28. 07.2017

Blog of Dmitry Vassiyarov.

SCSI is a fast and unusual interface

Hello.

From this article, you will learn the essentials about SCSI, what it is, where and why it is used, how many generations have come out since its inception, and how it is implemented in practice.

Read - suddenly, SCSI will come in handy for you too?

What does SCSI stand for?

This is a set of capital letters from the phrase Small Computer Systems Interface. In Russian, it sounds like "tell", and the decryption is a system interface for small computers.

This standard was created to combine computer components for various purposes on one bus: hard drives, disk drives, scanners, printers, etc. Why? To provide them with the same high speed of work as a single, but at the same time divisible mechanism. In addition, thanks to SCSI, you can use one device on several computers at once.

Other features

In addition to simply connecting iron, the technology allows you to exchange data and defines a set of commands that has become widespread. For example, in Windows it is used in a single stack for storage devices.

The most commonly used commands are writing, reading, checking devices, querying their characteristics, setting new parameters for them or returning previous ones, etc.

There is also the implementation of commands over wires and controllers of other standards. If we are talking about IDE, ATA or SATA, it is called ATAPI - ATA Packet Interface; if on top of the USB protocol - Mass Storage device. Thus, you can, for example, connect an external HDD via regular USB and the SCSI driver available in the operating system will be used for it.

Where is SCSI in demand?

On high performance servers and workstations. On servers belonging to the low price category, and even more so at home, this interface is extremely rare; in such cases, the best option is familiar to us.

But of course, no one forbids you to put such devices in your home computer. Or, for example, to a home server.

Technology in practice

All devices that you want to connect to the same bus work through a special adapter, which, in turn, is inserted into a free slot on the motherboard. The controller has its own BIOS, through which you can control devices. The operating system recognizes and communicates with them as usual with .

The presence of a SCSI adapter means that with CPU part of the load is removed, therefore, the iron works faster.

Since this technology is serial, the devices should be connected accordingly. Moreover, each must have a unique ID, and they all have the same interface.

History of appearance

I want to tell you the story of the creation of the interface not out of my tediousness, but because through it you will be able to understand more about the subject of our conversation.

So, in 1979, the inventor of 8-inch floppy disks and the manufacturer of magnetic drives, Alan Shugart, set himself the task of making a universal interface for his products that would not lose its position, taking into account the development of technology.

And he managed to solve it by creating a standard that supports logical and practical (head, cylinder, sector) addressing. It was based on protocols for 8-bit parallel sending of information along a path that included several lines.

The innovation received the name SASI (Shugart Associates Systems Interface), which is not very harmonious for the Russian-speaking population, that is, a connecting system interface named after the founding father.

After 2 years, he shared his development with the ANSI committee (American National Standards Institute - US National Standards Institute) - the same as GOST in our country. Based on this invention, ANSI specialists created SCSI.

Interface Generations

It is noteworthy that the technology was created almost half a century ago, and we are still talking about it. All because she was constantly changing. Since its inception, 10 versions have been released. I won't bore you with details about each of them. I will tell only what was originally, and what we have now.

SCSI-1

• It is possible to connect a maximum of 8 devices to one bus, including the controller.
• The maximum speed was 1.5 Mb / s in the asynchronous variation (“request-acknowledgment”), and 5 Mb / s in the synchronous - the same number of confirmations were returned for several requests.
• On the electrical side, there were 24 lines, including differential and unipolar, although signals of the second type were more often applied.
• The bus frequency was 5 MHz.
• The longest cable is 6m and for HVD differential bus is 25m.

Ultra-640SCSI

• The bus width has doubled, respectively, you can connect up to 16 devices at the same time.
• Its frequency is 160 MHz DDR.
• The speed also cannot be compared with the first modification - now it reaches 640 Mb / s.
• The connector consists of 68 pins.
• The length of the cable reaches 10 m.

Serial Attached SCSI (SAS)

• Added support for connecting SATA devices.
• The speed of this interface has already increased to 12.0 Gb / s.
• According to the developers, it is now possible to connect 16384 devices to one bus! In the previous generation, as described above, there were only 16.

Electrician

There are 3 ways to communicate information regarding electrical:

• SE (single-ended) - asymmetrical look. Each signal is sent on a separate line.
• LVD (low-voltage-differential) is a low voltage differential standard. The "+" and "-" signals are sent on separate wires. Each of them is assigned one twisted pair. They are transmitted under a voltage of ±1.8 V.
• HVD (high-voltage-differential) - an analogue of the previous version, but with special transceivers and increased voltage.

The load on the interface is distributed using terminators located at both ends of the bus. According to electrical characteristics, they are divided into:

• Passive - simple 132 ohm resistors;
• Active - stabilizers that produce the necessary signal, and each power line is connected to them with a resistance of 110 ohms;
• FPT (Forced Perfect Terminator). The name speaks for itself - an accelerated improved type. It has surge suppressors and is used in high frequency interfaces.

The 2nd model is the most commonly used.

SCSI Competitiveness

The SCSI standard has stood the test of time and is still popular today. Why?

• Has a high speed;
• You can create a chain of 15 devices;
• They are easy to manage;
• HDDs are highly reliable.

Still, such drives account for only about 30% of the modern market, since SCSI also has disadvantages:

• High cost. But you need to understand that you pay for quality. Although SATA hard drives have more capacity at a lower price, they cannot boast of such durability.
• Obsolescence. An improved competitor has appeared - SAS (Serial Attached SCSI) technology, which has more compact wires, does not need terminators, allows you to connect more devices and has better bandwidth.

That's all.

I look forward to seeing you on the blog pages as often as possible.

In this article, we'll look into the future of SCSI and look at some of the advantages and disadvantages of SCSI, SAS, and SATA interfaces.

In fact, the issue is a bit more complex than just replacing SCSI with SATA and SAS. Traditional parallel SCSI is a tried and tested interface that has been around for a long time. Currently, SCSI offers a very fast data transfer rate of 320 Megabytes per second (Mbps) using the modern Ultra320 SCSI interface. In addition, SCSI offers a wide range of features, including Command-Tag Queuing (a method of optimizing I/O commands to increase performance). SCSI hard drives are reliable; in a short distance, you can create a daisy chain of 15 devices connected to a SCSI link. These features make SCSI an excellent choice for high performance desktops and workstations, up to and including enterprise servers, to this day.

