"Go" Understanding physical I/O Components-host I/O bus

After the data leaves the system memory bus, it is usually transmitted to another bus, the host I/O bus. In today's product, the most common host I/O bus is the PCI bus, but there are several other buses, such as S-Bus, EIS a bus and VME bus. The host I/O bus implements several important functions, including:    allowing new cards to be added.     allows data to be entered and output from the memory bus.     allows data to be transferred between cards.      Host I/O bus is not the only intermediary between the device and the system memory bus, there is also a bridge controller chip between the host I/O bus and the system memory bus, which is responsible for exchanging data between the two buses. Figure 2-3 shows the host I/O bus with 4 I/O cards, all of which are connected to the system memory bus via a bridge controller chip.        Host I/O bus is a transport tool that transmits data between memory and peripherals. Many different peripherals are available in the industry, including video displays, sound cards, printers, scanners, and storage devices. If the system C P U is compared to the engine of a sports car, then the host I/O bus is similar to a sports car transmission device, the engine to the driving force of the wheel, and drive the wheel forward. Many data processing applications require large amounts of information to be transferred between storage devices. If a system fails to effectively transmit data between the host I/O bus and the memory bus, the system cannot provide the appropriate performance when the processing burden increases.      Sometimes, the number of expansion slots is limited, in order to have more expansion slots available, you can add the bridge controller, and 2-5 as the additional host I/O bus to join together.    whenever a business press conference is held, as long as there is new information on the host I/O bus technology, there are plenty of reasons for concern. Because of the changes in host I/O bus technology, many of the existing technologies associated with it will be at stake, including systems companies, adapter card manufacturers, CPU companies, software companies, chip manufacturers, and many important I-t agencies. In the P-C industry, users have to differentiate between the pros and cons of Isa, micro-path, EISA, PCI bus, and often mistakenly spend money on outdated server I/O buses. The following is a brief overview of some I/O bus technologies. 1.pci     Now most server systems have a memory bus design that provides efficient data transfer on the host I/O bus. Although there are several host I/O bus technologies that exist on a wide variety of systems, the interconnection of peripheral components is alwaysThe superior cost and performance advantages of the line PCI bus make it an industry standard. The number of PCI cards used by PC is relatively large, and the production is very large, resulting in a lower overall price of P C I components than other server bus technologies. As a result, most systems companies have adopted PCI technology for competitive pricing. The performance of PCI has proven to exceed the needs of many different purposes, including storage and I/O intensive applications.     A few years ago, Intel realized that if they were to remain competitive, they would have to provide better I/O capabilities in the x86 family of processors. However, on the one hand, to maintain compatibility with the existing PC bus technology, on the other hand, but also to provide higher I/O capability in the CPU, which presents a serious challenge to Intel company.     Intel starts with the 486-chip series and provides a high-speed I/O bus by integrating PCI bridge technology into the cpu/memory chipset. Other bus technologies, such as ISA, micro-path, EISA, etc., can be connected to the host memory bus via the PCI bridge controller. After the introduction of the new technology, Intel transferred the subsequent development of PCI to a standard agency, the PCI Professional Group (PCI-SIG). Subsequently, many other companies have accepted PCI as a standard and contributed to its development. In fact, PCI has been widely accepted by companies such as I Bmsun and Compaq, although these companies have their own I/O bus technology.     on PC systems, the performance of the PCI bus is difficult to predict because PC systems are often a mix of PCI and non-PCI components. Although the system performance of hybrid configurations is undoubtedly greatly affected, it is difficult to estimate the extent to which system performance using hybrid bus technology has been negatively impacted. Therefore, if you are not using a PCI card that is not suitable, do not use ISA card in a PCI system. Note For optimal performance, avoid using the ISA adapter in your server.  2.c-PCI    C-PCI, the abbreviation for dense PCI, is a derivative of the industry standard PCI bus, which is designed to serve the industrial and communications applications. Computers in these areas have high requirements for fault tolerance and environmental resilience because they are typically placed in small or unmanned locations, and C-PCI supports hot swapping of adapters, one of the salient features of which is that there is no motherboard in the system, and all system components are on a service-capable card.     In order to manage this particular hardware, C-PCI uses a real-time operating system, because the real-time operating system provides the basis for fault tolerance for these C-PCI cards, which include the bridge controller card, the CPU card (Lynx RealOne example of this system is the Lynx operating system produced by the TEMs company.  3.vesa Local bus     VESA Local bus is the immediate predecessor of PCI products on the market, it is developed by industry Standard Group, it can provide much faster speed than PC-based I/O bus, and some advantage performance comes from putting the interface directly in the processor. This will make the data transfer particularly fast. This is what "local area Bus" refers to: VESA is electronically connected to the CPU, and with the CPU to enjoy the common power requirements, resulting in the number of cards to support the serious limit, usually not more than two. This means that most server systems must mix older bus technologies with VESA, which is unacceptable in highly throughput-demanding systems. In short, VESA Local bus provides some advantages on the one hand, but there are several engineering obstacles on the other, making it difficult to compete with PCI.  4. One of the most important factors in the wide application of the isa,e I S A and micro path      PC is the high availability of the system and the low cost of peripherals, and the prosperity of the P-C machine will be impossible if I/O bus technology is not well interoperable. The following sections discuss the early development of P-C bus technology  1) ISA bus      The original PC bus is not called the ISA industry standard architecture, but after several other competitive buses appear, in order to make it different from these buses, To name it Isa. ISA is designed to transfer data between a relatively slow device and a slow processor, so it is very slow. But it has a fairly complete basic function. When the IBM Pc-at microcomputer appears, it has been upgraded to a 1 6-bit data bus, with the largest clock frequency reaching 8MHz. As processor speed continues to increase and the number of PC applications continues to grow, new I/O bus architectures need to be developed to keep pace with processor development.  2) Micro Path      1987, IBM announces its micro-path bus architecture, which combines bus control cards and higher clock frequencies. The micro Path introduces a new physical connector that makes it incompatible with existing ISA computers and adapters, and even after changing the design of its connector, the micro path must also maintain compatibility with the adapter with a slower clock speed, so not only is the micro-path and ISA physically incompatible, And its performance is limited by the existing ISA design. As a result, few other computer vendors have used the technology in their systems, and the development of micro-path adapters is far less than the level required by the PC industry. With the passage of time, the success of this bus technology in a few high-end UNIX systemsSystem, and these UNIX systems are almost all from IBM. Today's micro-path technology is actually an obsolete technology. Interestingly, though, the incompatible connector design of the micro-path technology doomed its failure in the PC market in the late 80, but it was later used as a PCI bus Connector specification and was a great success.  3) eisa    EISA represents the enhanced Isa, developed primarily by Compaq, but has been implemented in several PC companies to replace IBM's micro-path technology. The EISA allows the installation of either ISA card or E ISA card in the same system, both of which have the same bus structure. Because EISA technical requirements are compatible with the ISA card, its performance is severely limited. At the same time, the micro-path is compatible with the ISA design rather than the card, so it cannot be shown to be superior to EISA. But because the EISA price is cheaper and more widely available, it quickly occupies the market.     Because the PCI bus architecture not only provides stronger ISA backwards compatibility, but also has higher performance, it makes EISA gradually disappear from the market.  5.s-Bus
S-Bus is an I/O bus that is popular with sun products and provides better performance than other PC-based buses, including early implementations of PCI. Like PCI, Sun also opened S-bus technology to industry, followed by the formation of standard groups to take over the development of S-bus specifications. But the S-bus definitely does not really meet the scale required by the fact standard, and at this point it is different from PCI. PCI performance eventually catches up with the S-Bus, and for Sun, PCI is cheaper to implement.


