6.1 Taito-GPU hardware and operating system

The Taito cluster includes a separate partition of compute nodes with dedicated GPU accelerator cards. Currently in the system there are 26 nodes containing 4 Pascal P100 GPUs each, 12 nodes with 2 dual GPU K80  GPUs each and 38 nodes with two K40 GPUs.

The P100 nodes consists of 26 Dell PowerEdge C4130 servers with:

  • 2x Xeon E5-2680 v4 CPUs with 14 cores each running at 2.4GHz
  • 512 GB of DDR4 memory
  • 4x P100 GPUs connected in pairs to each CPU
  • 2x800GB of Sata SSD scratch space

The K80 nodes consists of 12 Dell PowerEdge C4130 servers with:

  • 2x Xeon E5-2680 v3 CPUs with 12 cores each running at 2.5GHz
  • 256 GB of DDR4 memory
  • 2x K80 GPU cards each with 2 GPUs for a total of 4 GPUs per node, these are all connected to the first CPU
  • 850GB of HDD scratch space

The K40 nodes consists of 38 Bullx B715 servers mounted in Bullx B700 Direct Liquid Cooled (DLC)  racks:

  • 2x Xeon E5-2620-v2 CPUs with 6 cores each running at 2.1GHz, these have hyperthreadding enabled allowing 12 threads to run per CPU
  • 32 GB of DDR3 memory
  • 2x K40 GPUs each connected to a separate CPU
  • 500 GB of HDD scratch space

The different kinds of nodes are connected to each other using FDR InfiniBand which is then connected to the Taito fabric allowing the use of the same storage environment. Note that the P100 nodes can be moved between Taito and the Pouta cloud environment meaning not all 26nodes will be present all the time in Taito. The nodes share the same operating system and offer a similar modules environment as the Taito cluster, with the some of the modules in Taito-GPU being built and optimized for GPU usage.

Previous chapter     One level up     Next chapter