Presentation on theme: "CLGrid: Chilean Grid initiative Grid Coordination : CMM Alejandro Jofre Juan Carlos Maureira, Gonzalo Hernandez, Eduardo Moreno."— Presentation transcript:
CLGrid: Chilean Grid initiative Grid Coordination : CMM Alejandro Jofre Juan Carlos Maureira, Gonzalo Hernandez, Eduardo Moreno
CMMGrid Computing2 CLGrid: Chilean Grid Initiative Dec 2005: Chilean Grid Initiative @ CMM Scientific applications of national interest At present: 3 Universities actives in CLGrid 4 Universities joining. Activities: Workshop 1 and 2 + Intel software college (Scientific presentations, coordination reunions, technical talks, tutorials) Grid programming: Testbeds Cluster implementation: hardware, OS, middleware Cluster programming: Parallel & distributed
CMMGrid Computing3 Current Universities Univ. de Chile (Uchile) Univ. Católica del Norte (UCN) Univ. Austral de Chile (AUCH) Univ. de la Frontera (UFRO) Univ. de Concepcion (Uconce) Univ. de la Serena Pontifícia Univ. Católica. (PUC) Univ. Federico Santa Maria.
CMMGrid Computing4 CLGrid: Chilean Grid Initiative Web site –http://clgrid.reuna.clhttp://clgrid.reuna.cl 5 Clusters Running –2 in Antofagasta (UCN) –2 in Santiago (CMM) –1 in Valdivia (UACH) 1 Testbed working using Ninf-G. –easy deploy of code and data. 3 new testbed on their way to run soon. –2 Bioinformatics –transportation –1 Sky exploring (image analysis)
CMMGrid Computing8 CLGrid: CMM Grid Initiative Applied research areas –Network Economics Energy and telecommunications –Ensemble Weather forecasting MM5, WRF, RCA, Match –Bioinformatics Blast, unfold, hmmler, interpf GAMESS, Gromacs –Transportation networks Stochastic network equilibrium Discrete event network simulation
CMMGrid Computing9 CLGrid: Chilean Grid Initiative Grid Research Areas –Cluster Operating Systems improvements. –Grid Filesystem deployment and performance. –Application deployment. –Central accounting system. –Grid Security and Trust. Grid Current Users –Bioinformatics Protein Threader (CMM), GAMESS (Fac. Quim), Gromacs (Fac. Quim) –Astronomers (UCN) Moving Objects in the Sky.
CMMGrid Computing10 CLGrid: Chilean Grid Initiative Future Applications –Network public transportation analysis –Ensemble Weather forecasting Parallel by using Mpich-Gx Distributed by using GridRPC. –Bioinformatic data analysis Blast Portal InterProScan Portal Protein Folding Base Stacking
MTE: Practical Issues Code originally written in Matlab Code rewritten in C using Intel MKL Libraries (VML, BLAS, PARDISO) Tuned for Intel Itanium 2 (but portable to other architectures) Parallelized using MPI Library Computation Time Chicago Network (546 destinations, 2176 streets) Santiago network (2159 destinations, 7587 streets)
The model engine that runs on the cluster feeds on-line a graphical interface to show the simulated parameters of the network. With this interface the Transit managers (or researchers) can analyze the simulated scenario and introduce new changes into the simulation boundary conditions. Parallel Discrete Event Simulation: Public Transportation System
Optimization and Equilibrium in the Telecommunications Market Keywords Deregulated market models Mobile and fixed local phone systems Equilibrium/optimization problem and high performance computing We have generated new models, calculation algorithms and processes for implementation and validation of efficient business models that allow the setting of rates for the services provided by landline and mobile phone companies.
Optimization and Equilibrium in the Telecommunications Market GENERAL MODEL – MSC LOCALIZATION Number of BTS (Base station Transceiver Station) Number of BSC (Base Station Controller) For each geographic zone Optimization Model Number & location of MSC (Mobile Switching Center) Regional Problem Separation
Optimization and Equilibrium in the Telecommunications Market GENERAL MODEL – FREQUENCY ASSIGNMENT BTS IDEAL CONFIGURATION There is no Interference BTS REAL CONFIGURATION Channel Interference Separation by District & Bands Santiago Graph
Optimization and Equilibrium in the Energy Market Keywords Energy Markets Regulation Hydrothermal Coordination Equilibrium/optimization problem and high performance computing We have created new models and improved some algorithms for coordinating the use of water reservoirs in hydrothermal energy systems. Especially using Stochastic Programming. We are also working on the problem of Optimal Expansion of Electrical Power Systems.
Optimization and Equilibrium in the Energy Market HYDROTHERMAL COORDINATION MODEL The main objective is to establish an appropriated policy of administration and use of the water reservoirs in an electrical power system
Bioinformatics GridHybrid –Virtual Microarray –Massive simulation of DNA-DNA binding –A several millions of energy minimization problems (5000 variables each one) –Currently applied to biomining research areas.