Planificación multinivel eficiente con aprovisionamiento dinámico en grids y clouds

La consolidación de las grandes infraestructuras para la Computación Distribuida ha resultado en una plataforma de Computación de Alta Productividad que está lista para grandes cargas de trabajo. Los mejores exponentes de este proceso son las federaciones grid actuales. Por otro lado, la Computación Cloud promete ser más flexible, utilizable, disponible y simple que la Computación Grid, cubriendo además muchas más necesidades computacionales que las requeridas para llevar a cabo cálculos distribuidos. En cualquier caso, debido al dinamismo y la heterogeneidad presente en grids y clouds, encontrar la asignación ideal de las tareas computacionales en los recursos disponibles es, por definición un problema NP-completo, y sólo se pueden encontrar soluciones subóptimas para estos entornos. Sin embargo, la caracterización de estos recursos en ambos tipos de infraestructuras es deficitaria. Los sistemas de información disponibles no proporcionan datos fiables sobre el estado de los recursos, lo cual no permite la planificación avanzada que necesitan los diferentes tipos de aplicaciones distribuidas. Durante la última década esta cuestión no ha sido resuelta para la Computación Grid y las infraestructuras cloud establecidas recientemente presentan el mismo problema. En este marco, los planificadores (brokers) sólo pueden mejorar la productividad de las ejecuciones largas, pero no proporcionan ninguna estimación de su duración. La planificación compleja ha sido abordada tradicionalmente por otras herramientas como los gestores de flujos de trabajo, los auto-planificadores o los sistemas de gestión de producción pertenecientes a ciertas comunidades de investigación. Sin embargo, el bajo rendimiento obtenido con estos mecanismos de asignación anticipada (early-binding) es notorio. Además, la diversidad en los proveedores cloud, la falta de soporte de herramientas de planificación y de interfaces de programación estandarizadas para distribuir la carga de trabajo, dificultan la portabilidad masiva de aplicaciones legadas a los entornos cloud...

Reconfigurable computing for Monte Carlo simulations: results and prospects of the Janus project

We describe Janus, a massively parallel FPGA-based computer optimized for the simulation of spin glasses, theoretical models for the behavior of glassy materials. FPGAs (as compared to GPUs or many-core processors) provide a complementary approach to massively parallel computing. In particular, our model problem is formulated in terms of binary variables, and floating-point operations can be (almost) completely avoided. The FPGA architecture allows us to run many independent threads with almost no latencies in memory access, thus updating up to 1024 spins per cycle. We describe Janus in detail and we summarize the physics results obtained in four years of operation of this machine; we discuss two types of physics applications: long simulations on very large systems (which try to mimic and provide understanding about the experimental non equilibrium dynamics), and low-temperature equilibrium simulations using an artificial parallel tempering dynamics. The time scale of our non-equilibrium simulations spans eleven orders of magnitude (from picoseconds to a tenth of a second). On the other hand, our equilibrium simulations are unprecedented both because of the low temperatures reached and for the large systems that we have brought to equilibrium. A finite-time scaling ansatz emerges from the detailed comparison of the two sets of simulations. Janus has made it possible to perform spin glass simulations that would take several decades on more conventional architectures. The paper ends with an assessment of the potential of possible future versions of the Janus architecture, based on state-of-the-art technology.

Robust fitting of Zernike polynomials to noisy point clouds defined over connected domains of arbitrary shape

A new method for fitting a series of Zernike polynomials to point clouds defined over connected domains of arbitrary shape defined within the unit circle is presented in this work. The method is based on the application of machine learning fitting techniques by constructing an extended training set in order to ensure the smooth variation of local curvature over the whole domain. Therefore this technique is best suited for fitting points corresponding to ophthalmic lenses surfaces, particularly progressive power ones, in non-regular domains. We have tested our method by fitting numerical and real surfaces reaching an accuracy of 1 micron in elevation and 0.1 D in local curvature in agreement with the customary tolerances in the ophthalmic manufacturing industry.

Tethered Monte Carlo: computing the effective potential without critical slowing down

We present Tethered Monte Carlo, a simple, general purpose method of computing the effective potential of the order parameter (Helmholtz free energy). This formalism is based on a new statistical ensemble, closely related to the micromagnetic one, but with an extended configuration space (through Creutz-like demons). Canonical averages for arbitrary values of the external magnetic field are computed without additional simulations. The method is put to work in the two-dimensional Ising model, where the existence of exact results enables us to perform high precision checks. A rather peculiar feature of our implementation, which employs a local Metropolis algorithm, is the total absence, within errors, of critical slowing down for magnetic observables. Indeed, high accuracy results are presented for lattices as large as L = 1024.

Tethered Monte Carlo: computing the effective potential without critical slowing down

We present Tethered Monte Carlo, a simple, general purpose method of computing the effective potential of the order parameter (Helmholtz free energy). This formalism is based on a new statistical ensemble, closely related to the micromagnetic one, but with an extended configuration space (through Creutz-like demons). Canonical averages for arbitrary values of the external magnetic field are computed without additional simulations. The method is put to work in the two-dimensional Ising model, where the existence of exact results enables us to perform high precision checks. A rather peculiar feature of our implementation, which employs a local Metropolis algorithm, is the total absence, within errors, of critical slowing down for magnetic observables. Indeed, high accuracy results are presented for lattices as large as L = 1024.

Janus: an FPGA-based system for high-performance scientific computing

This paper describes JANUS, a modular massively parallel and reconfigurable FPGA-based computing system. Each JANUS module has a computational core and a host. The computational core is a 4x4 array of FPGA-based processing elements with nearest-neighbor data links. Processors are also directly connected to an I/O node attached to the JANUS host, a conventional PC. JANUS is tailored for, but not limited to, the requirements of a class of hard scientific applications characterized by regular code structure, unconventional data manipulation instructions and not too large data-base size. We discuss the architecture of this configurable machine, and focus on its use on Monte Carlo simulations of statistical mechanics. On this class of application JANUS achieves impressive performances: in some cases one JANUS processing element outperfoms high-end PCs by a factor ≈1000. We also discuss the role of JANUS on other classes of scientific applications.