Exactly what it sounds like: an extremely powerful computer. Supercomputers are designed to do enormous amounts of calculations at the same time and are used by a wide range of organisations, including F1 racing, Disney/Pixar and Facebook. They are also extremely important to push medicine and science forward - without Supercomputers we would not be able to study the birth of stars, investigate climate change or design new medicines to name just a few.
Each Supercomputer is unique and around the world great Supercomputers have been given great names to match. Irish Supercomputers have previously been named after famous scientists. We are now asking you to help name Ireland's brand new Supercomputer.
The website has been filled with information on our previous Supercomputers and our new arrival and should give you lots of inspiration to help come up with a new name!
The competition is open to all schools/CoderDojos across Ireland. All you have to do is watch the video below and read about previous ICHEC Supercomputers; then get together with your classmates and teacher/mentor to see what name you think best suits our brand new Supercomputer. We have also provided some simple guidelines to follow.
Then you simply fill out the online entry form at the end of this page before the closing date of September 30th 2013. We will be announcing the winner entry at a special naming ceremony during Science Week - 10-17th November 2013. More details of the naming ceremony to follow.
The winning class or dojo and their teacher/mentor will be invited to the naming ceremony during Science Week - 10-17th November 2013.
Each student will also receive a specially designed T-shirt with the new name on the front and a photo of the Supercomputer on the back. They will also be taken on a trip to the TSSG hosting facility at the Waterford Institute of Technology to see the Supercomputer itself!
The following animation gives you an insight into the world of Supercomputing, how big Supercomputers are, what they can be used for and who uses them. It also highlights their importance and why having the chance to give our new Supercomputer it's new name is a great opportunity for you and your school/CoderDojo.
The new Supercomputer that you have been given the chance to name arrived at the hosting facility on 22nd August 2013. The machine was installed over a period of seven days. Over the coming weeks the machine will undergo testing and it will open to the research community in October 2013. The diagram below shows you how the new Supercomputer is pieced together and gives you an idea of just how powerful it is.
320 nodes (7680 cores; 20TB RAM)
16 nodes + 32 Intel Xeon Phi 5110P
16 nodes + 32 NVIDIA Tesla K20X
1 UV2000 node + 2 Intel Xeon Phi 5110P
DataDirectNetwork SFA12k-20 Storage, Lustre filesystem with 560TB
10GBE Switches, KVM Switch
The new Supercomputer was purchased from SGI after a competitive dialogue process and was funded by Science Foundation Ireland (SFI) at a total cost of €4.1M. This Supercomputer is a hybrid machine capable of running many different applications and workflows. The machine is made up of four components: Thin, Hybrid, Fat and Service. The technical details of each component are described above. The distinction between Thin and fat relates to the quantity of RAM each node contains, while the Hybrid name comes from the fact it contains various processing architectures. This is a commonly used convention within Supercomputing. More background information on the terminology used in Supercomputing can be found below.
The Thin component is the largest with 320 nodes or 7680 cores made of 2.4GHz Intel Ivy Bridge cores. Each node has 64GB of RAM and is connected using FDR InfiniBand. This amounts to 20TB of RAM! The Hybrid partition contains 32 nodes with 32 Intel Xeon Phi 5110P's plus 32 NVIDIA K20X's. The Fat section is a UV2000 where 1.7TB of RAM is accessible to 112 cores of Intel Sandy Bridge and 2 Xeon Phi's. In addition to these partitions, there will be additional service servers plus 560TB of DDN storage using a Lustre filesystem.
This new machine will quadruple computing resources previously available to scientists in Ireland (and this is only considering the Thin partition), besides providing access to the latest technology from Intel, Ivy Bridge processors + Xeon Phi coprocessors, and NVIDIA, Tesla accelerators, to researchers across Ireland. The storage component for the system is from DataDirect Networks (DDN). Overall, the new Supercomputer will enable scientists to solve their scientific problems quicker.
A wide range of new research and development (R&D) will be enabled on the new machine. 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 new machine will run non-stop for the next four years. It will provide an estimated 295,000,000 hours of computation not counting the power of the accelerators!!
Similarities: The hardware inside the new Supercomputer is very similar to that found in a laptop. All nodes have CPUs, RAM and network connectivity but are without a hard drive. Some nodes have graphics cards called GPGPUs that are slightly different to regular video cards but capable of solving scientific problems. Other nodes have co-processors or accelerators that are similar to GPGPUs where they allow some of the work to be offloaded from the CPUs.
Dissimilarities: The largest difference between a laptop and Supercomputer is the size. The new machine will span across 10 racks and require significantly more power. The mechanism which the individual components within the machine are connected is also different to standard house or small office networks.
