All "processors PC" use the same instructions as first 8086 CPU of INTEL (but often the operating system is not compatible any more with these old programs). The instructions which a processor is able to read are programmed out of assembler language. All

Internal structure of a basic processor

Let us take again our Z80 and distinguish the various parts: with the top, an internal bus of data, below an internal bus order and a bus interns of address. Knowing the diagram of the microprocessor-based systems, this is not astonishing. In the part of right-hand side, an UAL (Arithmetic Logic Unit). This UAL is a calculating unit, it is connected to a noted buffer memory the accumulator and through bus to a whole of memory called registers. In our case, these registers are noted B,C, D... The registers A', B', It... is specific to Z80, but also exists in the 8088. This makes it possible to work with two zones memories which one makes rock. The UAL of left makes it possible to make calculations on addresses. Notice the small incrementor +1 who allows to increase the pointer programs (the address where the next instruction of the program is).



The diagram of the processors of this generation showed the following diagram:



This structure charges an instruction code at the same time, that is to say for an instruction of the type LDA XX XX (loading in the accumulator of the data contained with the address memory XX XX), three cycles of instructions (for LDA and 2 for the address). This condition moreover is met only if the processor charges an instruction by cycle of clock, but it is the case in a system INTEL. As the instructions of the processor current are compatible with the 8088, the sizes of the registers are it too. It is thus impossible to increase the size of these memories to increase the speed of execution.

Which are the possibilities of increasing the speed of a processor without modifying the structure?

1. to carry out several instructions at the same time, but that is difficult since certain instructions are conditional.
2. to make pass via one to decode the instructions of the type CISC in faster RISC.
3. To make sure that the processor never awaits instructions, that they are directly accessible.
4. to increase integration by decreasing the size of the transistors. This increases the speed of transfer between the various parts.

All these tricks are currently used, as well by INTEL, as by AMD and Cyrix (VIA).