scholarly journals Software Managed Distributed Memories in MPPAs

Author(s):  
Robin Panda ◽  
Jimmy Xu ◽  
Scott Hauck
Keyword(s):  
2016 ◽  
pp. 261-280 ◽  
Author(s):  
Bjorn Nansen ◽  
Michael Arnold ◽  
Martin Gibbs ◽  
Tamara Kohn ◽  
James Meese

1997 ◽  
Vol 07 (04) ◽  
pp. 425-436 ◽  
Author(s):  
Yunheung Paek ◽  
David A. Padua

Due to the complexity of programming scalable multiprocessors with physically distributed memories, it is onerous to manually generate parallel code for these machines. As a consequense, there has been much research on the development of compiler techniques to simplify programming, to increase reliability, and to reduce development costs. For code generation, a compiler applies a number of transformations in areas such as data privatization, data copying and replication, synchronization, and data and work distribution. In this paper, we discuss our recent work on the development and implementation of a few compiler techniques for some of these transformations. We use Polaris, a parallelizing Fortran restructurer developed at Illinois, as the infrastructure to implement our algorithms. The paper includes experimental results obtained by applying our techniques to several benchmark codes.


2012 ◽  
Vol 58 (4) ◽  
pp. 403-410 ◽  
Author(s):  
Arkadiusz Bukowiec ◽  
Marian Adamski

Abstract In this paper a new method of Petri net array-based synthesis is proposed. The method is based on decomposition of colored interpreted macro Petri net into state machine subnets. Each state machine subnet is determined by one color. During the decomposition process macroplaces are expanded or replaced by doublers of macroplace. Such decomposition leads to parallel implementation of a digital system. The structured encoding of places is done by using minimal numbers of bits. Colored microoperations, which are assigned to places, are written into distributed and flexible memories. It leads to realization of a logic circuit in a two-level concurrent structure, where the combinational circuit of the first level is responsible for firing transitions, and the second level memories are used for generation of microoperations. Such an approach allows balanced usage of different kinds of resources available in modern FPGAs


2001 ◽  
Vol 11 (01) ◽  
pp. 155-167 ◽  
Author(s):  
EDUARDO MIZRAJI ◽  
JUAN LIN

Modular neural networks structured as associative memories are capable of processing inputs built from tensorial products of vectors. In this context, the operators of propositional and modal logic can be represented as modular distributed memories that can process not only classical Boolean but also fuzzy evaluations of truth-values of sentences. Furthermore, projecting memory outputs onto unit vectors yield discrete dynamical systems that exhibit varying degrees of complexity. As examples, we analyze outcomes of semantic evaluations in several self-referential systems including modal versions of the chaotic liar, antagonistic decisions and extended dilemmas. By studying these examples we hope to shed some light on the modeling of cognitive decisions.


Author(s):  
Jeong Joon Park ◽  
Ronnel Boettcher ◽  
Andrew Zhao ◽  
Alex Mun ◽  
Kevin Yuh ◽  
...  
Keyword(s):  

Sign in / Sign up

Export Citation Format

Share Document