BSC

PUMPS Summer School, July 7-11, 2014

  • First, let's thank all the participants, teachers and TA's for the great summer school we have had!

  • And specially, we want to congratulate the students that won the Best Poster Award. Thanks for their presentations at the poster session and the great work done during the lab sessions.

Best Poster Awards

Andrea Miele

Title: Cofactorization on graphics processing units
Author(s): Andrea Miele, Joppe W. Bos, Thorsten Kleinjung, and Arjen K. Lenstra
Affiliation: EPFL, Switzerland
Poster: Cofactorization on graphics processing units

Abstract: The number field sieve (NFS) is the most efficient algorithm for factoring RSA moduli. Understanding how efficiently it can be implemented on heterogeneous systems is key to the security assessment of the widely used RSA cypto-system. We show how the cofactorization step, a compute-intensive part of the relation collection phase of the number field sieve (NFS), can be farmed out to a graphics processing unit. Our implementation on a GTX 580 GPU, which is integrated with a state-of-the-art NFS implementation, can serve as a cryptanalytic co-processor for several Intel i7-3770K quad-core CPUs simultaneously. This allows those processors to focus on the memory-intensive sieving and results in more useful NFS-relations found in less time.

Piotr Przymus

Title: Lightweight compression algorithms for database systems supported by GPU devices
Author(s): Piotr Przymus, Krzysztof Kaczmarski, Paweł Rzążewski
Affiliation: Faculty of Mathematics and Computer Science, Nicolaus Copernicus University, Poland
Poster: Lightweight compression algorithms for database systems supported by GPU devices

Abstract: In past several years many authors have confirmed the effectiveness of GPU utilisation in experimental database systems. Due to limitations of a GPU programming model, in most of these applications CPU and GPU work as a host and a coprocessor respectively. This cooperation is dominated by memory operations and limited by a PCI-e bandwidth. Additionally, GPU memory space is usually smaller than CPU RAM. In this work, we discuss a usage of Lightweight compression algorithms for databases supported by the General-Purpose computation on Graphics Processing Units (GPGPU) in order to achieve significant savings of memory space, an improvement of DISK↔RAM↔GPU data transfer, cache utilisation and, in some cases, also processing speed. The poster covers basic concepts, current research directions as well as results of experiments.

 
awards.txt · Last modified: 2015/10/25 13:06 by nacho
Drupal Garland Theme for Dokuwiki