Systolic block Householder transformation for RLS algorithm with two-level pipelined implementation
نویسندگان
چکیده
منابع مشابه
Systolic block Householder transformation for RLS algorithm with two-level pipelined implementation
The QR decomposition, recursive least squares (QRD RLS) algorithm is one of the most promising RLS algorithms, due to its robust numerical stability and suitability for VLSI implementation based on a systolic array architecture. Up to now, among many techniques to implement the QR decomposition, only the Givens rotation and modified GramSchmidt methods have been successfully applied to the deve...
متن کاملA Pipelined Systolic Arrays Architecture for the Hierarchical Block-Matching Algorithm
This paper presents a pipelined architecture for the hierarchical block-matching motion estimation algorithm (HBMA). The hierarchical style leads enormous computation and complex data flow between hierarchy levels. Each stage of the proposed architecture consists of a systolic array for block-matching and an interpolation unit for bilinear interpolation. The interpolation unit regulates also th...
متن کاملSystematic Design of Two-level Pipelined Systolic Arrays with Data Contraflow
Many systolic algorithms and related design methodologies have been recently proposed. Frecuently, in these systolic algorithms practical considerations are not taken into account. Equitatively distributed load between processing elements, pipelined functional units etc, are desirable features when implementing systolic algorithms.In this paper we present a design methodology in which these fea...
متن کاملImplementation of Systolic Algorithms Using Pipelined Functional Units
A SA with simple synchronization cannot be directly and efficiently executed using PFUs because in every cell each operation is performed only when the previous one has been completed, even if those operations are independent. This fact will not permit to achieve a good utilization of the PFU in every cell unless the original SA is transformed into a two level pipelined SA. This transformation ...
متن کاملAugmented Block Householder Arnoldi Method
Computing the eigenvalues and eigenvectors of a large sparse nonsymmetric matrix arises in many applications and can be a very computationally challenging problem. In this paper we propose the Augmented Block Householder Arnoldi (ABHA) method that combines the advantages of a block routine with an augmented Krylov routine. A public domain MATLAB code ahbeigs has been developed and numerical exp...
متن کاملذخیره در منابع من
با ذخیره ی این منبع در منابع من، دسترسی به آن را برای استفاده های بعدی آسان تر کنید
ژورنال
عنوان ژورنال: IEEE Transactions on Signal Processing
سال: 1992
ISSN: 1053-587X
DOI: 10.1109/78.127965