Es1 or Einsteinium is an institutional GPU cluster that was deployed to meet the growing computational demand for researchers doing machine learning and deep learning. The system is named after the chemical element with symbol Es and atomic number 99 which was discovered at Lawrence Berkeley National Laboratory in 1952 and in honor of Albert Einstein who developed the theory of relativity.
Es1 is a 52-node partition consisting of multiple GPU node types to address the different research needs. These include:
Nvidia H100: New – 4 ea. new Dell XE9680 8-way H100 GPU servers have been added. With 80GB memory per GPU, these nodes are suitable for training and running LLMs.
Nvidia Tesla V100: Es1 contains 15 ea. Intel processor nodes each with two Nvidia 16GB V100 cards. These are the highest performing GPUs in Es1. Users using GPUs for computation should consider using the V100 nodes as they have ECC memory to detect errors. Similar users needing tensorflow cores should also use these nodes.
Nvidia A40: Es1 contains 15 ea. Intel processor nodes each with four Nivdia A40 cards. These GPUs are excellent for machine learning, deep learning, image processing and some applications such as Cryo-EM and molecular dynamic simulations.
Nvidia RTX2080Ti: Es1 contains 18 ea. AMD EPYC processor nodes each with four Nvidia 11GB RTX2080Ti GPU cards. Users can roughly expect these GPUs to be about 73% as fast as the Tesla V100 for FP32 training and 55% as fast as the Tesla V100 for FP16 training.