6.VME
The VME bus has been used for MIPS-based UNIX workstations and Dec company's VAX system. VME has also become an industry standard and widely used in industrial computing applications, and is now being replaced by PCI and C-PCI in industrial computing applications.

7.I/O Bus Comparison

A simple comparison of the bus techniques discussed above is given in the table. The PCI bus occupies three columns, each of which represents: version 2.0, which is the major version installed on the current machine, and version 2.1, which is a widely implemented version of 1999; C-PCI version. 8. Future I/O bus

As processor performance continues to increase, the I/O bus speed that is provided to the processor data has to be increased. This has happened in the past 20 years, and it may still be the case in the foreseeable future. In the fall of 1999, a new I/O bus specification standard, SYSTEMI/O (SIO), was developed by an industrial cooperation Organization with a large number of members.

Sio is a technology "consolidator" that combines two competitive I/O bus prototypes. One of them is called FUTUREI/O (FIO), presented by IBM, Compaq, and HP, and another I/O bus called NEXTGENERATIONI/O (Ngio), which is mainly presented by Intel, and Sun has given some support.

The same type of technology proposed in FIO and Ngio is expected to be achieved in 1999, although the details of Sio are not yet established. The results are likely to have a far-reaching impact on storage network technology, as both FIO and Ngio are designed for serial I/O transmissions. It can be expected that, without the host I/O adapter, the SIO can provide connectivity to the storage device and subsystem to the cable, as well as the ability to directly connect the cable to the system I/O bus.

"Go" Understanding physical I/O Components-host I/O bus