Founded on Oct 5, 2018, the Center of Computational Science and Engineering of SUSTech has two HPC clusters named “Qi-Ming” and “Tai-Yi” respectively. For the university’s faculty of the teaching and research series, the two clusters are free to use for testing purpose currently.

Please follow the format of the account application form to provide relevant information if you want to apply for an account. Should you have any questions, please feel free to contact hpc@sustech.edu.cn.

Current regulations of the clusters are as follows:

1. Account application: Each principle investigator (PI) can apply for a Group, and it is not allowed that multiple PIs apply for one Group together. In each Group, there is a User Account belonging to the PI. To add a User Account for a research group member to the Group, please fill in the attached form and send it to hpc@sustech.edu.cn. If the account application email is to be sent by a research group member, please also send a carbon copy to the corresponding PI at the same time.

2. Storage partitioning:

   • The storage space of the “Tai-Yi” cluster is divided into 3 partitions: data, scratch, and work. Each Group has a quota of 2.0 TB in the data partition and a quota of 500 GB in the work partition. No space limit applies to the scratch partition, but there is a regular clean-up which removes files generated prior to 2~4 weeks ago. (The configuration depends on specific use condition and data generated recently will not be cleaned.)

   • The “Qi-Ming” cluster has only one home partition for user storage currently. Each Group has a storage space of 2 TB (including the auto-backup space). CCSE may adjust the configuration according to space occupation situations at any time. A user can contact hpc@sustech.edu.cn to disable his/her auto-backup routine.

3. Job queuing: during the test, the parameters of the job submission & queuing system will be adjusted according to specific usage data to maximize the utilization of computational resources and meet our users’ needs to the utmost extent.

   • The “Tai-Yi” cluster currently opens the following 5 queues:

     1. smp: Queue of flat nodes (6TB RAM/node). Number of cores per node: 192; Suggested number of cores for jobs: multiple of 24; Maximum running time: 12 hours; Priority: medium.
     2. medium: Queue of long-time jobs. Number of cores per node: 40; Suggested number of cores for jobs: Multiple of 20; Maximum running time: 72 hours; Priority: low.
     3. short: Queue of short-time jobs. Number of cores per node: 40; Suggested number of cores for jobs: Multiple of 20; Maximum running time: 20 hours; Priority: high.
     4. debug: Queue of debug jobs. Maximum running time: 1 hours; Priority: high.
     5. ser: Queue of debug jobs. Maximum running time: 24 hours; Priority: low.

   • The “Qi-Ming” cluster currently opens the following 6 queues:

     1. fat: Queue of flat nodes (6TB RAM/node). Number of cores per node: 144; Suggested number of cores for jobs: multiple of 18; Maximum running time: 24 hours; Priority: medium.
     2. cal-s: Queue of short-time jobs. Number of cores per node: 24; Suggested number of cores for jobs: 12; Maximum running time: 24 hours; Priority: high.
     3. cal-l: Queue of long-time jobs. Number of cores per node: 24; Suggested number of cores for jobs: 12; Maximum running time: 72 hours; Priority: low.
     4. gpu: Queue of GPU computing jobs. Each node has 24 CPU cores and 4 Tesla K80 GPUs. Only jobs which use GPU devices are allowed to submit to this queue. Maximum running time: 24 hours; Priority: medium.
     5. debug: Queue of debug jobs. Number of cores per node: 24; Maximum running time: 1 hours; Priority: high.
     6. ser: Queue of serial jobs. Number of cores per node: 24; Maximum running time: 24 hours; Priority: low.

4. Software installation assistance: If you need assistance in software installation, please carefully fill in the application form of software installation assistance and submit it together with the source code of the software. Please upload the source code package to the cluster and set down the path correctly in the application form. For commercial software, please send related EULA to hpc@sustech.edu.cn and, at the same time, send a manually-signed printed application form to Ms. Li Yun at Room 307, Block 5, Wisdom Valley. We will try our best to offer relevant services. However, we cannot guarantee the timeliness of software installation due to the limited manpower of CCSE and we hope for your understanding.

Please refer to the user manual of “Tai-Yi” or “Qi-Ming” for detailed usage. Should you have any questions on the usage or find any abnormal runtime status of the clusters, please contact hpc@sustech.edu.cn or Mr. Zhou Wengui (18610351001) in time, or contact our resident engineers (Mr. Hu Changqing (13066918269) for problems concerning “Qi-Ming”; Mr. Zhong Huaming (15818796162) or Mr. Chen Qiwang (18129975513) for problems concerning “Tai-Yi”).