Design of Efficient Regular Arrays for Matrix Multiplication by Two-Step Regularization
نویسندگان
چکیده
215 given k and n, one can choose any value of nz in the range from 2 to n to construct a multiway merging network. However, use of certain values of m results in less delay than others for the same value of n. One choice of the value of m is m = n. In this case, can achieve this time bound for the given value of n, are m = 2' with c < [log, n1. We have also assumed that there are n keys in each ascending sequence and m divides n for simplicity and clarity of exposition. In general, a multiway merging network can be constructed to merge k sorted lists of different lengths with the ith list having n, keys not necessarily equal to n. One way of doing this is to construct a k-way merger of size n first using keys with the value of positive infinity to fill the difference, and then remove these positive infinity keys together with the comparison-exchange elements associated with them, resulting in a simpler k-way merging network. Thus, there are no restrictions on the property of n,, k, and m. A sorter can be constructed from k-way mergers: the keys are combined k at a time to form ordered lists of length k; these lists are merged k at a time to form ordered lists of length k 2 , etc., until all keys are merged into one ordered list. To sort k p keys using the mergers requires kp-' mergers of size 1 followed by ICp-' mergers of size k followed by kp-3 mergers of size k2 followed by kp-4 mergers of size k 3 , etc., etc. The longest path will go through steps. V. CONCLUSION The multiway merge described in this paper merges k ascending sequences into an ascending sequence for any integer k. This differs from existing merging networks that merge only two ascending sequences into one. Furthermore, the k-way merge represents a complete generalization of the odd-even merge, when k = 2. In this case, it uses m small two-way mergers to merge two ascending sequences into one, where m is not restricted to 2. The odd-even merge is a special case of the k-way merge, when k = 2 and m = 2. The multiway merges does not reduce the number of comparators nor does it reduce the delay over the odd-even …
منابع مشابه
A New Parallel Matrix Multiplication Method Adapted on Fibonacci Hypercube Structure
The objective of this study was to develop a new optimal parallel algorithm for matrix multiplication which could run on a Fibonacci Hypercube structure. Most of the popular algorithms for parallel matrix multiplication can not run on Fibonacci Hypercube structure, therefore giving a method that can be run on all structures especially Fibonacci Hypercube structure is necessary for parallel matr...
متن کاملFabrication of polymeric microneedle arrays containing Amphotericin-B for transdermal drug delivery
Background and Aim: Drug delivery through the microneedle array has been considered as an easy and non-invasive method in recent years. The purpose of this study was to design and construct an array of biodegradable polymeric microneedles containing Amphotericin-B to introduce this system and its use in the treatment of cutaneous lesions caused by Leishmania major parasite inoculation as a mode...
متن کاملDesign of Highly Uniform Platinum and Palladium Nanoparticle Decoration on TiO2 Nanotube Arrays: An Efficient Anode to the Electro-Oxidation of Alcohols
We explore electro-catalytic properties of a system consisting of platinum and palladium nanoparticles dispersed over a nanotubular self-organized TiO2 matrix. These electrodes prepared by electroess and microemulsion of palladium and palladium nanoparticles on to TiO2 nanotubes, respectively. Titanium oxide nanotubes were fabricated by anodizing titanium foil in ethylene glycol (EG) fluoride-c...
متن کاملTwo Cycle-Related Problems of Regular Data Flow Graphs: Complexity and Heuristics
1 Abstract1 A regular data flow graph (RDFG) is a graph with a highly regular structure that enables its description to be exponentially smaller than the description size for an ordinary graph. Such graphs arise when certain regular iterative algorithms (like matrix multiplication or convolution) are modeled using dependence graphs. These graphs can be implemented either on systolic arrays, or ...
متن کاملA New Approach for Representation of Multi-dimensional Matrix Multiplication Operations
In this paper, the extended Karnaugh Map representation (EKMR) scheme has been proposed as an alternative to the traditional matrix representation (TMR) which caused the multi-dimensional array operation to be inefficient when extended to dimensions higher than two. Multi-dimensional arrays are widely used in a lot of scientific studies but still some issues have been encountered regarding effi...
متن کاملAn Efficient LUT Design on FPGA for Memory-Based Multiplication
An efficient Lookup Table (LUT) design for memory-based multiplier is proposed. This multiplier can be preferred in DSP computation where one of the inputs, which is filter coefficient to the multiplier, is fixed. In this design, all possible product terms of input multiplicand with the fixed coefficient are stored directly in memory. In contrast to an earlier proposition Odd Multiple Storage ...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
عنوان ژورنال:
- IEEE Trans. Parallel Distrib. Syst.
دوره 6 شماره
صفحات -
تاریخ انتشار 1995