Power requirements: This Supercomputer will require 170kW of power to run. This is equivalent to over 350 average Irish homes after taking into account the data centre's Power Usage Effectiveness (PUE) of 1.4 and assuming the average household consumes around 5,300 kWh of electricity per year. Your average desktop requires approx. 50-250 watts while your laptop is significantly less power hungry with a requirement of 20-60 watts
Fast moving industry: the new machine would've been the most powerful in the world if it had been installed in June 2005! We expect that the general public will have the equivalent processing power on their mobile phones or tablets by 2040!! This is not unreasonable prediction as the Cray 2 was the fastest Supercomputer in 1985 which had a similar capability as today's Apple iPad!
Companies from areas including Aeronautics (NASA, Airbus Industry, Rolls-Royce, etc.), Automative industry (Mercedez-Benz, Porsche, BMW, Renault, Ferrari, etc.), Food Industry (e.g., Kellogg Company etc.), Cleaning products / Detergents (e.g. Proctor & Gamble, ), Oil & Gas Industry (Tullow Oil, Total, Shell, etc.), Social Media (Facebook, Twitter, etc.) and Search (Google, Yahoo, etc.) use Supercomputing to solve industry related problem.
Even Animation Studios such as Pixar and Dreamworks use Supercomputers! Watch how a bunch of penguins explain the world of High Performance Computing or HPC (Supercomputing and HPC are synonymic)!!
In Ireland, each day Met Éireann, the Irish Meteorological Service, determines the weather forecast by running complex mathematical models every six hours on a Supercomputer. This provides accurate predictions of such things like wind speed/direction and rain. Forecasts also give advanced warning of severe weather such as storms or frost.
InfiniBand Networking: The new Supercomputer has a FDR InfiniBand network set up with an enhanced hypercube which reduces the overheads of passing information from one node to another. This is essential when your calculations span across 100s of nodes and the program passes information across the network. The FDR InfiniBand is capable of sending 56Gbps from one node to another. The latency or time delay to send information within the network is 1 microseconds. Access to the machine from outside the data centre will be provided by a 10Gb Ethernet connection from HEANet, Ireland's National Eduction & Research Network. Access will be restricted to researchers within higher education institutes.
General-Purpose Graphics Processing Units: Also known as GPGPUs, they are similar to graphics cards found in desktops and laptops that people use to process photos or play video games. Each Tesla card or K20X can provide 1.31 TFLOPS of peak-double precision floating-point performance and has 2,699 CUDA cores.
Intel Xeon Phi Coprocessors: Each 5110P card can provide 1.01 TFLOPS of double-precision floating-point performance.
Shared memory: Modern software applications and scientific algorithms require more and more RAM. The UV2000 will provide researchers with over 1.7TB of RAM. These resources will open up new research directions for scientists and allow them to carry out work that would have otherwise been impractical or impossible.
There are hundreds of Supercomputers around the world - here is a small selection from the top 100:
For details of more Supercomputers visit the TOP500.org site for the latest rankings and names for inspiration.
Performance: The field of Supercomputing is littered with facts and figures. You'll see a lot tera's, peta's, exa's etc. These describe the number of computations or floating point (FLOPS) calculation that the machine performs in one second. The fastest machine in the world at present (Tianhe-2 or MilkyWay-2) can perform over 33 PetaFLOPS. This is 33 with 15 zeros or 1015!! More on prefixes like peta can be found here.
Node/Socket/Core: A node can be thought of as a computer and the core is the processing unit where the calculations are performed. A socket or processor is a collection of cores. Each node of the Thin section of the new machine will have two sockets, each with 12 cores.
GPGPUs/Coprocessors: Both terms relate to processing units different to that of the traditional CPU. GPGPUs come from a graphic cards background, while the coprocessors are very similar to modern CPU cores but at much lower performance. Both are very green and provide large performance with lower power requirements than traditional cores. The downside to these technologies is that they are quite new and can be difficult to enable new programs to take advantage of this computational power.
As we mentioned, great machines deserve great names and the following information relates to previous generations of Supercomputers that were operated by ICHEC and have since been decommissioned. Each of these machines, in their own way, played a part in scientific advances and have left their mark on history - just as their name sakes did. We hope these magnificent machines will help inspire discussion between you and your class to help you come up with a suitable name for our newest Supercomputer.
Named after the famous Austrian physicist, Erwin Schrödinger developed important underlying theories of modern quantum chemistry and is most famous for the Schrödinger equation. He also spent a lot of time thinking about cats and worked in the Dublin Institute of Advanced Studies from 1939 to 1955.
The Schrödinger Supercomputer was an IBM BlueGene/P machine that was operational from January 2008 to January 2011. The system used PowerPC 450 processors running at 860MHz, providing 4096 cores with 2TB of RAM. It featured the BlueGene Tree/Torus interconnect system to enable super fast inter-processor communication that is essential for supercomputing workloads. The system had a peak performance of 13.83 TFLOP.
More about the work of Erwin Schrödinger can be found here.
This system was named after Sir George Gabriel Stokes, the Irish mathematician, physicist and politician. Some of his most famous work was in the field of fluid dynamics where he helped develop the Navier-Stokes equations. These equations are used to this day to perform weather predictions. A machine called Navier was used in the past as a backup system to Stokes.
