Beijing Tiantan Hospital, Capital Medical University, was founded on 23 August 1956. It is a Grade-A tertiary general hospital led by neurosurgery, characterized by neuroscience, and integrating medical care, teaching, research, and prevention. It is one of the world’s three major neurosurgery research centers and Asia’s base for neurosurgery clinical care, research, and teaching.

In March 2018, Sugon won the bioinformatics HPC cluster procurement for Beijing Tiantan Hospital, Capital Medical University. The cluster’s overall double-precision peak performance reaches 423.168 trillion operations per second, including 80 dual-socket compute nodes with a peak of 186.368 trillion operations per second, 80 high-frequency blade compute nodes with a peak of 15.36 trillion operations per second, 5 fat nodes with a peak of 26.2656 trillion operations per second, and 10 GPU compute nodes with a peak of 195.1744 trillion operations per second (CPU+GPU). It uses the Sugon ParaStor300 distributed parallel storage system with 7.5 PB raw capacity. The compute network uses 100 Gb/s EDR InfiniBand with modular InfiniBand switches and full-system line-rate switching.

Once built, the system can substantially meet computational-simulation needs of frontier emerging life-science disciplines at Beijing Tiantan Hospital, Capital Medical University, including:

(1) Storage, processing, and computing needs of the hospital’s Clinical Trials and Research Center (CTRC) for clinical diagnosis and treatment information, omics data (whole genome, exome, RNA, and proteome) and high-resolution imaging data (about 10,000 people per year);

(2) Processing and analysis needs of the hospital’s monogenic neurological-disease diagnosis center for pathogenic-mutation and drug-metabolism gene DNA testing data from about 3,000 patients per year;

(3) The National Clinical Research Center for Neurological Diseases’ need to analyze large-scale multimodal medical imaging data, build an intelligent assisted-diagnosis system based on medical data, and support related deep-learning algorithm development, platform construction, and applications;

(4) Future needs for data collection, upload, sharing, and management in nationwide or regional projects.