Systematic design approaches for algorithmically specified systolic arrays
Computer architecture
A new matrix vector product systolic array
Journal of Parallel and Distributed Computing
Designing of processor-time optimal systolic arrays for band matrix-vector multiplication
Computers & Mathematics with Applications
Parallel and systolic solution of normalized explicit approximate inverse preconditioning
The Journal of Supercomputing - Special issue: Parallel and distributed processing and applications
The Journal of Supercomputing - Special issue: Parallel and distributed processing and applications
The Journal of Supercomputing
Hi-index | 0.00 |
One of the most challenging problems in systolic processing is the development of a methodology for mapping an algorithm into a systolic architecture. It becomes more and more obvious that systolic arrays can be produced not only using trial and error approaches, but also that they can be mechanically produced through the use of optimal synthetic methods. Herein, a new approach is investigated regarding the synthesis of processor-time optimal systolic arrays. The proposed procedure is based on the dependence method, but nonlinear transformations and the concept of double processing streams for the input data of the algorithm are also used. The result is efficient systolic designs from both aspects, the hardware used and the overall time complexity achieved.