Logo of Science Foundation Ireland  Logo of the Higher Education Authority, Ireland7 CapacitiesGPGPU Research Projects
Ireland's High-Performance Computing Centre | ICHEC
Home | News | Infrastructure | Outreach | Services | Research | Support | Education & Training | Consultancy | About Us | Login

Fionn Supercomputer

Hardware

Fionn (fionn.ichec.ie) is a heterogenous machine made up of four components: Thin, Hybrid, Fat and Service.

  • The Thin component is an SGI ICE X system of 320 nodes or 7680 cores made of 2.4GHz Intel Ivy Bridge cores. Each node has 2x12 core processors, 64GB of RAM and is connected using FDR InfiniBand. This amounts to 20TB of RAM across the partition.
  • The Hybrid partition contains 32 nodes. Each of these nodes has 2x10 core 2.2GHz Intel Ivy Bridge with 64GB of RAM. This partition has accelerators from Intel and NVIDIA. 16 nodes have 32 Intel Xeon Phi 5110P's while the other 16 have 32 NVIDIA K20X's.
  • The Fat section is an SGI UV2000 where 1.7TB of RAM is accessible to 112 cores of Intel Sandy Bridge (14x8 cores processors) and 2 Intel Xeon Phi 5110P's.
  • The final component contains a set of service and administrative nodes to provide user login, batch scheduling, management, tape backup, switches, etc. Storage is provided via a DDN SFA12k-20 with 560TB of capacity to all components of the machine via a Lustre filesystem.

Fionn quadruples the computing resources previously available to scientists in Ireland (and this is only considering the Thin partition). It also provides access to the latest technology from Intel, Ivy Bridge processors + Xeon Phi coprocessors, and NVIDIA, Tesla accelerators, to researchers across Ireland.

A wide range of new research and development (R&D) will be enabled on Fionn. These potentially include an 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 will also be 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.

The name, Fionn, stems from the Irish word Fionnachtana which means the action of discovery. This name was chosen from a competition run for primary and secondary students in Ireland and aptly describes the work carried out on the machine.

Software

A large number of scientific software packages, libraries, compilers and other tools are installed and maintained on Fionn by ICHEC. A full listing of these is available on the software page.

Scheduling Policy

The following queues are available on Fionn:

  Job Size (cores) Maximum Walltime Maximum Available Cores
DevQ 24 - 96 30 minutes 192
ShortQ 24 - 1032 24 hours 6912
ProdQ 24 - 1032 72 hours 6912
LongQ 24 144 hours 768
GpuDevQ 20 30 mins 20
GpuQ 20-160 48hrs 320
PhiDevQ 20 30 mins 20
PhiQ 20-160 24hrs 320
ShmemQ 8-112 48hrs 112

Note:

  • Users should avoid specifying a queue in their PBS scripts. Specifying the walltime and processor requirements will result in the job being routed to the optimum queue.
  • Processing cores on Fionn are allocated in multiples of 24 (one node). This means that users can only request core counts of 24, 48, ...
  • Hard and soft throttling limits are in place to prevent a single user from flooding the entire system.
  • Queued jobs gain priority at different rates depending on many factors.

Documentation

Introductory documentation for Fionn is provided in our documentation section.

Photos

Some photos of the Fionn system:

Acknowledgement

Fionn was purchased from SGI after a competitive dialogue process and was co-funded by Science Foundation Ireland (SFI) at a level of €3.7M and industry funds at a total cost of €450k. The naming competition was funded by an outreach project from SFI Discover Science and Engineering.