Magerit is the name of the one of the most powerful supercomputers in Spain. It also reached the second best Spanish position in the
TOP500
The TOP500 project ranks and details the 500 most powerful non-distributed computing, distributed computer systems in the world. The project was started in 1993 and publishes an updated list of the supercomputers twice a year. The first of these ...
list of
supercomputer
A supercomputer is a computer with a high level of performance as compared to a general-purpose computer. The performance of a supercomputer is commonly measured in floating-point operations per second ( FLOPS) instead of million instructions ...
s.
This computer is installed in
CeSViMa
The Supercomputing and Visualization Center of Madrid (CeSViMa), also called Madrid Supercomputing and Visualization Center (in Spanish, Centro de Supercomputación y Visualización de Madrid), depends on the computer science faculty of the Techn ...
, a research center of the
Technical University of Madrid
The Technical University of Madrid or sometimes called Polytechnic University of Madrid ( es, Universidad Politécnica de Madrid, UPM) is a public university, located in Madrid, Spain. It was founded in 1971 as the result of merging different Te ...
.
Magerit was first installed in 2006 and reached the 9th fastest in Europe and the 34th in the world,
the second best position of a Spanish supercomputer in the list. It also reached the 275th position in the first
Green500
The Green500 is a biannual ranking of supercomputers, from the TOP500 list of supercomputers, in terms of energy efficiency. The list measures performance per watt using the TOP500 measure of high performance LINPACK benchmarks at double-precisi ...
list published. It is no longer among the TOP500.
The second version, installed in 2011 reached the 1st position of Spain, 44th of Europe and 136th fastest of the world.
It also reached the 18th position in the
Green500
The Green500 is a biannual ranking of supercomputers, from the TOP500 list of supercomputers, in terms of energy efficiency. The list measures performance per watt using the TOP500 measure of high performance LINPACK benchmarks at double-precisi ...
list.
''Magerit'' (for ''*Materit'' or ''*Mageterit'') is the most ancient recorded name of the current city of
Madrid
Madrid ( , ) is the capital and most populous city of Spain. The city has almost 3.4 million inhabitants and a metropolitan area population of approximately 6.7 million. It is the second-largest city in the European Union (EU), and ...
. The name comes from the Celtic name of a fortress built on the Manzanares River in the 9th century
AD, and means "Place of abundant water".
History
First steps (2005)
Magerit was created as a collaboration between
Technical University of Madrid
The Technical University of Madrid or sometimes called Polytechnic University of Madrid ( es, Universidad Politécnica de Madrid, UPM) is a public university, located in Madrid, Spain. It was founded in 1971 as the result of merging different Te ...
and
IBM. The computer is housed in the newly created
CeSViMa
The Supercomputing and Visualization Center of Madrid (CeSViMa), also called Madrid Supercomputing and Visualization Center (in Spanish, Centro de Supercomputación y Visualización de Madrid), depends on the computer science faculty of the Techn ...
. This first version had only 124 nodes and was housed temporarily in the Computer Science School of Madrid. The funding was provided by the Spanish Ministry of Education and Science and the Autonomous Region of Madrid.
Joining the Spanish Supercomputer Network (2006–2007)
Late 2006 CeSViMa joins
Spanish Supercomputing Network
The Spanish Supercomputing Network (RES) is a distributed infrastructure involving the interconnexion of 12 supercomputers which work together to offer High Performance Computing resources to the scientific community. It is coordinated by the B ...
(Red Española de Supercomputación or RES in Spanish) and the supercomputer was upgraded. The new configuration has 1204 nodes reaching a speed of 14
TFLOPS
In computing, floating point operations per second (FLOPS, flops or flop/s) is a measure of computer performance, useful in fields of scientific computations that require floating-point calculations. For such cases, it is a more accurate meas ...
. This is considered the first version due to its inclusion in the TOP500 list in the 34th position, the second best position of a Spanish supercomputer in the list.
In 2007 the first users from the access committee of Spanish Supercomputing Network (the agreement makes that the Network can schedule the use of the 68% of the resources) and users managed at local (CeSViMa) access committee (using the other 32%).
Migration and small upgrades (2008–2010)
In May 2008, CeSViMa and Magerit supercomputer migrated to a new building in the same campus (only 500 meters from previous location at Computer Science School).
The computer was upgraded: change of communication switch, storage subsystem and replacement of some blades with a new version. This upgrade increase the power of the supercomputer near 2
TFLOPS
In computing, floating point operations per second (FLOPS, flops or flop/s) is a measure of computer performance, useful in fields of scientific computations that require floating-point calculations. For such cases, it is a more accurate meas ...
reaching 15.95 TFLOPS. This upgrade did not avoid the fall from the TOP500 list in November 2008
In this configuration the 59.7% of the supercomputer CPU time is assigned via RES access committee and 40.3% is assigned via CeSViMa policies.
One year later, in 2009, the operating system and other system software was upgraded (migrating to
SUSE Linux Enterprise Server
SUSE Linux Enterprise (often abbreviated to SLE) is a Linux-based operating system developed by SUSE. It is available in two editions, suffixed with Server (SLES) for servers and mainframes, and Desktop (SLED) for workstations and desktop compu ...
