Table 1 - uploaded by Mats Brorsson
Content may be subject to copyright.
Source publication
Software Distributed Shared Memory (S-DSM) systems support parallel programming by implementing a shared memory on top of distributed system. It frees programmer from communi- cation complexities to concentrate to parallel algorithms. How- ever, there is a drawback: current S-DSM systems waist memory on all processors. Therefore it is hard to be im...
Contexts in source publication
Context 1
... The number of pages which the system pre-allocates to use each time: NUM_PAGES • The maximum number of processors the system can use: MAX_PROCS • The maximum number of locks to be used: MAX_LOCKS We used two different sets of parameter to investigate the memory usage according to Table 1. The first set takes common parameters using in standard CVM system. ...
Context 2
... time spent on each part of each proc- essor in one iteration. Processor time (ms) 0 1 2 3 500, black 2782 2841 2838 2829 500, red 2749 3398 2781 3223 504, black 2815 2828 2846 2797 504, red 2772 2783 2802 2786 By using a number of rows divisible by 8, e.g. m=504, this local load imbalance was isolated and we thus got a speedup of 3.9 on 4 processors instead of 3.5 (see Figure 5). ...
Similar publications
Addition is an essential operation in all cryptographic algorithms. Higher levels of security require larger key sizes and this becomes a limiting factor in GF(p) using large integers because of the carry propagation problem. We propose a novel and efficient attribute-based large integer representation scheme suitable for large integers commonly us...