M.M. Hafizur Rahman

Network Performance of Pruned Hierarchical Torus Network

The complexity of an interconnection network often determines the size of the parallel computer and thus the attainable performance of a parallel computer is limited by the characteristics of the interconnection network. Pruning technique reduces the complexity and hence increases the performance. In this paper, we apply the pruning technique on Hierarchical Torus Network (HTN) and study the architectural details of the pruned HTN. We have explored the network diameter, average distance, bisection width, peak number of vertical links, and VLSI layout area of different HTN. It is shown that the pruned HTN possesses several attractive features including small diameter, small average distance, small number of wires, a particularly small number of vertical links, and economic layout area as compared to its non-pruned counterpart.

Symmetric and Folded Tori Connected Torus Network

Hierarchical interconnection networks provide high performance at low cost by exploring the locality that exists in the communication patterns of massively parallel computers. A Symmetric Tori connected Torus Network (STTN) is a 2D-torus network of multiple basic modules, in which the basic modules are 2D-torus networks that are hierarchically interconnected for higher-level networks. We also fold the STTN to minimize the length of the longest wire of STTN. The folded network is called Folded Tori connected Torus Network (FTTN). In this paper, we present the architecture of the STTN and FTTN, addressing of node, routing of message, and evaluate the static network performance of STTN, FTTN, TTN, TESH, mesh, and torus networks. It is shown that both the STTN and FTTN possess several attractive features, including constant node degree, small diameter, low cost, small average distance, moderate bisection width, and high fault tolerant performance than that of other conventional and hierarchical interconnection networks. We further evaluate the longest wire length of FTTN and compare it with other networks. We found that the longest wire length of the FTTN is far lower than that of other networks considered in this paper while keeping good static network performance. It is just about (1 / 2 m th) of its rival STTN.

Symmetric Tori connected Torus Network

A Symmetric Tori connected Torus Network (STTN) is a 2D-torus network of multiple basic modules, in which the basic modules are 2D-torus networks that are hierarchically interconnected for higher-level networks. In this paper, we present the architecture of the STTN, addressing of node, routing of message, and evaluate the static network performance of STTN, TTN, TESH, mesh, and torus networks. It is shown that the STTN possesses several attractive features, including constant degree, small diameter, low cost, small average distance, moderate bisection width, and high fault tolerant performance than that of other conventional and hierarchical interconnection networks.

Particle Swarm Optimization with partial search to solve Traveling Salesman Problem

Particle Swarm Optimization (PSO) is population based optimization technique on metaphor of social behavior of flocks of birds and/or schools of fishes. For better solution, at every step each particle changes its velocity based on its current velocity with respect to its previous best position and position of the current best particle in the population. PSO has found as an efficient method for solving function optimization problems, and recently it also studied to solve combinatorial problems such as Traveling Salesman Problem (TSP). Existing method introduced the idea of Swap Operator (SO) and Swap Sequence (SS) in PSO to handle TSP. For TSP, each particle represents a complete tour and velocity is measured as a SS consisting with several SOs. A SO indicates two positions in the tour that might be swap. In the existing method, a new tour is considered after applying a complete SS with all its SOs. Whereas, every SO implantation on a particle (i.e., a solution or a tour) gives a new solution and there might be a chance to get a better tour with some of SOs instead of all the SOs. The objective of the study is to achieve better result introducing using such partial search option for solving TSP. The proposed PSO with Partial Search (PSOPS) algorithm is shown to produce optimal solution within a less number of generation than standard PSO, Genetic Algorithm in solving benchmark TSP.

A deadlock-free routing algorithm using minimum number of virtual channels and application mappings for Hierarchical Torus Network

A Hierarchical Torus Network (HTN) is a 2D-torus network of multiple Basic Modules (BM), in which the BM are 3D-torus networks that are hierarchically interconnected for higher level networks. In this paper, we present a deadlock-free dimension-order routing using minimum number of Virtual Channels and evaluate the networks dynamic communication performance under the uniform traffic pattern by computer simulation and compare it with other contemporary conventional and hierarchical networks. We find that the dynamic communication performance of the HTN is better than that of the H3D-mesh, TESH, mesh, and torus networks. We also present the mapping of some primitive applications on the HTN. It is shown that the number of communication steps for various advanced applications mapping on the HTN is lower than that of those networks.