SAS hard drives use the SCSI command set and have the same reliability and performance as SCSI drives, but use a serial version of the SCSI interface at 300 Mbps. While slightly slower than 320 Mbps SCSI, the SAS interface is capable of supporting up to 128 devices over longer distances than the Ultra320 and can expand to 16,000 devices per channel. SAS hard drives offer the same reliability and rotation speeds (10000-15000) as SCSI drives.

SATA drives are a little different. Where SCSI and SAS drives focus on performance and reliability, SATA drives trade them off in favor of massive capacity increases and cost reductions. For example, a SATA drive has now reached a capacity of 1 terabyte (TB). SATA is used where maximum capacity is needed, for example, for Reserve copy data or archiving. SATA now offers point-to-point connections at speeds up to 300 Mbps, and easily outperforms the traditional parallel ATA interface at 150 Mbps.

So what will happen to SCSI? It works great. The problem with traditional SCSI is that it's just coming to the end of its useful life. Parallel SCSI at 320 Mb/s will not run much faster on current SCSI cable lengths. In comparison, SATA drives will reach 600 Mb/s in the near future, SAS have plans to reach 1200 Mb/s. SATA drives can also work with the SAS interface, so these drives can be used simultaneously in some storage systems. The potential for increased scalability and data transfer performance far exceeds that of SCSI. But SCSI isn't going away anytime soon. We will see SCSI in small and medium servers for a few more years. As hardware upgrades, SCSI will be systematically replaced by SAS/SATA drives to get faster and more convenient connections.

"Courageously stepping on uncharted land" - IDE drives on SCSI controllers

With each new generation of drives, hard drive manufacturers pull out new aces from their sleeves: the latest models are faster, quieter and more voluminous than their predecessors. They have already reached the capacity of 200 GB - and soon we will see 300 GB disks. But disks of this size with SCSI interface are not available, and SCSI is the standard for the server market.

Powerful server systems must be reliable, fast, and resourceful in terms of power and capacity. The first two parameters are easily achieved by using the best SCSI controllers and the best hard drives. But increasing storage can cost a pretty penny.

So why don't we try to use cheaper IDE Solutions- they do the same job as their more expensive SCSI counterparts. However, several arguments speak against the use of IDE drives: the maximum number of devices, the reliability of modern hard drives, and the lack of controller functionality.

Taiwanese manufacturer Acard has developed an adapter that allows IDE drives to work on SCSI controllers.

In fact, such problems do not concern home users. Even if SCSI systems work faster, they are not so attractive due to their high cost. In addition to the money that you pay for a modern hard drive, you will also need to buy a node controller. If you need a RAID controller, then be prepared to shell out at least the cost of a Pentium 4.

With two Ultra160 SCSI lanes, the Adaptec 39160 provides a level of flexibility that is hard to beat.

Today, IDE drives are characterized by high speed and volume. And as for the price, SCSI is not a competitor to them.

But the server segment dictates completely different rules of the game. It's not about extra gigabytes - the priority is given to maximum reliability and performance, since even an insignificant downtime of the server will cost serious money, and in the worst case, even call into question the existence of the company.

That is why SCSI solutions are so expensive: expensive development, high-quality components, and the market is relatively small.

However, not so long ago Maxtor announced its entry into the server segment of the market with a new line of drives with an IDE interface. With a low minimum performance and adequate reliability, the goal is to achieve a much higher capacity than SCSI drives (where the current maximum is 147 GB). In theory, the plan is good, because for the price of five Ultra320 SCSI drives, each 147 GB, you can buy 15 state-of-the-art IDE drives, each 200 GB.

The only thing missing today is the right controllers. There is little chance that manufacturers will release IDE versions of their high-end controllers. However, there are a huge number of SCSI host controllers on the market.

Aside from the IDE2SCSI adapters below, Acard is mainly known for its SCSI and IDE controllers and related products, as well as unusual data handling solutions such as CD or DVD ripping stations.

Also from Acard: dual-channel IDE RAID controller AEC-6880.

Unusual thing: IDE2SCSI adapter AEC7722, front view.

The adapter is as wide as a 5.25" drive and connects directly to the IDE hard drive. However, the current on the IDE bus is not enough to power the controller, so an external power connection is required.

For tests, we used an IBM (Hitachi) hard drive.

As you can see, the connected adapter protrudes slightly on the left. Before purchasing an adapter, be sure to check that you have enough space in your computer case.

Be careful when connecting the adapter, because the board will bend slightly under pressure.

There are no components placed on the back of the adapter. Just an IDE connector.

According to Acard, the adapter's maximum interface speed is 80 MB/s. Even if the peak transfer rate of modern disks may be higher, this throughput will be sufficient for most applications.

Chip, BIOS and jumpers (top). The last two are used to set the SCSI-ID.

The heart of the IDE2SCSI adapter is a controller made by Achip (ARC765-D).

View of the adapter from the front and back.

Upside down: Adaptec's SCSI host adapter looks for available drives. A 180 GB IDE drive from IBM was found.

The SCSI connector has 80 small pins (top). In contrast, the IDE has only 40 pins.

A typical Ultra160 SCSI cable has three to five drive connectors. For more expensive versions, the number of connectors can reach up to 15.

The SCSI specification provides for termination of both ends of the bus, that is, there must be a special resistor there to prevent signal reflections.

Testing

test system
CPU	Intel Pentium 4, 2.0 GHz 256 KB L2-Cache (Willamette)
mother board	Intel D845EBT, ​​845E chipset
Memory	256 MB DDR/PC2100, CL2, Infineon
Controller	IDE: i845E UltraDMA/100-Controller (ICH4) SCSI: Adaptec AHA-39160 Ultra160-SCSI
video card	NVIDIA GeForce2 MX 400
Network Card	3COM 905TX PCI 100MBit
OS	Windows XP Pro 5.10.2600, SP1
Tests
High-end applications	ZD WinBench 99 - Highend Disk Winmark 1.2
Performance	HD Tach 2.61, PC Mark 2002 (HD Test)
I/O performance	Intel I/O Meter
Drivers and settings
Video driver	NVIDIA reference driver 29.42
IDE driver	Intel Application Accelerator 2.2.2
DirectX Version	8.1
Permissions	1024x768 16bit 85Hz refresh

To see how a modern IDE hard drive would perform on a SCSI controller with a typical setup, we tested the IBM IC35L180 test drive in both configurations.

Conclusion: useful but expensive

The result of the tests is clear: the difference between a hard drive running on an IDE and on an Adaptec 39160 SCSI controller is negligible in all important tests.

