core NVIDIA Grace™ CPU on NVIDIA GB300 Grace Blackwell Ultra Desktop Superchip
- Chassis: Tower or 5U Rackmount
- Drive: 2 1.92TB M.2 NVMe PCIe 4.0 + 2 960GB M.2 PCIe 4.0 NVMe Solid State Drives
- RAM: 496GB LPDDR5x SOCAMM Memory
- GPU: 1 Onboard NVIDIA Blackwell Ultra GPU, 252GB GPU HBM3e Memory
- Network Ports: NVIDIA ConnectX®-8 SuperNIC™ + 1 RJ45 10 GbE LAN Port + 1 Dedicated IPMI LAN Port
Data Center-Level Performance in a Local AI Supercomputer
Built on the same GB300 architecture that powers large AI factories around the world, the Super AI Station delivers up to 20 PFLOPS of FP4 AI performance and up to 3x the fine tuning and inference performance of H200 to support autonomous agents, AI development, AI inference, and data science. A 748GB combined coherent memory pool supports trillion-parameter models locally, allowing always-on agents to run 24/7, generating up to 1200tps at no additional token cost. Full local AI development capabilities eliminates reliance on cloud-based AI services, reducing security risks for proprietary data and offloading cloud subscription costs. The system offers native support for the full suite of NVIDIA AI Enterprise (NVAIE) software stack, including software tools, libraries, and frameworks for enterprise AI deployments.