10)
During 2010, CeSViMa acquire a new massive storage system with 1 petabyte of capacity in parallel with the own storage of Magerit.
Upgrade (2011)
In the first half of 2011, the supercomputer was fully upgraded replacing all computer nodes and interconnexion networks with the latest technologies in only one month (a record time)
[Nota de prensa de la ]UPM UPM may refer to:
* Ultra-pure metal
* UPM (company), UPM-Kymmene Oyj, a pulp and paper company
* Union pour la méditerrannée, Mediterranean Community
* Union for a Popular Movement, opposition party of France
* Unit production manager, someone ...
Technical University of Madrid installs the most powerful supercomputer in Spain
This configuration reached the 136th position in the TOP500 list
and the 18th position in the related
Green500
The Green500 is a biannual ranking of supercomputers, from the TOP500 list of supercomputers, in terms of energy efficiency. The list measures performance per watt using the TOP500 measure of high performance LINPACK benchmarks at double-precisi ...
list (both widely used as the supercomputer reference ranking) becoming the most powerful supercomputer and ecological supercomputer in Spain
The new distribution of use is 80% managed by CeSViMa-UPM access committee and 20% managed by Spanish Supercomputing Network. Although the RES managed percent is lower, the resources donated to the network increased 4–5 times.
The upgrade does not include the storage subsystem (maintain the storage upgraded in 2008). There is a small upgrade planned in next few years to adapt the storage system to the new requirements.
Architecture
Two versions of the supercomputer can be considered:
* The original 2006 (the 124 nodes of the agreement of 2005 was included in this configuration) with a small upgrade in 2008.
* The full upgrade in 2011 that makes Magerit the first supercomputer of Spain.
First version (2005–2010)
This setup reached the second best position in the TOP500 list (34th, November 2006). When this version enters in production it reach the 2nd of Spain, 9th of Europe and 34th of the world in the TOP500
list and the 275th position in the first Green500 list
The final version setup (reached after the upgrade of 2008) is a cluster of 1204 nodes
eServer
IBM eServer was a family of computer servers from IBM. Announced in 2000, it combined the various IBM server brands (AS/400, Netfinity, RS/6000, S/390) under one brand. The various sub-brands were at the same time rebranded from:
*IBM RS/6000 to ...
BladeCenter
The IBM BladeCenter was IBM's blade server architecture, until it was replaced by Flex System in 2012. The x86 division was later sold to Lenovo in 2014.
History
Introduced in 2002, based on engineering work started in 1999, the IBM eSe ...
(1036 JS20 and 168 JS21, both
PowerPC
PowerPC (with the backronym Performance Optimization With Enhanced RISC – Performance Computing, sometimes abbreviated as PPC) is a reduced instruction set computer (RISC) instruction set architecture (ISA) created by the 1991 Apple Inc., App ...
64-bit
In computer architecture, 64-bit Integer (computer science), integers, memory addresses, or other Data (computing), data units are those that are 64 bits wide. Also, 64-bit central processing unit, CPUs and arithmetic logic unit, ALUs are those ...
) under
SUSE Linux Enterprise Server
SUSE Linux Enterprise (often abbreviated to SLE) is a Linux-based operating system developed by SUSE. It is available in two editions, suffixed with Server (SLES) for servers and mainframes, and Desktop (SLED) for workstations and desktop compu ...
9.
* Each JS20 node has two processors
IBM PowerPC single-core 970FX (two cores) with 2.2
GHz
The hertz (symbol: Hz) is the unit of frequency in the International System of Units (SI), equivalent to one event (or cycle) per second. The hertz is an SI derived unit whose expression in terms of SI base units is s−1, meaning that one he ...
, 4
GB of
RAM
Ram, ram, or RAM may refer to:
Animals
* A male sheep
* Ram cichlid, a freshwater tropical fish
People
* Ram (given name)
* Ram (surname)
* Ram (director) (Ramsubramaniam), an Indian Tamil film director
* RAM (musician) (born 1974), Dutch
* ...
and 40 GB of local
hard disk
A hard disk drive (HDD), hard disk, hard drive, or fixed disk is an electro-mechanical data storage device that stores and retrieves digital data using magnetic storage with one or more rigid rapidly rotating platters coated with magnet ...
.
* Each JS21 node has two processors IBM PowerPC dual-core 970FX (four cores) with 2.2 GHz, 8 GB of RAM and 80 GB of local hard disk.
The system has a distributed storage system with a capacity of 190
TB under
GPFS
GPFS (General Parallel File System, brand name IBM Spectrum Scale) is high-performance clustered file system software developed by IBM. It can be deployed in shared-disk or shared-nothing distributed parallel modes, or a combination of these. It ...
. The access to this shared storage is provided by a high bandwidth switch that allows peaks of 1
Tbit/s
In telecommunications, data-transfer rate is the average number of bits (bitrate), characters or symbols (baudrate), or data blocks per unit time passing through a communication link in a data-transmission system. Common data rate units are multi ...
.
