MSI Unveils EdgeXpert MS-C931 Based on NVIDIA DGX Spark: Detailed Review of a Compact AI Supercomputer
NVIDIA
Micro-Star International (MSI)
Realtek
MediaTek
MSI has introduced the EdgeXpert MS-C931, a version of the NVIDIA DGX Spark with improved cooling and Wi-Fi. This compact AI supercomputer, powered by the Grace Blackwell GB10 SoC with 128 GB of RAM and performance up to 1000 TOPS (FP4), can run models up to 200 billion parameters locally, ensuring privacy.
MSI has released the EdgeXpert MS-C931, based on the NVIDIA DGX Spark (formerly Project DIGITS). The device uses the NVIDIA Grace Blackwell GB10 SoC with 20 Arm cores (Cortex-X925/A725) and 6144 CUDA cores, 128 GB of LPDDR5x-8533 memory, and a 4 TB NVMe SSD. It delivers up to 1000 TOPS in FP4 and 32 TFLOPS FP32 performance. MSI has improved the cooling system (up to -15°C difference) and optimized the Wi-Fi antennas (speed boost up to +37%). The computer supports clustering via 200GbE QSFP56 and runs on NVIDIA DGX OS based on Ubuntu 24.04. It is designed for AI developers, data processing, and Edge AI, including machine vision and autonomous systems. Power consumption under load is up to 180 W, with low noise. It can run LLMs up to 200 billion parameters and fine-tune models up to 70 billion parameters.
Source: ServerNews —
original