Kay

Kay is ICHEC's primary supercomputer and Ireland's national supercomputer for academic researchers. Information about how to access Kay and HPC resources under the National Service can be found here.

Aug. 2018
Installation Date
665 TFlop/s
Linpack Performance

System

Kay (kay.ichec.ie) is comprised of a number of components:

  • "Cluster" -  A cluster of 336 nodes where each node has 2x 20-core 2.4 GHz Intel Xeon Gold 6148 (Skylake) processors, 192 GiB of RAM, a 400 GiB local SSD for scratch space and a 100Gbit OmniPath network adaptor. This partition has a total of 13,440 cores and 63 TiB of distributed memory.
  • "GPU" - A partition of 16 nodes with the same specification as above, plus 2x NVIDIA Tesla V100 16GB PCIe (Volta architecture) GPUs on each node. Each GPU has 5,120 CUDA cores and 640 Tensor Cores.
  • "Phi" - A partition of 16 nodes, each containing 1x self-hosted Intel Xeon Phi Processor 7210 (Knights Landing or KNL architecture) with 64 cores @ 1.3 GHz, 192 GiB RAM and a 400 GiB local SSD for scratch space.
  • "High Memory" - A set of 6 nodes each containing 1.5 TiB of RAM, 2x 20-core 2.4 GHz Intel Xeon Gold 6148 (Skylake) processors and 1 TiB of dedicated local SSD for scratch storage.
  • "Service & Storage" - A set of service and administrative nodes to provide user login, batch scheduling, management, networking, etc. Storage is provided via Lustre filesystems on a high-performance DDN SFA14k system with 1 PiB of capacity.

A wide range of new research and development (R&D) is enabled on Kay. These include increased resolution in weather and climate forecasting, larger and longer simulations for research in areas such as medical device development, nanotechnology, genomics, drug design, etc. The machine is also capable of running heterogeneous workflows that require large compute power and large amounts of memory either during the pre- or post-processing phases of researchers work. Detailed list of the software available can be found here.

 

Scheduling Policy

Queue Node Type Max nodes per Job Max Walltime
DevQ Cluster 4 1 hour
ProdQ Cluster 40 72 hours
LongQ Cluster 1 144 hours
GpuQ GPU 4 48 hours
PhiQ Phi 16 24 hours
ShmemQ Large Mem 1 72 hours