All the nodes are interconnected with a low latency (2.6 – 3.2
μs) and high bandwidth network called
Myrinet
Myrinet, ANSI/VITA 26-1998, is a high-speed local area networking system designed by the company Myricom to be used as an interconnect between multiple machines to form computer clusters.
Description
Myrinet was promoted as having lower protocol ...
. This network is used only for
MPI messages of users' tasks.
Finally, an auxiliary
Ethernet
Ethernet () is a family of wired computer networking technologies commonly used in local area networks (LAN), metropolitan area networks (MAN) and wide area networks (WAN). It was commercially introduced in 1980 and first standardized in 198 ...
network is deployed for administration tasks.
Second version (2011)
This setup converts Magerit into the most powerful supercomputer of Spain.
When this setup enters in production stage in 2011, it reach the first position of Spain, 44th of Europe and 136th of the world.
The system maintains the cluster architecture with 245 PS702 nodes, each one with 16 cores in two 64-bit processors
POWER7
POWER7 is a family of superscalar multi-core microprocessors based on the Power ISA 2.06 instruction set architecture released in 2010 that succeeded the POWER6 and POWER6+. POWER7 was developed by IBM at several sites including IBM's Roche ...
(eight cores each) 3.0
GHz
The hertz (symbol: Hz) is the unit of frequency in the International System of Units (SI), equivalent to one event (or cycle) per second. The hertz is an SI derived unit whose expression in terms of SI base units is s−1, meaning that one he ...
, 32
GB of
RAM
Ram, ram, or RAM may refer to:
Animals
* A male sheep
* Ram cichlid, a freshwater tropical fish
People
* Ram (given name)
* Ram (surname)
* Ram (director) (Ramsubramaniam), an Indian Tamil film director
* RAM (musician) (born 1974), Dutch
* ...
and 300
GB of local hard disk. Each core provides 18.38 Gflops.
The interconnection was replaced with an
Infiniband
InfiniBand (IB) is a computer networking communications standard used in high-performance computing that features very high throughput and very low latency. It is used for data interconnect both among and within computers. InfiniBand is also used ...
network, a high-bandwidth (40
Gbit/s
In telecommunications, data-transfer rate is the average number of bits (bitrate), characters or symbols (baudrate), or data blocks per unit time passing through a communication link in a data-transmission system. Common data rate units are multi ...
) and low latency (0.3
μs). The system maintains two independent Gigabit Ethernet for auxiliary tasks: deployment of images and access to storage subsystem.
The storage system remains the same (192
TB under
GPFS
GPFS (General Parallel File System, brand name IBM Spectrum Scale) is high-performance clustered file system software developed by IBM. It can be deployed in shared-disk or shared-nothing distributed parallel modes, or a combination of these. It ...
) with a bandwidth near 1
Tbit/s
In telecommunications, data-transfer rate is the average number of bits (bitrate), characters or symbols (baudrate), or data blocks per unit time passing through a communication link in a data-transmission system. Common data rate units are multi ...
.
The upgrade includes an update of the software: operating system (
SLES11SP1), deployment system (
xCAT, eXtreme Cluster Administration Toolkit) and all software and libraries used in the system.
Third version (2019)
Magerit upgraded with Lenovo
ThinkSystem SD530 nodes.
Use
Magerit processes
batch job
Computerized batch processing is a method of running software programs called jobs in batches automatically. While users are required to submit the jobs, no other interaction by the user is required to process the batch. Batches may automatically ...
s with large processing requirements, such as models of the universe, simulations of materials and climate models. An example of project is the project
Cajal Blue Brain
The Blue Brain Project is a Swiss brain research initiative that aims to create a digital reconstruction of the mouse brain. The project was founded in May 2005 by the Brain and Mind Institute of ''École Polytechnique Fédérale de Lausanne'' (EP ...
(Spanish participation in Blue Brain Project).
These jobs are organized by a queue manager. Due to the characteristic of the jobs (runs in hundred of CPUs a few days) its impossible to use more conventional access to the resources. The supercomputer must be running jobs without interrupts all the year.
The use of a queue manager of batch jobs allows a global scheduling of the resources increasing the use of the resources and a fair play between users.
Access to resources
The system is available to any person, institution or company that requests access via:
* Directly
CeSViMa
The Supercomputing and Visualization Center of Madrid (CeSViMa), also called Madrid Supercomputing and Visualization Center (in Spanish, Centro de Supercomputación y Visualización de Madrid), depends on the computer science faculty of the Techn ...
, filling the request for access forms on CeSViMa
web page.
* As a collaboration agreement with CeSViMa
* Via Spanish Supercomputing Network. This is a competitive process. The access committee evaluates all the projects and can assign resources on any other supercomputer of the network so it can be scheduled in the 20% of Magerit resources managed by RES.
References
External links
{{commons, UPM#CeSViMa, Magerit
CeSViMa – Centro de Supercomputación y Visualización de Madridwhere Magerit is installed
Hardware and software environmentTimelapse of the building of second version of Mageritarchived i
Ghostarchive.orgon 17 April 2022
IBM supercomputers
Lenovo supercomputers
Spanish Supercomputing Network
Computer-related introductions in 2005
2005 establishments in Spain