Software configuration
About the OS
As of March 2019, our Supercomputer System has been using Linux as its OS, and the Thin compute nodes use the RedHat Enterprise 6 distribution. This is common to all compute nodes.
| System | OS | OS Version |
|---|---|---|
| Gateway node | Red Hat Enterprise Linux 7 | 7.5 |
| Thin compute node | CentOS 7 | 7.5 |
| Medium compute node | Red Hat Enterprise Linux 7 | 7.5 |
| Fat compute node | Red Hat Enterprise Linux 7 | 7.5 |
About the job management system
Our Supercomputer System utilizes Univa Grid Engine (UGE) as the job management system. UGE is a commercial product derived from the open source software Sun Grid Engine6.2U5 owned by the former Sun Micro Systems, and so its underlying system concept, user command system, and so forth, are identical to those of Sun Grid Engine. For basic procedures for using the system utilizing UGE, please reference How to use the system.
Job-related upper limit settings in UGE
The number of job slots that can be used simultaneously by one user is as follows:
| Setting item | Setting value |
|---|---|
| Upper limit on the number of job slots (accounts for general research) | 300 |
| Upper limit on the number of job slots (accounts for large-scale use) | 300 |
| Upper limit on the number of jobs to enter per user | 5000 |
| Upper limit on the number of tasks to enter per user | 75000 |
By applying for an expansion of computer resources, the upper limit on the number of job slots can be expanded to the level of accounts for large-scale use. Please make an application if necessary. (* We are currently preparing. Please wait for a while.)
Queue configuration
The following queues are set up depending on the type of compute node used and the period of use (as of Mar. 11, 2019)
Phase3 System
| Queue name | Number of job slots | Maximum memory | Upper limit for execution time | Purpose | Options for queue specification |
|---|---|---|---|---|---|
| epyc.q | 4,224 | 512G | 62 days | When there is no resource request in two months of the execution period | No specification or -l month (default) |
| intel.q | 1,472 | 384G | 62 days | Use of Intel Xeon | -l intel or -l month -l intel |
| gpu.q | 384 | 512G | 62 days | Use of GPU | -l gpu or -l month -l gpu |
| medium.q | 160 | 3T | 62 days | Use of Medium compute node | -l medium or -l month -l medium |
| login.q | 258 | 512G | Unlimited | Used to execute qlogin from the gateway node | |
| login_gpu.q | 48 | 384G | Unlimited | Qlogin from gateway node when using GPU | |
| short.q | 744 | 192G | 3 days | Short period job | -l short |
For specific procedures for entering jobs and the job management system, please see How to use the system.
File transfer software
For data transfer between the system and an external entity, sftp, scp, and aspera can be used. For specific transfer procedures, see How to transfer file.
