A new network topology with multiple meshes

Das, Debraj; De, Mallika; Sinha, Bimal K.

doi:10.1109/12.769436

Cited by 37 publications

(15 citation statements)

References 10 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The algorithm due to Tsai et al [33] takes O(log N+2(log 3 NÂlog m)&3 log 2 N+log N log m) time to sort N elements on a mesh-connected computer with multiple broadcasting (MCCMB) using Nm processors, where m is the number of processors in each dimension of the MCCMB. A sorting algorithm on the Multi-Mesh (MM) network has been proposed by De et al [12] which sorts N data elements on a MM [10,11] SIMD model in 58N…”

Section: Introductionmentioning

confidence: 99%

Parallel Sorting Algorithm Using Multiway Merge and Its Implementation on a Multi-Mesh Network

Sinha

Mukherjee

2000

Journal of Parallel and Distributed Computing

View full text Add to dashboard Cite

Section: Introductionmentioning

confidence: 99%

Parallel Sorting Algorithm Using Multiway Merge and Its Implementation on a Multi-Mesh Network

Sinha

Mukherjee

2000

Journal of Parallel and Distributed Computing

View full text Add to dashboard Cite

“…Similarly, it is also true for even β. Thus the 4-neighbor property of the mesh interconnection can be emulated [3] by the multi-mesh network in constant time. The initial distribution of data to preserve desired adjacency among different data elements on an MM network with n = 3 is shown in Figure 2.…”

Section: Emulation Of An N 2 X N 2 Mesh To a Multi-mesh Networkmentioning

confidence: 99%

Faster numerical weather forecasting using parallel computing with multi-mesh topology

De¹

2017

EESRJ

View full text Add to dashboard Cite

Today for different Meteorological observations attempts are being made for a faster parallel computing method for the purpose of numerical weather forecasting. Highly faster computation technique is required to study raw observation data into easily understandable and accurate weather forecast today. Faster technique will make it possible for meteorological data investigations successfully as well as to run complex weather forecasting models in lesser time, using highly complex set of weather models. The methodology that is used by meteorological study of weather forecasting consists usually of technique called numerical weather prediction. Numerical modeling of atmosphere is a complex process of solving a number of equations involving the future changes of weather variables, like temperature, pressure, humidity, wind speed etc. A simultaneous calculation of different parameters is a necessary part for accurate weather prediction. The multi-mesh topology is getting popular for its efficient topological properties for example: existence of Hamiltonian cycle, simple routing etc. In an n  n multi-mesh network used n 2 meshes of size n  n each, which themselves are again arranged in n rows and n columns so that there will be n 4 processors in total. Each n  n mesh in this network is termed as a block. Using this topology different numerical equation for weather prediction can be solved very faster. This paper presents an efficient and faster way for solving different numerical equation using multi-mesh topology with multiple processors acting simultaneously.

show abstract

“…As examples of real-life applications, simple problems like those of calculating the sum, average, minimum and maximum of n 4 data values have been implemented in O(n) time on the MM network having n 4 processors [3]. Numeric problems like matrix multiplication and Lagrange's interpolation have been implemented in MM having n 4 processors in O(n) and O(p 0.6 ) time for n 2 point Lagrange's interpolation and p × p matrix multiplication, respectively, where p = n 5/3 .…”

Section: The MM Networkmentioning

confidence: 99%

“…In a MM network [3], [4], [11], [12] shown in Fig. 1, there are n 2 meshes of size n × n each, which themselves are again arranged in n rows and n columns so that there will be n 4 processors in the network.…”

Section: The MM Networkmentioning

confidence: 99%

“…The parallel computation of the linear transformation A × B = C takes O(n) time for implementing in an n × n 2D mesh [2]. The authors Somen De et al in the present paper initially propose a parallel implementation of the same using multi-mesh (MM) architecture [3], [4] having n 4 processors, all having degree 4. The processors in the MM network having n 2 meshes of size n × n are arranged in rows and columns of meshes of size n × n. The time complexity of the algorithm is O(n) for n 2 points, thereby reducing the time complexity by an O(n).…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation