Efficient
Total-Exchange in Wormhole-Routed Toroidal Cubes
F.
Petrini
Abstract.
The
total-exchange is one of the most dense communication
patterns and
is at the heart of numerous applications and programming models in parallel
computing. In this paper we present a simple randomized algorithm to efficiently
schedule the total exchange on a toroidal mesh with wormhole switching. This
algorithm is based on an important property of the wormhole networks that reach
high performance
under uniform traffic using adaptive routing. The experimental results,
conducted on a 256 nodes bi-dimensional torus, show that this algorithm reaches
a very high level of performance, around 90% of the optimal bound, and is more
efficient than other algorithms presented in the literature.
A
universal fixpoint semantics for ordered
logic
E.
Laenens, D. Vermeir
Ordered
logic is the theoretical foundation of the LOCO programing language [9] which
combines the declarative elegance and power of logic programming with asvantages
of object-oriented systems. Ordered logic is based on a partially ordered
structure of logical theories or objects. Objects are entities that may contain
positive as well as negative information represented by rules. The partial order
allows for the definition of a preference structure on these objects and
consequently also on the information they contain.
The result is a simple yet powerful logic that models classical as well
as non-monotonic inference mechanisms. The central issue of this paper is the
definition of a universal fixpoint semantics for ordered logic programs which
constitutes an important extension and generalization of the fixpoint semantics
prresented in [11, in the sense that it computes all partial models
(well-founded and stable partial models included) instead of only ītotalī
models (a possibly empty subset of the stable partial models), thus overcoming
the limitations of the previous approach.
ON
DISCONTINUOUS OPTICAL FLOW
S.
S. Beauchemin, J. L. Barron
Abstract.
Retinal
image motion and optical flow as its approximation are
fundamental concepts in the field of vision, perceptual and
computational. However, the
computation of
REAL
TIME SPEED OF A CONSERVATIVE PARALLEL
N.
Kalantery
Abstract. This paper examines the real time speed of the conservative parallel simulation of a telecommunications network. Real time speed is defined as the ratio of the simulated time to the execution time. A generic simulation model of SS7 networks is executed under the conservative mechanism. A technique used to secure a stable lower bound on the speed of the SS7 simulator is introduced and analyzed. Using this technique simulation models ranging in size, from 16 to 64 nodes, are executed on similarly sized and configured transputer networks. Empirical data on the real time performance of the system are presented. The results confirm that under the given conditions, the performance boundaries of the simulator are both stable and scalable: independent of work load density and size of the modeled network, simulation of a given period of model activity can be guaranteed within a predetermined period of real time.