Small but powerful, Pearl is a deep learning system with the capacity to greatly speed up the training process of complex artificial intelligence models. It comprises two tightly coupled, state-of-the-art NVIDIA DGX2 nodes, and a large array of highspeed storage. The system is expected to accelerate the research from multiple scientific domains. For instance, a complex model that would require about 500 hours on a single CPU can be performed on Pearl in under an hour.
For more information, find and download the Pearl leaflet here: 
Pearl flyer.pdf
