Nvidia Tesla M40 12GB CUDA Accelerator

Deep learning models typically take days to weeks to train, forcing scientists to make compromises between accuracy and time to deployment. The NVIDIA Tesla M40 GPU accelerator, based on the ultra-efficient NVIDIA Maxwell™ architecture, is designed to deliver the highest single precision performance. Together with its high memory density, this made the Tesla M40 the world’s fastest accelerator for deep learning training in its generation.

Ideal for Machine Learning, Large Language Models and AI Environments, the Tesla M40 has a large 12GB of vMem which can be utilised for compute or virtualised to accelerate virtual desktops.

The Tesla M40 is best utilised in a server environment, it is fully passively cooled in front to back configuration with the server standard layout and requires one PCI-E 8-pin GPU power cables to a total draw of up to 250W.

The Tesla M40 is a FHFL PCI-E card, which means it requires two full height full length PCI-E slots with the card stretching 10.5" (267mm).