Batch System

OpenPBS software optimizes job scheduling and workload management in high-performance computing (HPC) environments – clusters, clouds, and supercomputers – improving system efficiency and people’s productivity. Built by HPC people for HPC people, OpenPBS is fast, scalable, secure, and resilient, and supports all modern infrastructure, middleware, and applications.

Batch queue list

TEM farm provides multiple batch queues with different characteristics to users who submit jobs to analyze large-scale Cryo-EM data. A batch queue means a logical set of CPU and GPU computing resources. Users interact with a specific queue to manange their own jobs. Within each queue, submitted jobs are executed in order (First-in-first-out). Note that multiple jobs requiring CPU and/or GPU resources can be executed concurrently if there are enough available resources in the queue.

_images/batch.png

Category

Queue Name

Assigned Computing Resources

Remarks

Shared

cpuQ

  • tem-wn[1001-1002]-el7.sdfarm.kr (36 cores and 384GB memory per node)

  • tem-wn[1003-1013]-el7.sdfarm.kr (28 cores and 192GB memory per node)

  • 380 Physical CPU cores

gpuQ

  • tem-gpu[01-03]-el7.sdfarm.kr (28 cores, 2 P100 GPUs and 384GB mem.)

  • tem-gpu[04-05]-el7.sdfarm.kr (28 cores, 2 P40 GPUs and 256GB mem.)

  • tem-gpu[06-07]-el7.sdfarm.kr (32 cores, 2 A100 GPUs and 256GB mem.)

  • tem-gpu[08-10]-el7.sdfarm.kr (32 cores, 4 A100/V100 GPUs and 384GB mem.)

  • 300 Physical CPU cores

  • 26 GPUs

  • P100 16GB device memory

  • P40 24GB device memory

  • V100 32GB device memory

  • A100 40GB device memory

  • A100 80GB device memory