General Information

Name

Aurelio Vivas

Area

Parallel Computing

Status

Graduated

Email

aurelio.vivas@correounivalle.edu.co

Affiliation

Universidad del Valle

Laboratory

Lascilab

Contributions

Prototype of a Parallel Programming Environment for the Construction of General Purpose Cellular Automata - Meritorious - BSc. Thesis

The study of dynamic systems such as fluids, climate, gases, among others, has allowed throughout history the observation and discovery of a variety of less physical phenomena. Among the various existing methods to carry out these observations is the Cellular Automata method. This particular method allows to observe the compartment of the system, and in many cases find the numerical solution to the differential equation that governs the behavior of the same. While it is true, this type of systems in the field of physics are modeled normally under the paradigm of sequential programming, which is contradictory since the models of Cellular Automata can take advantage of the computational power available in current parallel computers. That is why the present research work focused on the development of a tool that would allow these models of cellular automata to take advantage of the parallelism present in parallel computers.

Parallel Programming Techniques Applied to Physics Problems - Research Practice

In the last decade and a half computer architectures went from mono processors (single core) to be multi core systems in order to overcome the physical limitations imposed by increasing the number of transistors on a single processing unit. The number of cores by processors is then incremented and the computer clock frequency is decreased in order to reduce power consumption and heat generation. This change in processors design causes that applications running under the new architectures have worse performance than previous processors or single core. ¿How to solve this dilemma? New architectures requires to think about the traditional algorithmic solutions again, since mono-threaded algorithmic though not exploit the potential of multi core architectures. This practice aims to highlight the importance of popular parallelization technologies (classic and current), in order to illustrate how these can be used for problem solving in the scientific word.