OVERVIEW OF USAGE OF NSCC SUPERCOMPUTER IN THE FIRST YEAR
The National Supercomputing Centre Singapore (NSCC) has introduced the first-ever 1 PFlops high-performance computing (HPC) system (named Aspire01) in Singapore to researchers and scientists from the key stakeholders, NUS, NTU, SUTD and AStar, for about one year.
After the NSCC system has been in service, many researchers have carried out large scale and impactful computational jobs on it. Researchers from the Cancer Science Institute of Singapore (CSI) and the Department of Civil and Environmental Engineering (CEE) had demonstrated significant performance and benefits of the system on their big scale analyses involving huge data-set and large scale parallel computations, respectively. Please refer to their showcase project here.
In the following, a quick overview of the facts about the utilization of the system are presented based on the weekly usage statistics data.
Top Used Applications
It is not surprised to notice that all the top used applications on the NSCC system are open-source program, except VASP which is usually licensed to be used within a research group.
Departments that Consumed the Most Computing Resources
The researchers from the following departments consumed the most computing resources on the NSCC system in the past one year.
• The Department of Mechanical Engineering (MPE)
• The Department of Civil & Environmental Engineering (CEE)
• The Department of Chemistry (CHM)
• The Department of Biological Sciences (DBS)
• The Department of Biomedical Engineering (BIE)
• The Science Faculty
• Cancer Science Institute of Singapore (CSI)
• The Duke-NUS (GMS)
Overview of Job Size and Waiting Time
The aim of building NSCC system is to support large scale cutting edge computational research work. With this in mind, higher priority is given to large scale parallel computing jobs that require hundreds or thousands of CPU cores for execution. Comparatively, lower priority is given to small jobs and median sized jobs.The below table gives the overview of job sizes and the computing resources consumed and mean waiting time of these jobs.
| Job Size | CPU Cores Required Per Job | Computing Resources Consumed | Mean Waiting Time (Hours) |
|---|---|---|---|
| Small | ≤ 24(1 compute node only) | 21% | 5~38 |
| Median | ≤ 240(≤ 10 compute nodes) | 22% | 13 ~ 132 |
| Large | > 240 (> 10 compute nodes) Up to 3000 cores or more |
57% | 14 ~ 38 |
Support of Large Data Analyses
Apart from the supporting of large scale parallel computational simulations, NSCC system provisions large and high performance storage space in supporting large data analyses.In the first year of service, four users have consistently used more than 100 TB storage space for their large data related analyses, with the top two usage cases at about 600+ ~ 800+ TB. There are also another dozen users who have used more than 50+ TB storage space.
Enrollment and Access of NSCC System
NUS staff and student researchers can access and use the NSCC system free of charge for the time being. Researchers can enroll the access of NSCC system using their NUS-ID/password at help.nscc.sg. After the enrollment, you can log into the system via the login nodes installed in NUS at nus.nscc.sg via SSH connection or via the web portal at https://nusweb.nscc.sg. Both login methods are limited to NUS campus network.Detailed instructions of use of NSCC system can be found at NSCC’s webpage.