The slightly reduced I/O performance is due to the need to convert interface protocols, which is quite important in a server environment. Each disk access operation is handled by the Achip controller. So the IDE hard drives with an adapter should not be used in disk-intensive applications (i.e. databases or web servers). In these areas, SCSI drives have a distinct advantage over their IDE counterparts, as they can provide more I/O operations per second.

SCSI adapters and IDE hard drives with adapters are interesting in applications where large hard drives are required. If you install larger hard drives, you can equip your storage with fewer drives, and more importantly, it will cost you much less than the SCSI option. Even if you install a few spare drives in case of IDE hard drive failure (in a large RAID cluster), you will still save a significant amount of money. Of course, the transition to such a configuration is, first of all, a matter of confidence in the manufacturer of hard drives.

If you are interested in the IDE2SCSI adapter, then we will disappoint you a little: it is by no means cheap. On the Acard website, prices start at $69, a pretty substantial price for a controller designed for budget-friendly solutions.

Therefore, using an Acard adapter makes sense only in cases where you save a lot of money by ditching SCSI drives and switching to bulky IDE drives, without taking into account additional expensive security measures (redundancy, mirroring, hot-swappable bays).

What is SCSI?

A: The [SCSI Basics] section is dedicated to answering this question.
What is SAS, which is better than SCSI or SAS, and how do they differ?
A: The [ SAS or SCSI ] section is dedicated to answering this question.
What is eSATA?
A: eSATA is a SATA interface designed to connect external SATA devices. It provides a 3 Gb/s link, which eliminates the bandwidth bottlenecks associated with modern devices external data storage.

What is Unified Serial?
A: All Unified Serial controllers allow you to connect SATA and SAS drives using a point-to-point interface. It uses an extended SCSI command set to provide powerful data management, error handling, and performance.

The flexibility provided by SATA and SAS drive support makes it easy for companies to standardize their I/O infrastructure for both primary storage of mission-critical data and secondary storage, depending on whether SATA or SAS drives are installed. Customers can standardize their infrastructure with unified I/O controllers and storage systems to reduce training and maintenance costs.

Can SATA drives be used with SAS controllers?

A: Yes, you can, while on the same controller you can simultaneously use both SAS and SATA drives. This allows you to start the transition to SAS technology right now for moderate money.

Can SAS drives be used with SATA controllers?
Oh no.

Is it possible to connect SAS disks to a controller without using a hotswap bucket?
O: Yes, you can. To do this, you need to use a special cable with an SFF-8482 connector on the side of the drives. The connector on the other end of the cable is determined by the SAS controller.

What is the difference between SCSI-1, SCSI-2, Fast, Wide,Ultra Wide and Ultra2 SCSI?
A: The main difference is in the set of SCSI commands and the bus width (respectively, in speed).
SCSI-1 5MB/Sec 8 bit SCSI bus
SCSI-2 5MB/Sec 8 bit SCSI bus
SCSI-2 Fast 10MB/Sec 8 bit SCSI bus
SCSI-2 Fast Wide 20MB/Sec 16 bit SCSI bus
SCSI Ultra 20MB/Sec 8 bit SCSI bus
SCSI Ultra Wide 40MB/Sec 16 bit SCSI bus
Ultra2 Wide 80MB/sec 16 bit SCSI bus
Ultra160 160MB/sec 16bit SCSI bus
Ultra320 320MB/sec 16bit SCSI bus

When should I use a Low Voltage Differential (LVD) controller?
A: In case:
High data transfer rate required - 80 - 320 MB/s
There is a very high level of electromagnetic noise in the environment, which affects data transmission. LVD mode provides much better noise immunity than Single Ended (SE) SCSI
It is necessary to provide a significant distance of SCSI devices from the computer. LVD devices can be removed from the SCSI controller up to 12 meters (this is the maximum allowed length of the LVD SCSI cable.

What is SCSI terminator and why is it needed?
A: The SCSI Terminator is a small electronic device that must be located at both ends of the SCSI bus and there must be exactly two of them (terminators) per SCSI bus. Most often, the first SCSI Terminator is the SCSI controller (as a rule, this function can be "turned off" in the BIOS of the controller, but it is enabled by default), and the second is the terminator connected to the last (from the SCSI controller) SCSI connector of the cable.

Some SCSI devices (legacy disks, floppy drives, tape drives) have a built-in terminator that can be enabled by a corresponding jumper on the device. In this case, you need to ensure that the device with the terminator enabled is located at the very end of the SCSI bus.

And everything works for me even without SCSI terminator, maybe it will work like that?
A: For the time being, it can and will do, especially if you only have one disk and it is not used very intensively. But as the number of devices on the SCSI bus increases, or as the load on it increases, you eventually run the risk of losing data, so don't skimp on it.

What is a SCSI ID and why is it needed?
A: SCSI ID is a unique (within one SCSI bus) identifier (number) of a SCSI device. It is needed to provide addressing to devices on the SCSI bus.

A SCSI ID is assigned either automatically (for example, if you use hotswap drive cages that support this feature) or by manually setting the appropriate jumpers on SCSI devices. The SCSI ID has nothing to do with the physical order of the devices on the SCSI bus (for example, a SCSI controller typically has a default SCSI ID of 7, although most often, but not always, it is located at the beginning of the SCSI bus), it only matters so that there are no devices with the same SCSI ID on the same SCSI bus.

SCSI ID values ​​can be:
0 to 15 (16 in total) for Wide (W) and UltraWide (UW, U2W, U160, U320) SCSI buses;
0 to 7 (8 total) for Narrow (U, U2) SCSI bus;

What happens if two devices with the same SCSI ID are connected to the same SCSI channel?
O: Nothing good. In the best case, the SCSI controller recognizes one of these devices, but still cannot work with it correctly; in the worst case, it will not "see" any of these devices. Neither the controller nor the disks will be damaged, but the risk of corrupting data on SCSI disks remains.

It should be taken into account that the vast majority of controllers do not report such an error in any way, so when connecting new devices to the SCSI bus, you need to pay attention to observing the uniqueness of the SCSI ID.

Please note that the SCSI controller itself also has a SCSI ID (usually it is 7, and can be changed in the controller's BIOS), so you should not assign the same SCSI ID to disks.

What is SAF-TE?
A: SAF-TE - SCSI Accessed Fault-Tolerant Enclosure is an "open" specification designed to provide a comprehensive and standardized method for monitoring and displaying information about the status of drives, power supplies, and cooling systems used in high availability servers and storage subsystems. The specifications are independent of the I/O hardware, operating systems, and server platform because the chassis itself appears to be just another device on the SCSI bus. SAF-TE specifications have been adopted by many leading manufacturers of servers, storage devices and RAID controllers. Products that meet the SAF-TE specification reduce the cost of monitoring chassis health, simplify the work of the network administrator, provide alarm notification and equipment status information.