Stokes is an SGI Altix ICE 8200EX cluster with 320 compute nodes. Each compute node has two Intel (Westmere) Xeon E5650 hex-core processors and 24GB of RAM. This results in a total of 3,840 cores and 7,680GB of RAM available for jobs. The system had a peak performance of 41.01 TFLOP.
Stokes was installed in December 2008 and will be decommissioned in late 2013. More details can be found here.
More details of the work of Sir George Gabriel Stokes can be found here.
Lanczos was named after the Hungarian mathematician and physicist, Cornelius Lanczos. He also worked in the Dublin Institute of Advanced Studies succeeding Schrödinger. He was a pioneer in the field of mathematics and created various algorithms for digital processors.
The Lanczos system was an IBM BlueGene/L. The system used PowerPC 440 processors running at 700MHz, providing 2,048 cores with 1TB of RAM. It was operational from January 2008 to July 2010 and had a peak performance of 5.73 TFLOP. Like the Schrödinger machine, it also featured the BlueGene Tree/Torus network.
Additional information of the life and work of Cornelius Lanczos can be found here.
The Nobel laureate Ernest Walton gave his name to the first ICHEC Supercomputer. Walton was the first person to split the atom that has led to nuclear power.
The Walton machine was an IBM Cluster 1350 consisting of 479 IBM e326 compute nodes. Each compute node was a dual AMD Opteron 250 - 2.4 GHz single core CPUs. A total of 415 of these nodes had 4 GB of RAM, while the remaining 64 nodes had 8 GB of RAM.
The nodes were connected together with Gigabit Ethernet using a Force10 E600 switch. Storage was provided via the IBM GPFS parallel filesystem running on a set of dedicated storage servers connected to an IBM DS4500 storage controller. An additional 14 nodes provided load-balanced login, cluster management, scheduling, filesystem and other facilities for the cluster.
The system achieved a peak performance of 4.464 TFLOP.
More information on Ernest Walton can be found here.
William Hamilton is most famous for engraving the equations for Quaternions on the Broome Bridge in Dublin (1843). These equations are still in use today; as an example, they are often part of the underlying algorithms that allow objects to be rotated in 3D in computer graphics.
The Hamilton machine was a Bull NovaScale 6320 that provided 32 Intel Itanium2 CPUs and 256 GB of RAM as a single system image. Each CPU ran at 1.5 GHz and had 6 MB of L3 cache.
The processors were connected using a three tier NUMA topology providing inter-process communication via shared memory. Storage was provided by 9 TB of directly attached disks. This provides optimal performance for codes using large scratch files.
Hamilton had a Bull NovaScale 4400 for use as a login node with 4 more Itanium2 CPUs and 8 GB of RAM for compilation, batch submission along with pre- and post- processing.
More information of how some of William Hamilton's work can be seen every day in computer graphics can be found here.
George Stoney worked on electricity and is most noted for introducing its basic unit, the electron.
Stoney is a Bull Novascale R422-E2 cluster with 64 compute nodes. Each compute node has two 2.8 GHz Intel (Nehalem EP) Xeon X5560 quad-core processors and 48 GB of RAM. This results in a total of 496 cores and 2,976 GB of RAM available for jobs.
The nodes are interconnected via InfiniBand (DDR) providing high bandwidth and low latency for both computational communications and storage access. Storage is provided by an EMC CLARiiON CX4-120 SAN with 21 TB (formatted) of capacity to the compute nodes via two Lustre filesystems. Each compute node also provides 870 GB of local scratch capacity on a directly attached hard disk.
Stoney also provides ICHEC's National GPU Service with 24 of the compute nodes having two NVIDIA Tesla M2090 cards each. Both cards provide 512 GPU cores each, 6 GB of local GDDR5 memory and a theoretical peak double-precision performance of 665 GFLOPS. Stoney will be decommissioned in 2014.
More information regarding George Stoney's work can be found here.
We have a few guidelines to naming our new Supercomputer that we would like you to consider before you submit your entry. They are very simple and will help you with choosing a name.
To create a good name try and keep your entry to one word only so that it is catchy and easy to remember.
Try and avoid any special characters or numbers. Stick with the A-Z alphabet.
A very long name doesn't have the same impact - 8 characters or less is best.
Supercomputers don't like being called 'Blinky' or 'Fluffy'.
As well as having the chance to name a Supercomputer, ICHEC are also running a second competition with CoderDojo where students will compete to gain access to the new Supercomputer.
The competition is split into two phases where the first will challenge students in writing algorithms and optimising code. Use of the C/C++ language is required as this lends itself to working on GPGPUs and coprocessors that will be available to winners of the first phase during a coding challenge during Science Week 2013. More details of the coding competition can be found on the Super Coder Competition page.
Have you come up with a winning name? Then fill in the form below and click submit. It's that simple! We will notify you of the results shortly after the competition closes on September 30th, 2013.