Pipeline processor

In one application (Maitre, 1981), the parallelism of noncoherent optical convolution methods is combined with video techniques in a totally analog application to images. Matsuoka et al (1982) have developed a hybrid optical-video-digital system based on similar principles. Pipeline processors that perform image convolution at high speed are ideal for digital applications. The inherent simplicity of the method adapts it to high-speed implementations by use of other hybrid techniques. It is even possible to devise a transversal filter that has all the desirable nonlinear properties of the method. We describe this filter in the next section. 

Regarding an implementation within a pipelined processor, we can execute the redundant computation in the following clock cycle (in an extra stage within the arithmetic pipeline) or in the current cycle, if the adder is not in the critical path. The normalizing of the result can be done in the suc-cessing cycle. The throughput of the adder is 1, since in each cycle an addition can be started. 

Fechner, B. Analysis of Checksum-Based Execution Schemes for Pipelined Processors. In Proc. of the 11th Workshop on Dependable Parallel, Distributed and Network-Centric Systems (DPDNS-II), 2006. 

Several types of parallel computing architectures have been used over the years, including pipelined processors, specialized SIMD machines and general MIMD computers. Most parallel computing is done on hybrid supercomputers that combine features from several basic architectures. The best way to define parallel processing in detail is to explain how a program executes on a typical hybrid parallel computer. 

Research about time behavior of workflows started in the middle of the 1970s. Ramchandani introduced timed petri nets [17] which he used to model the time response of asynchronous pipelined processors. Later, this method got adapted... 

The user has to choose a style, pipelined or non-pipelined, for each processor. Different styles call for different synthesis algorithms. In a pipelined processor, a data instance enters the processor before the completion of previous instances. In a non-pipelined processor, on the other hand, a data instance has to wait for the completion of the previous data instance. 

---