	General information about interfaces……………………………………….
	Classification of interfaces………………………………………………
	The history of the creation of the SCSI interface……………………………………
	Evolution of SCSI standards………………………………………………..
	What does a SCSI controller look like and what does it consist of…………………….
	SCSI concept………………………………………………………….
	SCSI bus phases…………………………………………………..
	SCSI Commands………………………………………………………………
	Host adapters………………………………………………………….
	SCSI cables……………………………………………………………...
	Software support for SCSI devices……………………………...
	Hardware programming peripherals…
	SCSI vs IDE……………………………………………………………
	Bibliography………………………………………………………

1. General information about interfaces

The creation of modern computer facilities is associated with the task of combining into one complex various computer blocks, information storage and display devices, data equipment and the computer itself. This task is assigned to unified interface systems - interfaces. An interface is understood as a set of circuitry means that provide direct interaction between the constituent elements of a computer system. An interface provides an interconnection between the constituent functional units or devices of a system.

The main purpose of the interface is the unification of intra-system and inter-system communications and interface devices in order to effectively implement progressive methods for designing functional elements of a computer system.

2. Classification of interfaces

1) Machine interfaces are designed to organize connections between the constituent elements of a computer, i.e. directly for their construction and connection with the external environment.

2) Peripheral equipment interfaces perform the functions of interfacing processors, controllers, storage devices and data transmission equipment.

3) The interfaces of multiprocessor systems are basically trunk interface systems oriented into a single complex of several processors, memory modules, memory controllers, limited in space.

4) Distributed aircraft interfaces are designed to integrate information processing facilities located at a considerable distance.

The development of interfaces is carried out in the direction of increasing the level of unification of interface equipment and standardization of compatibility conditions, modernization of existing interfaces, creation of fundamentally new interfaces.

3. History creating an interface SCSI

The name Shugart is familiar to many: it belongs to one of the brightest pioneers and ideologists of the "storage" industry - the legendary silicon Olympian (in the sense of the inhabitant of the Silicon Valley Olympus) Alan F. Shugart, who led the development of floppy and RIGID at IBM, then worked at Memorex. In 1973, Shugart raised capital from outside and formed a 5.25-inch FDD drive company, Shugart Associates. This firm worked under his management for a year, after which Shugart was kicked out by the very people who invested in the undertaking. Shugart recovered from the blow for six years - during this period he even bought a fishing boat and became a professional fisherman. But the craving for high-tech did not go away: in 1979, together with Finis Conner, he founded Seagate Technologies (originally Shugart Technologies), after which he remained its leader for almost two decades, during which the company became the largest independent manufacturer of hard drives (albeit from Seagate in 1998, Shugart was "flooded", but that's a completely different story).

We are more interested in Shugart Associates, since it was they who developed the SASI interface, the earliest version of the SCSI bus, in 1979. Expanding the SASI abbreviation is currently difficult, the first two letters reliably mean Shugart Associates, the fourth - Interface, and the third stands for System, Systems or Standard in different sources (I think the correct version is still the latter). The capabilities of SASI were very modest even compared to the first version of SCSI - the transfer rate was only 1.5 MB / s, the interface had a very limited set of commands. However, the ideas embedded in SASI carried a lot of progressive: instead of the then ubiquitous analog serial transmission, 8-bit parallel digital was used, instead of a bunch of control lines, the interface provided a set of commands, and it worked at the logical level, allowing you to address blocks, not physical heads , cylinders and sectors.

Two years later, in late 1981, to spur industry acceptance of the interface, Shugart Associates partnered with NCR (National Cash Register) to apply to ANSI to form a technical committee to refine and standardize the interface. Such a committee, X3T9.2, was formed in 1982, and the interface name changed to the impersonal descriptive SCSI. Over the next few years, the standard was refined and improved: the bandwidth expanded, instruction sets were added - for printers, streamers, processors, WORM and ROM devices. (It should be noted that SCSI, unlike SASI, is no longer just disk interface, but by the kind of system bus: theoretically, on a “bare” SCSI, you can assemble a full-fledged system by connecting a processor, memory, drives, and peripherals.) protostandard. The first official standard - X3.131-1986 - was adopted in 1986 (with the advent of subsequent versions, it became known as SCSI-1).

Subsequent additions and improvements led to the creation of the SCSI-2 specification.

4. Evolution of SCSI standards

The SCSI specifications strictly define the physical and electrical parameters of the interface and the minimum number of commands. The use of these commands became the main advantage of the SCSI interface, as it made it manageable. Developed in December 1985, the SCSI-1 specification provided for data transfer over a bus with a width of 8 bits and a frequency of 5 MHz. The data transfer rate on the SCSI bus in standard asynchronous mode (or handshake mode, i.e. when acknowledgment is required after each data send) is about 3 Mb / s. When transmitted in synchronous mode, the SCSI bus is capable of developing a throughput of about 5 Mb / s.

The devices were connected in a chain one after another. The first device connected to the SCSI interface on the host computer, the second connected to the first, and so on (see Figure 1). The first and last devices in the chain had to be terminated. On all other devices, termination had to be disabled. Devices were identified by a jumper or switchable ID (from 0 to 7), with the bus adapter on the host typically assigned ID=7 as giving the highest priority for bus access.

Figure 1. Typical scheme for connecting SCSI devices in the form of a chain.

The standard did not oblige the use of any particular type of connectors (connectors), but only described the purpose of the contacts. The most widely used D-Ribbon type Centronics connectors for PCs, as well as DB-25 for Macintosh. Termination was predominantly passive, while active or controlled termination was used only by individual manufacturers.

In March 1990, the SCSI-2 (Fast SCSI) specification was developed and officially approved in 1992, which defines 18 basic SCSI commands (Common Command Set, CCS), mandatory for all peripheral devices, as well as additional commands for CD- ROM and other peripherals. It became possible to exchange data without the participation of the central processor. There were "queues" - the ability to accept chains of up to 256 commands and process them autonomously in an optimized order. And if the controller of the destination executive device received a command that does not require any external interactions, then this controller will not occupy the bus until it becomes necessary to transfer some data. Here you can see a major advantage of SCSI over IDE, especially in multitasking environments: the IDE bus acts as a passive signal path from the CPU - it must execute one command first before initiating another.

