What is a PCI bus?

Trends in the development of computer technology are obvious - it is unification, simplification for manufacturers (reduction of overhead costs) and the unification of many devices in a single package. There is nothing surprising in that many modern users-beginners do not even know what a PCI bus is. Those who found the time of the formation of IBM-compatible computers based on 286, 386, etc. processors, no doubt remember the abundance of various tires and hardware interfaces used in computer systems of the time. This is the famous 8-bit ISA bus, then its revision VLB (known as VESA), PCI bus, Intel dedicated connector for soft modems, AGP, etc. And, all of them, with rare exceptions, were needed. But at the present time on motherboards the striking "poverty" is one PCI Express bus, however, different revisions and with different number of signal lines. Alas, for such unification to pay, in a literal sense, it is necessary to the end user. After all, even if there is, for example, a high-quality sound card representing a PCI-device, it simply has nowhere to connect (inconvenient adapters do not count) and you have to buy a version for another bus or, in general, go to a frankly flawed, Built-in solution. The situation is reminiscent of the story of the "forcible" transfer of everyone from AGP to PCI-E.

What is so famous for the PCI bus? Its history began back in 1991 when, with the advent of productive Pentium class processors, it became clear that a common 8-bit ISA can no longer provide an acceptable speed of interaction of all components among themselves. Recall that at that time there was neither DMI nor Hyper Transport, and the components on the motherboard were connected via the system bus. Although attempts were made to bypass ISA restrictions, for example, the VESA bus appeared, but due to limitations they did not take root.

In 1992, Intel introduced the PCI bus version 1.0 based on the open standard. Its clock frequency was huge at that time 33 MHz (which simplified the processor frequency scaling), depending on the hardware implementation, 32 or 64 bits were supported (compare with 8 bits of ISA). The voltage of the data lines was 5 and 3.3 V, and the throughput was 133 MB. And, most importantly, it was implemented Plug & Play support (off the jumper!).

In 1993 there appeared an improved version 2.0.

The PCI bus has gained worldwide fame since 1995 (version 2.1). Depending on the implementation, the frequency was 66 MHz. When using a 64-bit version, 533 MB could be transmitted via this bus. The upcoming Windows 95, finally, fully ensured the correct operation of PCI devices with Plug & Play technology. Signal lines allowed to switch from 5V to 3.3V.

Further, revisions 2.2 - 3.0 were issued. The connector in version 2.2 had a "key" blocking the installation of incompatible voltage devices. In 2.3, the transition to 3.3 V continues, so using such devices in versions up to 2.1 (5 V) resulted in damage to both the card and, sometimes, the bus. In version 3.0, the 5 V voltage was no longer used.

In 1997, PCI had to make room, because the same Intel was released AGP bus for video cards that could not fully reveal their potential on PCI.

Now PCI bus, as mentioned earlier, is practically not used, giving way to its successor - PCI Express. The software addressing mechanism remained the same, but the physical implementation has undergone significant changes. The number of conductors has been changed, the operating frequency has been increased (along with the reduction of the current). The way of configuring the initiators of transactions (connected devices) has also changed, allowing more convenient processing of their requests to the bus.

By the way, on some motherboards it is still possible to meet the PCI bus connector. However, the native support from the chipset and the processor is not talking - in this way of implementation a special bridge chip is used that redirects PCI requests to PCI Express.