System Description
The Mayflower®-LLMW33-SP3 is a powerful 19″ liquid-cooled GenAI server, specially designed for AI inference, model fine-tuning and training. At the heart of the system is the NVIDIA® H100 NVL GPU, which offer 188 GB HBM3 memory and 33792 CUDA cores. This GPU ensures outstanding AI performance and offers up to 7916 TFLOPs. If requested, the system can also be configured with two NVIDIA® H100 NVL GPUs to further increase performance.
The server is based on an AMD® EPYC™ processor and supports up to 1 TB of memory. The robust infrastructure includes a 2000 watt power supply unit, which can be configured redundantly on request. The system is equipped with hybrid cooling, consisting of two 120 mm fans and liquid cooling for the CPU and GPU. The GenAI server offers numerous interfaces, including 2x 10 GBit LAN and 2x 100 GBit LAN with NVIDIA® ConnectX®-6 Dx, as well as various USB and VGA connections. Both Microsoft® Windows® Server and Linux® are supported as operating systems.
The Mayflower®-LLMW33-SP3 combines high-performance components, industrial robustness and flexible configuration options, making it an ideal solution for demanding AI applications.
Find out more
about Generative AI.