Specification extensions have also appeared, the designations of which can often be seen in price lists. The basic 8-bit version - Fast SCSI (SCSI-2) - has a throughput of 10 Mbps. The Wide SCSI-2 modification is a 16-bit version of Fast SCSI (SCSI-2) and, accordingly, has a double data transfer rate, and also allows you to connect up to 15 peripheral devices. The Ultra prefix denotes an operating frequency increased to 20 MHz, and Ultra2 controllers are capable of transmitting data at a frequency of 40 MHz. Very often there are designations Ultra Wide or Ultra2 Wide. This means that combinations of options are used. So, for example, Ultra2 Wide devices can exchange information with a maximum speed of 80 Mb / s.

The Ultra160/m SCSI specification was adopted on September 14, 1998. The main components of Ultra160/m SCSI were: double synchronization during data transfer (Double Transition Clocking), data integrity control through the use of cyclic redundancy code (CRC), environment control (Domain Validation). A data transfer rate of 160 Mbps is achieved by using both edges of the challenge/acknowledge signal to synchronize the data. Accordingly, this allows developers to increase performance or reliability, as it becomes possible to use bus bandwidth up to 160 Mbps with existing Ultra2 SCSI connecting cables, or to increase the reliability of the Ultra2 SCSI interface (80 Mbps) by reducing the frequency at which synchronization occurs.

With regard to data integrity control through the use of cyclic redundancy code (CRC), the Ultra160/m uses the same method that is used in FDDI, in local networks based on the CSMA-CD protocol and in fiber-optic data transmission channels. Environment Control is an intelligent technology that checks the storage subsystem, including connecting cables, terminators, etc. This technology controls the operation of the system to the required specifications, and in case of danger of data loss, even lowers the transfer rate.

According to the method of communication with the controller, SCSI devices are divided into two types: using single-ended and differential (differential, D) electrical interfaces. The single-ended interface uses one wire for each bit of transmitted data or control signals and a corresponding wire for ground, with information carried on only one signal wire. In the differential interface, the signal is divided into positive and negative components and transmitted over a pair of conductors, which makes it possible to transmit a signal over long distances without interference. The choice of SCSI transceiver type determines the maximum bus length and the number of connected devices. Most existing SCSI devices use single-ended transceivers, resulting in shorter cable lengths as the transfer rate increases. Differential transceivers overcome this limitation, but are much more expensive. Low-Voltage Differential (LVD) technology, which is a hybrid of the above two technologies, is designed to solve this problem. Most of the newer devices support universal transceivers that can operate as both single-ended and LVD transceivers.

bit depth,

Maximum transfer rate, Mb/s

Maximum cable length/number of devices, m/piece

Number of pins in the connector

6/7.25/6(0), 12/6 (LVD)

3/7.25/6(0), 12/6 (LVD)

Fast SCSI-2, Fast SCSI

3/15.25/15(0), 12/15 (LVD)

3/3.1.5/7.25/6(D),12/6(LVD)

Wide Ultra SCSI-2

3/3.1.5/7.25/15(D), 12/15(LVD)

Fast-20 Wide SCSI

Wide Ultra2 SCSI-2

Fast-40 Wide SCSI

Ultra3 Wide SCSI

There is also an 80-pin connector for connecting devices in Hot Swap mode. A feature of this connector is the presence of power contacts along with contacts for transmitting data and control signals.

5. What does a SCSI controller look like and what does it consist of?

Here is a picture of the simplest FastSCSI controller on PCI bus.

As you can see, most of the space is occupied by connectors. The largest (and oldest) is the 8-bit internal connector, often referred to as narrow, it is similar to the IDE connector, only it has 50 pins instead of 40. Most controllers also have an external connector, as the name implies, you can and should connect external SCSI devices to it. The picture shows a mini-sub D connector with 50 pins.

For Wide devices, a similar one is used, but with 68 pins, and the fastening is also used not in the form of latches, but on screws - like in COM mice and printers. It is even smaller than narrow due to the higher contact density. (By the way, despite the name, wide plume is also narrower than narrow). Sometimes you can also find the old version of the external connector - just centronix. You can meet the same (outwardly, but not functionally :) on your printer. Some devices, such as the IOmega ZIP Plus and those designed for Macs, use a regular 25 pin Cannon (D-SUB) like a modem. Mini-centronics is also used for external high-speed connections. Here is the complete table:

(almost original dimensions)

Internal

Low Density 50-pin

connection of internal narrow devices - HDD, CD-ROM, CD-R, MO, ZIP. (like IDE, only 50 pins)

High Density 68-pin

connection of internal wide devices, mainly HDD

External

connection of external slow devices, mainly scanners, IOmega Zip Plus. most common on Mac. (like a modem)

Low Density 50-pin

or Centronics 50-pin. external connection of scanners, streamers. usually SCSI-1.

High Density 50-pin

or Micro DB50, Mini DB50. standard external narrow connector

High Density 68-pin

or Micro DB68, Mini DB68. standard external wide connector

High Density 68-pin

or Micro Centronics. according to some sources it is used for external connection of SCSI devices.

For the operation of any device, as you know, software support is required. For most IDE devices, the minimum one is built into the BIOS of the motherboard, for the rest, drivers for various operating systems are required. For SCSI devices, things are a little more complicated. For initial boot from SCSI hard drive and work in DOS requires its own SCSI BIOS. There are 3 options here.

1. A microcircuit with SCSI BIOS is on the controller itself (like on VGA cards). When the computer is booted, it is activated and allows you to boot from a SCSI hard disk or, for example, CDROM, MO. When using a non-trivial operating system (Windows NT, OS/2, *nix), drivers are always used to work with SCSI devices. They are also required for non-hard disk devices to work under DOS.

2. The SCSI BIOS image is flashed into the flash BIOS of the motherboard. Further on p.1. Usually, a SCSI BIOS is added to the motherboard BIOS for a controller based on the NCR 810 chip, Symbios Logic SYM53C810 (it is on the first picture) or Adaptec 78xx. You can control this process and change the SCSI BIOS to a newer version if desired. If there is a SCSI controller on the motherboard, this approach is used. This option is also more cost-effective :) - a controller without a BIOS chip is cheaper.

3. There is no SCSI BIOS at all. All SCSI devices work only with operating system drivers.

In addition to supporting booting from SCSI devices, the BIOS usually has several other functions: configuring the adapter, checking the surface of disks, low-level formatting, setting SCSI device initialization parameters, setting the boot device number, etc.

