Ksilink acquires one high performance GPU/TPU cluster from locally-based 2CRSI company to implement deep learning approaches to its drug discovery programs for cardiovascular and muscle diseases. Equipped with 4 NVIDIA Tesla V100 Tensor Core delivers up to 125 TeraFlops, the GPU cluster will accelerate AI, high performance computing (HPC), and graphics. In parallel, Ksilink upgrades Checkout, its C++ based in-house image analysis software co-developed with WD Expertise & Recherche to a multi-server version. Now acquired images could be processed by up to 184 CPUs with a 4x10Gb/s bandwidth which significantly decreases analysis time. “This is a unique asset as speed becomes a crucial parameter with more and more complex images to be analyzed” says Dr Arnaud Ogier, Ksilink’s Head of Data and Image Mining.