Elimination of the computational broadcast in systolic arrays: an application to the qr decomposition algorithm
Computational broadcast elimination is defined in terms of the system of affine recurrence equations. It is shown that it can be eliminated by introducing a new recurrence equation in the single assignment code. Several examples are given to demonstrate the proposed procedure. The QR decomposition algorithm, where computational broadcast occurs for two variables, is localized and implemented on a processor array by using the data dependence method. It is shown that, in some cases, the algorithm can be realized by using only half the processors at the cost of doubling the complexity of processors used.
- A Rotation Method for Computing the QR-Decomposition
- Algorithm transformations for the data broadcast elimination method
- Localized algorithms for vlsi processor arrays
- A systolic array architecture for QR decomposition of block structured sparse systems
- Systolic computation of QZ matrix decomposition
- A triangular processor array for computing singular values
- Algorithm transformations for computational and data broadcast
- Double pipelines and fast systolic designs on linear arrays
- scientific article; zbMATH DE number 3628373 (Why is no real title available?)
- scientific article; zbMATH DE number 3892457 (Why is no real title available?)
- On the Analysis and Synthesis of VLSI Algorithms
- Synthesizing systolic arrays from recurrence equations
- The Organization of Computations for Uniform Recurrence Equations
- VLSI processor array IPS cells
This page was built for publication: Elimination of the computational broadcast in systolic arrays: an application to the qr decomposition algorithm
Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4489982)