The next remark follows from the first. As you know, usually motherboards ah there is CMOS. In it, the BIOS stores the board settings, including the configuration of hard drives. The SCSI BIOS often needs to store the configuration of the SCSI devices as well. This role is usually performed by a small chip like 93C46 (flash). It connects to the main SCSI chip. It has only 8 legs and several tens of bytes of memory, but its contents are preserved even when the power is turned off. In this SCSI chip, the BIOS can store both SCSI device settings and its own. In the general case, its presence is not associated with the presence of a chip with a SCSI BIOS, but, as practice shows, they are usually installed together.

Here you can see the ASUSTeK UltraWide SCSI controller. It already has a SCSI BIOS chip. You can also see the internal and external Wide connectors.

The last (I couldn't find any more quickly :) picture shows a dual-channel Ultra Wide SCSI controller. Its specification includes the following items: RAID levels 0,1,3,5; Failure Drive Rebuilding ; Hot Swap and on-line Rebuilding; cache memory 2, 4, 8, 16, 32 Mb; Flash EEPROM for SCSI BIOS. The 486 processor is very clearly visible, which is apparently trying to manage all this stuff.

You can also find on the SCSI controller board

• SCSI bus activity LED and/or connector for its connection
• memory module connectors
• floppy disk controller (mostly on older Adaptec boards)
• IDE controller
• sound card(on ASUSTeK cards for MediaBus)
• VGA card

Other SCSI cards

Scanners and other slow SCSI devices often come with a simple SCSI controller. Usually this is a SCSI-1 controller on the ISA bus 16 or even 8 bits with one (external or internal) connector. It does not have a BIOS, eeprom, it often works without interruptions (polling mode), sometimes it supports only one (not 7) device. Basically, such a controller can only be used with your device, because. there are drivers for it. However, with a certain skill, you can connect to it, for example, a hard drive or streamer. This is justified only in the absence of money and the availability of time (or sports interest :), because. a standard SCSI controller, as already mentioned, can be purchased for $20-40 and have an order of magnitude fewer problems and much more features.

6. Concept SCSI

The SCSI bus is an I/O bus, not a system bus or instrument level interface. Interface facilities such as SCSI bus are especially effective for machines that require the connection of multiple disk drives or other controllers. The SCSI interface increases the flexibility and processing power of the system, since it allows several different PUs to be connected to the same bus, which can directly communicate with each other. The bus transfer rate will certainly not be a limiting factor, as the SCSI bus rate currently reaches 40MB/s.

The SCSI bus provides the ability to connect up to eight devices. At first glance, this may seem like a rather severe limitation, but considering that each device can represent eight logical blocks, and each logical block can represent 256 logical subblocks, it is obvious that the expansion possibilities here are more than ample.

Each SCSI bus device must be assigned an individual ID, the value of which is usually set using jumpers directly in the device. The ID has two functions: it identifies the device on the bus and determines its priority in arbitration for bus access (the higher the device number, the higher its priority).

Each of the eight possible bus devices can play the role of initiator, target, or both. The initiator is part of the SCSI host (master) adapter that connects the host computer to the SCSI bus. In a typical system, one or more workers are connected to one initiator. An advanced system can contain more than one SCSI host adapter (many initiators). In such systems, not only any processor can interact with any PU, but also host adapters can interact with each other, since the host adapter itself is a SCSI bus device and can play the role of both an initiator and an executor. Two PUs (both slaves), however, cannot interact with each other, since only a pair of initiator - slave can communicate on the bus at any given time.

The host adapter contains the hardware and software to interface with the CPU.

The interface between the SCSI controller and the system bus can be either very simple (based on the principle of software polling of the I / O channel) or more complex (providing for high-speed data exchanges in direct memory access mode, DMA). Such controllers accept high-level commands and relieve the CPU of the need to process and control SCSI bus signals.

The host computer software is simplified because it does not have to take into account the physical characteristics of a particular device. The SCSI interface provides for the use of logical rather than physical addresses for all blocks of data.

7. Bus phases SCSI

The SCSI bus protocol has eight distinct phases:

Bus Free - "The bus is free"

Arbitration - "Arbitration"

Selection - "Selection"

Reselection - "Reverse selection"

Command - "Team"

Data - "Data"

Status - "Status"

Message - "Message"

The last four phases are called the information transfer phases. The SCSI bus can only be in one of these eight phases at any given time.

The Bus Free phase means that no device is currently active on the SCSI bus, and the bus is free to access. This phase usually occurs after a system reset or after a bus reset by the RST signal. The indication of the Bus Free phase is the absence of busy signals BSY and sampling SEL.

The bus switches to the Arbitration phase when a SCSI device wants to take control of the bus, i.e. become an initiator on the bus. This occurs in cases where the initiator wants to select an executor or the executor wants to reselect the initiator that previously requested it. The bus can only switch to the Arbitration phase from the Bus Free phase. After the device determines that the bus is free, the "Arbitrage" phase begins. For this, a BSY signal is generated on the corresponding data line

SCSI ID - device identifier (ID - bit) is issued. At the same time, each

of the eight possible SCSI bus devices can give out its ID - bit

only to the data line assigned to him as a sign of his participation

in arbitration. The device with the highest ID wins the arbitration and takes control of the bus.

The Fetch phase allows the initiator to select an executor to initiate the execution of the appropriate function, such as a READ command or a READ write command. According to the SCSI-2 specification protocol, the Sampling phase always comes after the Arbitration phase. The SCSI-1 specification provides for a variant of the system with a single initiator, where there is no need for arbitration, and the fetch phase can be entered immediately after the Bus Free phase. In both cases, to fetch the slave, the initiator issues its ID bit on the corresponding SCSI bus data line and generates the SEL sample signal.

An optional refetch phase is possible when an executor wants to re-establish communication with the initiator that previously sent the command to it. This phase is similar in principle to the Sampling phase, with the exception that the I/O line goes active along with the SEL signal, which makes it possible to distinguish between the two phases.

The "Command", "Data", "Status" and "Message" phases form a group of information transfer phases since they are all used to transfer data or control information over the data bus. To distinguish between them, the signals C / D - control, I / O - input / output and MSG - a message are used, generated by the executors, which thereby controls all transitions from one phase to another. To control the transfer of data between the executor and the initiator in the phases of information transfer, the signals of the REQ / ACK lines - request / confirmation are used (in the SCSI-2 version, the REQB / ACKB lines are additionally used).

Real data exchange can be carried out in a synchronous and asynchronous way. In both cases, the ACK and REQ signal lines are used to perform the handshake. Synchronous transfer mode is optional for the performer. The initiator MAY request that the worker perform a synchronous transfer, but if the worker rejects this request, asynchronous mode will be used.

