Hardware processor unit

Figure 3. Schematic logic diagram of a hardware interface between a Zilog Z80A microprocessor and the Am9511A arithmetic processor unit.

Figure 4. Logic diagram of an Am9511A arithmetic processor unit to S-100 bus hardware interface.

Another option might have been the implementation of a triple modular redundant [4] (TMR, fault masking structural redundancy) system with robust majority voting schemes at the actuators. Such systems can be built either with synchronously running processor units and distributed local code/data memories, or asynchronously running processor units with local access to code and shared access to triple redundant data memories. The complexity and hardware demand of such a system architecture is so high, however, that this idea had to be rejected for technical reasons (mass, volume, harness, power consumption, etc.). 

A simple, distributed (triple redundant) hardware logic decides on the current role (primary vs. secondary) of the two identical (dual hot redundant) processor units (DPUs). Both can access and control any other functional subunits, and a serial bidirectional cross-link is established between the two DPUs to keep them in sync. 

System hardware consists of the central processor, the input devices (usually a keyboard), the output devices (probably both a video display terminal and a hardcopy printer), long-term storage devices, and perhaps communications components. In smaller systems, more than one of these components may be built in to one unit, while in larger systems there may be many units each of several components associated with the system. 

Computer architectures have evolved over the years from the classic von Neumann architecture into a variety of forms. Great benefits to operating speed have accrued. The major contributions to speed have been the introductions of parallel processors and of pipelining. For many years, these innovations were transparent to the programmer. For example, to program in Fortran to run on a CDC 6600 (10, one did not take cognizance of the existence of multiple functional units, nor did one consider the I/O channels when writing Fortran applications for the IBM 360 series (2). This was because the parallel processors were hidden behind appropriate hardware or software. 

Hardware. Computers built to work like neural nets are called "parallel processors". A parallel processor uses a large number of small, interconnected processing units rather than a single CPU. Prominent among these is Thinking Machines Corporation s "Connection Machine", which can realize a variety of neural net models. It is programmable in LISP, and is well-suited to database tasks. Hecht-Nielsen produces the ANZA board, a co-processor which allows the PC/AT to emulate a parallel processor. 

Modern processor architectures exploit the parallelism inside the instruction stream by executing independent instructions concurrently using multiple functional units. This independence relation can be computed from the program in the situation of pure floating-point arithmetic instructions considered in this paper, it can be inferred from the program text, and there is a trade-off between compiler- and hardware-measures to exploit it. As soon as data-dependencies across load-store instructions are considered, data- dependencies can only be computed at run-time. In a later paper, we show how to extend the simple model presented here to also cope with such dynamic dependencies, as well as speculative execution of instructions as resulting from branch-prediction. 

Hardware parallelism can be incorporated within the processors through pipelining and the use of multiple functionalunits, as shown in Fig. 19.2. These methods allow instruction execution to be overlapped in time. Several instructions from the same instruction stream or thread could be executing in each pipeline stage for a pipelined system or in separate functional units for a system with multiple functional units. 

With so-called one-chip processors, modern semiconductor technology offers an inexpensive module which can perform all the functions of the central processor of a process computer. As a result, many new fields of application for computer-based solutions are being opened up, and the end of this development is not in sight. Hence the introduction of microprocessors into the control of cement manufacturing plant obviously suggests itself, and has indeed been successfully accomplished for certain special purposes. It should be borne in mind, however, that with the use of a process computer in a cement works rather less than 10% of the capital cost is spent on the central processor, as against more than 50% on software and engineering. The actual cost of a microprocessor is therefore only to a very limited extent determined merely by the hardware cost. Even so, the microprocessor is assured of a future in cement works, but more particularly as a powerful component unit of process control systems. 

---