Congratulations to the winners of the 32 Core Testing Plan contest. Take a look at the descriptions of their projects as they get them posted to the forums and let us know what you think. We'll be reporting out their progress and posting their results soon.
|Computation of Automorphism Groups of q-ary Codes
Yekaterina Zhmud, Novosibirsk State University, has basic computations of symmetry groups that are estimated to take around 21 days. Without this data, testing the whole algorithm is impossible. What will it take for Yakaterina to change and improve the algorithm (using OpenMP) within hours or days, rather than weeks?
|Fast Spectral Estimation of Genetic Homology
Anton Pankratov and Ruslan Tetuev from Moscow State University and Russian Academy of Sciences wonder what it would be like to view the homology of all known genetic sequences as one single, albeit humongous image. Once generated, how would you browse such an image quickly?
|Matrix Multiplication, Performance and Scalability in OpenMP
Nicolas Wolovick, National University of Cordoba introduced matrix multiplication to his students this year and challenged them to create the fastest serial code. How will optimizing this code to fully utilize 32 cores change the exercise?
|Parallel Implementation of Lattice Boltzmann Methods
Miguel Montes, Instituto Universitario Aeronautico, wants to try out a parallel approach for the Lattice Boltzmann Method. Will the collision model proposed to address continuum-level physics of fluids benefit from manycore parallelism?
|Parallel Algorithms of Asynchronous Cellular Automata Simulation
Konstantin Kalgin, Novosibirsk State University, knows that there are several sophisticated parallel algorithms out there that address his questions, however without access to a true manycore environment he can only guess at the scalability. Can he create a new algorithm that fully utilizes the benefits of manycore?