To send data to the initiator in asynchronous mode, the slave issues it on the SCSI bus data lines along with the REQ signal. Data must be held on the bus until an ACK is received from the initiator. After that, the next data is output to the bus, and the process is repeated. If data transfer is to occur in the opposite direction, the worker issues a REQ request signal indicating that it is ready to receive data. The initiator issues data on the SCSI bus data line, and then generates an ACK signal. The initiator continues to hold data on the bus until the REQ line switches to the passive state. The worker then resets the REQ signal, the initiator emits new data, and the process repeats.

If the devices agreed to use synchronous mode during the Message phase, then the slave will not wait for an ACK signal before issuing a REQ signal to receive the next data. It can generate one or more REQ pulses without waiting for the corresponding ACK pulses (up to a predetermined maximum called the REQ/ACK offset).

When issuing all scheduled REQ pulses, the worker compares the number of REQs and ACKs to verify that each data frame was received successfully. When preparing the synchronous exchange mode, the devices set the REQ/ACK offset and the transmission period. The transmission period determines the time interval between the end of the transmission of the next byte and the beginning of the transmission of the next.

8. SCSI Commands

Previous interface specifications for hard drives (like ESDI already mentioned) called for serial transmission one bit at a time, with drive control carried out on separate wires (lines), each of which performed a specific function. For example, one specific signal line set the offset of the hard disk read / write head, another - the direction of the offset, the third - the type of operation (read or write), the fourth served to transfer data in the required format. Thus, the controller used depended on the type of hard drive.

SCSI, on the other hand, is capable of performing high-level commands, such as querying the type of device connected to the bus using the Inquiry command. Thus, in addition to specifying the physical characteristics of the bus (type of connector, voltage levels, pin assignments, etc.), the standard for each type of peripheral (hard disk, CD-ROM, etc.) defines supported commands and their corresponding responses (of the order 12 for each kind of peripheral). The standard SCSI-1 commands are grouped according to the six types of devices, as shown in Table 1.

Table 1. Groups of commands according to the types of supported devices.

Device type

Name

Typical Function

Random read/write access (hard disk)

Logical block addresses, writable block length

Sequential access (tape drive)

Reading the next entry

Page layout control

CPU

Sending and receiving

WORM (Writable CD-ROM)

Large size, removable

Random read-only access

Logic block addresses, read block length

When the target device requests a command, as in the PC disk access example, the initiator responds by sending 6 bytes of command information. These bytes serve to set the command and identify the device. Together they are called the Command Descriptor Block (CDB). The first byte (more precisely, byte number 0) determines the type of command or operating code (opcode). Some of the more common codes have the following meanings (in hexadecimal):

00 Test device ready;

03 Formatting;

08 Reading;

0A Write;

0B Search.

The meaning of the remaining bytes depends on the particular opcode. For example, in the case of the Write command (code 0A), they have the following meaning:

Byte 0 Operation code 0A;

Byte 1 LUN number in bits 5 and 6,

bits 1 to 4 set the address of the logical block;

Byte 2 Logic block address;

Byte 3 Logic block address;

Byte 4 Bits 2 to 5 specify the transmission length;

Byte 5 Bit 1 - flag; bits 6 and 7 are assigned by the manufacturer.

The transmission of commands is carried out in an asynchronous mode. However, if the response contains data, it can be transmitted synchronously, as is the case with the Inquiry command, in which the target device responds with an ASCII string identifying its type (this response is often displayed on the PC monitor when loading SCSI drivers).

9. Host adapters

The host adapter implements the functions of interfacing the SCSI bus with system resources, primarily with the system bus and operating system computer. It usually acts as the initiator on the SCSI bus, although in complex (for example, multiprocessor and multimachine) SCSI systems, it can dynamically change (initiator / executor).

The main functions of the host adapter, which determine its structure and characteristics, include:

Implementation of the SCSI bus protocol, as well as the physical and electrical specifications of the standard;

Interfacing with hardware and software system resources

The implementation of the SCSI bus protocol is usually carried out by a specialized LSI of the SCSI bus controller. Typically, this circuit also provides the implementation of the electrical specifications of the standard.

Interfacing with system hardware means, first of all, matching the bit width and bandwidth of the SCSI bus and the system bus of the host system, as well as the implementation of advanced means of accessing system memory. The structure of the bus width negotiation node depends on the purpose of the host adapter and the version of the SCSI standard used (8 bits for SCSI-1; 16 or 32 bits for SCSI-2). The main means of matching the throughput of the system and SCSI buses is buffer memory, which is usually implemented as a FIFO buffer or dual-port RAM. The most common system memory access algorithm is direct access, most often implemented using the host system's DMA controller.

Interfacing with software systems requires the presence of a SCSI driver for a particular OS.

Characteristics of modern host adapters

Among the LSI SCSI controllers used for the AT bus, NCR models dominate. Following are well-known WD33C93 from Western Digital and ALC 6250/60 from Adaptec (USA). The host adapter most often supports both synchronous and asynchronous modes of exchange over the SCSI bus. The exchange rate significantly depends on the type of controller used. In simple host adapters, it ranges from 0.25 to 1 MB / s in asynchronous mode and synchronous modes, respectively.

The size of the data buffer also varies within a fairly wide range: from using the internal buffers of the small-capacity BIC SCSI controller to a large-capacity RAM (1 MB). The presence of a large buffer significantly increases the cost of the host adapter.

10. SCSI cables

In addition to using twisted pairs, external SCSI cables are arranged in three concentric layers to provide noise immunity (see Figure 2). The central, inner layer contains three pairs: Request ("Request"), Acknowledge ("Confirmation") and Ground ("Earth"). The middle - intermediate - layer serves to transmit control signals. The third - outer - layer is designed to transmit data and parity information. In the middle layer, the pairs are twisted in the opposite direction compared to the outer and inner layers adjacent to it to reduce the capacitive coupling between the layers. The placement of the cores for transmitting control signals in the middle layer ensures that there is no interference between data and Request/Acknowledge signals.

Figure 2. Cross-sectional view of an external SCSI cable.

Although the entire cable is insulated with PVC, it is not suitable for individual pairs, as its electrical characteristics are highly dependent on temperature, and in addition, it has a very high capacitance. This design of the cable ultimately affects its price. However, we are not rich enough to buy cheap things.

11. Software support SCSI devices

The task of programming SCSI systems and devices is multilevel and can be divided into the following relatively independent subtasks:

Programming hardware peripherals.