Bangla handwritten numeral recognition using convolutional neural network

Recognition of handwritten numerals has gained much interest in recent years due to its various application potentials. Although Bangla is a major language in Indian subcontinent and is the first language of Bangladesh study regarding Bangla handwritten numeral recognition (BHNR) is very few with respect to other major languages such Roman. The existing BHNR methods uses distinct feature extraction techniques and various classification tools in their recognition schemes. Recently, convolutional neural network (CNN) is found efficient for image classification with its distinct features. It also automatically provides some degree of translation invariance. In this paper, a CNN based BHNR is investigated. The proposed BHNR-CNN normalizes the written numeral images and then employ CNN to classify individual numerals. It does not employ any feature extraction method like other related works. 17000 hand written numerals with different shapes, sizes and variations are used in this study. The proposed method is shown satisfactory recognition accuracy and outperformed other prominent exiting methods.

High and stable performance under adverse traffic patterns of tori-connected torus network

A Tori-connected Torus Network (TTN) is a scalable network constructed from hierarchically-interconnected 2D-torus networks. Static network performance of fixed size TTN and dynamic communication performance under very common traffic patterns have been evaluated in our previous study. A generalized study of static network performance of TTN of any size and dynamic communication performance under adverse traffic patterns are the main focus of this paper. The TTN possesses several attractive features, including constant node degree, small diameter, low cost, small average distance, moderate bisection width, and better fault-tolerant performance than other networks. We compared with hierarchical TESH and MH3DT and conventional mesh and torus networks, and found that under adverse traffic patterns, the TTN yields high throughput and low latency, providing better dynamic communication performance compared to those networks. Also, the adverse traffic patterns have more stable throughput than uniform traffic in the TTN.

On Dynamic Communication Performance of a Hierarchical 3D-Mesh Network

A Hierarchical 3D-Mesh (H3DM) Network is a 2D-mesh network of multiple basic modules (BMs), in which the basic modules are 3D-torus networks that are hierarchically interconnected for higher-level networks. In this paper, we evaluate the dynamic communication performance of a Hierarchical 3D-Mesh (H3DM) network using a deadlock-free routing algorithm with minimum number of virtual channels under the uniform and non-uniform traffic patterns; and compare it with other networks to show the superiority of the H3DM network over other networks. We have also evaluated the dynamic communication performance of the mesh and torus networks. It is shown that H3DM network yields low average transfer time than that of mesh and torus networks. The trade-off between throughput and latency of these networks shown that H3DM network provide better dynamic communication performance than that of mesh and torus networks before saturation.

Routing performance enhancement in hierarchical torus network by link-selection algorithm

A hierarchical torus network (HTN) is a 2D-torus network of multiple basic modules, in which the basic modules are 3D-torus networks that are hierarchically interconnected for higher-level networks. The static network performance of the HTN and its dynamic communication performance using the popular dimension-order routing algorithm have already been evaluated and shown to be superior to the performance of other conventional and hierarchical interconnection networks. In this paper, we propose a link-selection algorithm for efficient use of physical links of the HTN, while keeping the link-selection algorithm as simple as the dimension-order routing algorithm. We also prove that the proposed algorithm for the HTN is deadlock-free using three virtual channels. We evaluate the dynamic communication performance of an HTN using dimension-order routing and link-selection algorithms under various traffic patterns. We find that the dynamic communication performance of an HTN using the link-selection algorithm is better than when the dimension-order routing algorithm is used.

Convolutional neural network training with artificial pattern for Bangla handwritten numeral recognition

Recognition of handwritten numerals has gained much interest in recent years due to its various application potentials. The progress of handwritten Bangla numeral is well behind Roman, Chinese and Arabic scripts although it is a major language in Indian subcontinent and is the first language of Bangladesh. Handwritten numeral classification is a high-dimensional complex task and existing methods use distinct feature extraction techniques and various classification tools in their recognition schemes. Recently, convolutional neural network (CNN) is found efficient for image classification with its distinct features. In this study, a CNN based method has been investigated for Bangla handwritten numeral recognition. A moderated pre-processing has been adopted to produce patterns from handwritten scan images. On the other hand, CNN has been trained with the patterns plus a number of artificial patterns. A simple rotation based approach is employed to generate artificial patterns. The proposed CNN with artificial pattern is shown to outperform other existing methods while tested on a popular Bangla benchmark handwritten dataset.