GPU Small Queue Now Available

  • by: 
    Matthew Edeker
  • Published: 
      Lonestar6 has a new queue, "gpu-a100-small" designed for users who do not need a node's entire 3 GPUs in the "gpu-a100" queue. The jobs in this queue consume 1/3 of the full resources of a full node. 

      If your job uses only a single GPU and less than 70GB of memory, you should try running in the "gpu-a100-small" queue. 

      The charge rate for the "gpu-a100-small" queue is 1.5 SUs per node hour. 

      Please submit any questions you may have through the TACC Consulting System or feedback form.