Instruction-set architecture exploration of VLIW ASIPs using a genetic algorithm

Abstract

Abstract-Genetic algorithms are commonly used for automatically solving complex design problem because exploration using genetic algorithms can consistently deliver good results when the algorithm is given a long enough run-time. However, the exploration time for problems with huge design spaces can be very long, often making exploration using a genetic algorithm practically infeasible. In this work, we present a genetic algorithm for exploring the instruction-set architecture of VLIW ASIPs and demonstrate its effectiveness by comparing it to two heuristic algorithms. We present several optimizations to the genetic algorithm configuration, and demonstrate how caching of intermediate compilation and simulation results can reduce the exploration time by an order of magnitude.

Parallelism in embedded microprocessor systems based on clustering

“…To single out the clusters and their images, we employ a set-theoretical mode based on the notion of equivalence [4]. Then, Expression (4) serves as a criterion for the evaluation of the partitioning quality.…”

“…The number of equivalence classes being L , and the order of each l th class equaling | | l a , each i th data processing algorithm in the ECS is associated with a set of data defined according to (4). Obviously, this is a generic case, and this is typical of a multiprocessor structure, as it presupposes associations not only between the interclass elements, as in the vector structure, but also among elements of various classes, as in the pipeline structure.…”

“…The emerging picture boils down to the following: performance of conventional systems is bridled by certain physical limits. To resolve the problem of performance growth (i.e., to improve the price/performance ratio), various system architectures have been designed (RISC, CISC, VLIW, EPIC et al) [2][3][4][5].…”

Cluster Parallelism of the Neuroprocessor Systems Based on the Fuzzy Logic

“…(3) To define and to introduce the clusters the authors suggest using the theoretic-set model based on the conception of equivalence [4].…”

“…The vast majority of the memory cells are inactive during the process. So, the resource usage of the hardware is quite low [4]. Besides, the developer is forced to adapt the decision algorithm to the hardware structures available.…”