.

Artemis

  • A magnificently flexible architecture adapts to the extensibility from the all-in-one to the cloud scales to empower the wide-spectrum AI training applications​
  • The most powerful AI cluster platform with up to 72 NVIDIA latest GPUs fully connected through 5th generation NVLink switch trays.​
  • Based on MGX 1U platform design for compute trays. ​
  • Adapt diverse liquid cooling solutions: In rack / Sidecar/ Row based​
  • Proven liquid cooling design with the Inventec-designed leakage detection and handling features​
  • Great serviceability with blind-mate design: implement manifold and NVLink Cartridge to reduce rack level cabling and piping per system. ​
  • 19’ chassis for compute tray and NVLink switch: can slot into 19” EIA rack and 21” ORv3 rack.

Artemis, a new 44OU rack that adopts the NVIDIA MGX-based rack solution and complies with 1U 2P4GPU with the NVIDIA GB200 Grace Blackwell Superchip.

 

Inventec has been making significant strides in research and development across AI, automotive, electronics, 5G and the metaverse. Inventec’s flexible and agile engineering and production capabilities further meet the diverse needs of growing large language model (LLM) applications and AI users.      

 

Artemis contains the NVIDIA GB200 NVL72, a 72-GPU fifth-generation NVIDIA NVLink-connected GPU system , and supports the new NVIDIA Quantum-X800 and Spectrum-X800 networking platforms to deliver powerful AI communication performance.

 

The NVIDIA Blackwell architecture features six transformative technologies for generative AI and accelerated computing that will help unlock breakthroughs in many areas, such as quantum computing, electronic design automation, and computer-aided engineering. Moreover, Artemis will tap support from NVIDIA AI Enterprise, an end-to-end software platform for production-grade AI, for the NVIDIA GB200 Superchip.

 

Item
Description
Title
Document Type
Date
Back