Institute of Physics

Versión en Español

Advance Computing Project - Physics Institute, UNAM

The Physics Institute initiative for advance computing has its origins in a computing support project submitted to CONACYT (National Scientific Council) in the year 1999, this was motivated by the tradition of having "on the edge" tools to support the research at the Physics Institute and at the University (UNAM), so the Institute staff could explore and implement new technologies to be used in scientific research that were being used all around the globe. A proposal was made to adquire and implement two computing clusters with the Beowulf architecture to do numeric calculations in parallel. This has evolved to two clusters using Intel x86, x86_64 AMD64 and Opteron, Motorola G4 and G5. This technology has been serving the academic body of the institute for almost 10 years and has given us the opportunity of having access to a computing paradigm that is of great support to our labor in scientific research.

The project was registered with the title "Laboratorio de simulación de sistemas de muchos cuerpos: de átomos a materia condensada" ("Simulation lab of systems of many bodies: from atoms to condensed matter") (CONACYT / G32723-E) it has now expanded sucessfully to other project in the institute. The team in the original proposal were Matias Moreno PhD, Rocio Jauregui PhD, Karo Michaelian PhD, Ernesto Belmont PhD, Carlos Bunge PhD, Cecilia Noguez PhD, Octavio Miramontes PhD, Rubén Santa María PhD, Guillermo Ramírez PhD, Víctor Romero PhD and Manuel Torres PhD.

The project now has a teaching lab (masterlab) with an small cluster for this purpose and has two clusters of Beowulf architecture.

The ollin cluster is formed by 12 nodes, one master and 11 slaves, of which, 8 are Dual Core Intel Xeon (TM) and 4 nodes more with Dual Core AMD Opteron.

The brodix cluster named after Tomás Brody Phd, initially was an OpenMosix cluster, but due to the limitations found in this architecture we moved it to a Beowulf cluster again it is actually formed by 25 nodes, a master and 24 slave nodes for processing from which 18 nodes are Dual Core Intel Pentium 4 and 7 have Dual AMD Athlon MP 2600+.

The complete description of the hardware of both cluster is decribed in the corresponding section at this site.