Peer-to-peer (P2P) networking is a distributed architecture that partitions tasks or data between peer nodes. In this paper, an efficient Hypercube Sequential Matrix Partition (HS-MP) for efficient data sharing in P2P...Peer-to-peer (P2P) networking is a distributed architecture that partitions tasks or data between peer nodes. In this paper, an efficient Hypercube Sequential Matrix Partition (HS-MP) for efficient data sharing in P2P Networks using tokenizer method is proposed to resolve the problems of the larger P2P networks. The availability of data is first measured by the tokenizer using Dynamic Hypercube Organization. By applying Dynamic Hypercube Organization, that efficiently coordinates and assists the peers in P2P network ensuring data availability at many locations. Each data in peer is then assigned with valid ID by the tokenizer using Sequential Self-Organizing (SSO) ID generation model. This ensures data sharing with other nodes in large P2P network at minimum time interval which is obtained through proximity of data availability. To validate the framework HS-MP, the performance is evaluated using traffic traces collected from data sharing applications. Simulations conducting using Network simulator-2 show that the proposed framework outperforms the conventional streaming models. The performance of the proposed system is analyzed using energy consumption, average latency and average data availability rate with respect to the number of peer nodes, data size, amount of data shared and execution time. The proposed method reduces the energy consumption 43.35% to transpose traffic, 35.29% to bitrev traffic and 25% to bitcomp traffic patterns.展开更多
The new generation of computing devices tends to support multiple floating-point formats and different computing precision.Besides single and double precision,half precision is embraced and widely supported by new com...The new generation of computing devices tends to support multiple floating-point formats and different computing precision.Besides single and double precision,half precision is embraced and widely supported by new computing devices.Lowprecision representations have compact memory size and lightweight computing strength,and they also bring opportunities to the optimization of BLAS routines.This paper proposes a new sparse matrix partition approach based on IEEE 754 standard floating-point format.An input sparse matrix in double precision is partitioned and transformed into several sub-matrices in different precision without loss of accuracy.Most non-zero elements can be stored in half or single precision,if the most significant bits of exponent and the least significant bits of mantissa are zeros in double-precision representation.Based on this mixed-precision representation of sparse matrix,we also present a new SpMV algorithm pSpMV for GPU devices.pSpMV not only reduces the memory access overhead,but also reduces the computing strength of floating-point numbers.Experimental results on two GPU devices show that pSpMV achieves a geometric mean speedup of 1.39x on Tesla V100 and 1.45x on Tesla P100 over double-precision SpMV for 2,554 sparse matrices.展开更多
Some rank equalities are established for anti-involutory matrices. In particular, we get the formulas for the rank of the difference, the sum and the commutator of anti-involutory matrices.
文摘Peer-to-peer (P2P) networking is a distributed architecture that partitions tasks or data between peer nodes. In this paper, an efficient Hypercube Sequential Matrix Partition (HS-MP) for efficient data sharing in P2P Networks using tokenizer method is proposed to resolve the problems of the larger P2P networks. The availability of data is first measured by the tokenizer using Dynamic Hypercube Organization. By applying Dynamic Hypercube Organization, that efficiently coordinates and assists the peers in P2P network ensuring data availability at many locations. Each data in peer is then assigned with valid ID by the tokenizer using Sequential Self-Organizing (SSO) ID generation model. This ensures data sharing with other nodes in large P2P network at minimum time interval which is obtained through proximity of data availability. To validate the framework HS-MP, the performance is evaluated using traffic traces collected from data sharing applications. Simulations conducting using Network simulator-2 show that the proposed framework outperforms the conventional streaming models. The performance of the proposed system is analyzed using energy consumption, average latency and average data availability rate with respect to the number of peer nodes, data size, amount of data shared and execution time. The proposed method reduces the energy consumption 43.35% to transpose traffic, 35.29% to bitrev traffic and 25% to bitcomp traffic patterns.
文摘The new generation of computing devices tends to support multiple floating-point formats and different computing precision.Besides single and double precision,half precision is embraced and widely supported by new computing devices.Lowprecision representations have compact memory size and lightweight computing strength,and they also bring opportunities to the optimization of BLAS routines.This paper proposes a new sparse matrix partition approach based on IEEE 754 standard floating-point format.An input sparse matrix in double precision is partitioned and transformed into several sub-matrices in different precision without loss of accuracy.Most non-zero elements can be stored in half or single precision,if the most significant bits of exponent and the least significant bits of mantissa are zeros in double-precision representation.Based on this mixed-precision representation of sparse matrix,we also present a new SpMV algorithm pSpMV for GPU devices.pSpMV not only reduces the memory access overhead,but also reduces the computing strength of floating-point numbers.Experimental results on two GPU devices show that pSpMV achieves a geometric mean speedup of 1.39x on Tesla V100 and 1.45x on Tesla P100 over double-precision SpMV for 2,554 sparse matrices.
文摘Some rank equalities are established for anti-involutory matrices. In particular, we get the formulas for the rank of the difference, the sum and the commutator of anti-involutory matrices.
