Specifications include, but are not limited to: Deep Learning Server (NIGP commodity code 204-00) to help on research projects. Guo’s group is working on two research projects, both of which require many Graphics Processing Unit (GPU) computing resources. The first one is acceleration for distributed deep learning with GPUs, which aims at developing novel algorithms using multiple GPU cards at the same moment to learn one large deep neural network (DNN). The second project is to design novel DNNs for protein identification. The input is very high dimensional (> 50,000), and the training sample size is large (>100,000), which requires lots of GPU memory (>40 GB). It also needs computation resources for tuning the hyperparameters, which may be accomplished by multiple GPUs.