MSI AI Edge: 126 TOPS AI Performance in a 4-Liter Chassis

Compact Infrastructure for Edge AI

MSI announced its new desktop computer series optimized for Edge AI computations. Featuring a compact 4-liter form factor, the AI Edge model is designed especially for AI projects, engineering work, and development processes that require data to be processed on the local device.

Hardware Architecture and Memory Management

At the core of the system are AMD Ryzen™ AI Max+ 300 series processors. The top model of the series, Ryzen AI Max+ 395, reaches a total AI compute power of 126 AI TOPS thanks to its unified platform structure. An RDNA™ 3.5 architecture with 40 CUs and an XDNA 2 NPU are used as the graphics processing unit.

The device’s most critical technical advantage is its memory architecture, which aims to overcome traditional VRAM bottlenecks:

• Compound Memory: The system comes with up to 128 GB of LPDDR5X 8000 MHz onboard memory.
• GPU Sharing: Up to 96 GB of the total memory can be dynamically allocated to the GPU.
• Model Capacity: This structure enables Large Language Models (LLM) with up to 120 billion parameters to run on the device and offers an output speed of about 15 TPS (tokens/second).

Software and Operating System Support

Designed with developers’ need for flexible work in mind, MSI AI Edge supports both Windows and Linux operating systems. Thanks to RAG (Retrieval-Augmented Generation) technology, the system can perform secure and contextual queries on local datasets. MSI’s Glacier Armor thermal cooling solution is used to ensure stability under prolonged load.