Implementation of SCSI bus protocols.

Implementation of SCSI commands.

Access to SCSI devices for OS and application tasks.

Unfortunately, the solutions used in practice at all these levels are poorly unified. Many reputable firms offer their original, but often incompatible approaches. Taking into account that at present the standard has not yet been formed in the field of programming SCSI devices, it is advisable to consider the most interesting solutions at each of the levels.

12. Hardware programming of peripheral devices

Due to the specificity of the physical principles of their implementation, highly specialized low-level programs are inevitably the final link in the means of software support for PU. Due to the fact that programming at this level is difficult even for general system, not to mention application programmers, there is a tendency to increase the level of PU programming tools by masking the specifics of PU at the level of the so-called firmware (internal software- VPO). An example is the masking of the functions of direct control of disk drives at the level of internal commands of disk controllers WD2010,8272, etc.

However, only specialized programs reach the level of controller registers. Currently, PUs are usually programmed at the level of system BIOS functions, and higher-level programs generally use standard OS functions.

The use of the SCSI interface further increases the level of programming of the PU by using a set of general commands defined by the standard. For an application programmer, using standard BIOS functions becomes almost impossible.

However, as an element of device control, of course,

are stored at the HPO level of the PU controller and are implemented either by the local microprocessor (MP) of the controller, or by a microcontroller built into the basic LSI of the PU controller.

In order to preserve the developed software tools for controlling the PU electronics, emulation of standard PU interfaces is currently widely used, which involves converting SCSI logical addresses into physical addresses of a specific device. An example is the SmartConnex/ISA controller from Distributed Processing E Technology. It uses the interface of the well-known Western Digital WD1003 disk controller, as a result of which the computer “sees” the controller as a normal device compatible with the ST-506 interface.

In reality, interface emulation is performed by a driver invisible to the user, which is remembered during formatting in the last NMD block. Appropriate drivers are available for the most common operating systems

(MS-DOS, OS/2, Xenix/Unix, Novell NetWare). Installation of the SmartConnex controller into the system is carried out using a special utility supplied by the company.

Known WD 33C92/93 controllers from Western Digital even have a built-in command for converting logical address formats into physical ones.

Thus, to implement various PUs in the SCSI standard,

use fragments of ready-made programs that support such standard control functions in MS-DOS as INT 13, INT 11, etc.

It should be noted that this approach apparently does not fully correspond to the SCSI ideology, and in the future special programs for direct control of a SCSI device based on SCSI commands will be used.

13. SCSI vs. IDE

The debate "Which is better: IDE or SCSI" is one of the most common in many newsgroups. The number of messages and articles on this topic is very large. However, this question, like the famous "Windows NT or OS/2 or Unix", is insoluble in this formulation. The most frequent and correct reaction to them is "What for?". After considering this issue in more detail, you will be able to make a decision for yourself about the need for SCSI for yourself.

Let's tell you in more detail what a simple SCSI controller can give compared to an IDE and why you should choose it or not.

SCSI offer

EIDE/ATAPI objections

SCSI response

the ability to connect 7 devices to one controller (to Wide - 15)

it is not difficult to install 4 IDE controllers and there will be 8 devices in total

for each IDE controller you need an interrupt! And only 2 will be with UDMA/33. And 4 UWSCSI is 60 devices :)

wide range of connected devices

IDE has CDD, ZIP, MO, CD-R, CD-RW

Do you have drivers and programs for all this? and more? but for SCSI you can use any, including those included in the OS

Ability to connect both internal and external devices

Removable rack or LPT-IDE

the total length of the SCSI cable can be up to 25 meters. In normal versions 3-6m *

if you do not overclock the PCI bus, you can also by a meter

you can use caching and RAID technologies to dramatically improve performance and reliability

there used to be caching Tekrams, and now there are RAID for IDE

it doesn't work and it's not serious at all

* It should be noted that when using the Ultra or Ultra Wide SCSI interface, there are additional restrictions on the quality of connecting cables and their length, as a result of which the maximum connection length can be significantly reduced.

In order not to give the impression that the IDE is very bad and you should be ashamed of using it, we also note the positive qualities of the IDE interface, partly in the light of the above table:

1. Price. Undoubtedly sometimes it very important.

2. Not everyone needs to connect 4 HDDs and 3 CDDs. Often two IDE channels are more than enough, and all sorts of scanners come with their own cards.

3. It is difficult to use a cable in the minitower case, it is longer than 80cm :)

4. IDE HD is much easier to install, there is only one jumper, not 4-16 like on SCSI :)

5. Most motherboards already have an IDE controller

6. For IDE devices, the bus is always 16 bits, and for models comparable in price, IDE wins in speed.

Now about the price. The simplest SCSI to ISA bus costs about $20, but now nobody needs such ones, so you can find cheaper ones. The next option is a controller on the PCI bus. The simplest version of FastSCSI costs about $40. However, there are now many motherboards that can install the Adaptec 7880 UltraWideSCSI for as little as +$70. Even the famous ASUS P55T2P4 and P2L97 have SCSI options. For UWSCSI cards, the price ranges from $100 to $600. There are also dual-channel (like IDE on Intel Triton HX/VX/TX) controllers. Their price is naturally higher. Note that in the case of SCSI, in contrast to IDE, where it is difficult to come up with something new, for additional money, controllers can be expanded with the functions of a cache controller, RAID-0..5, hotswap, etc., so talking about the top controller cost limit is not quite correct.

And finally, about speed. As you know, today the maximum information transfer rate on the IDE bus is 33Mb / s. For UWSCSI, a similar parameter reaches 40Mb / s. The main advantages of SCSI are manifested when working in multitasking environments (well, not much in Windows95 :). Many of the benchmarks given under WindowsNT show the undoubted advantage of SCSI. Perhaps this is the most popular OS today, for which the use of SCSI is more than justified. There may also be specific tasks (related, for example, to video processing) in which it is simply impossible to use the IDE. We will not talk about differences in internal architectures that also affect performance in this article, since there are too many special terms. We only note that while watching the development of the IDE, we are surprised to notice that it acquires many features of SCSI, but, let's hope, they will not merge at all.

Bibliography

1. Mikhail Guk: “PC interfaces. Directory "" Peter ", 1999.

2. A.P. Pyatibratov:

"Computers, systems and networks"

3. A.A. Myachev, V.N. Stepanov:

"Personal computers and microcomputers"

M .: "Radio and communication", 1998.

4. A.A. Myachev:

"Interfaces of the IBM PC", 1992.

5. Stefan Feutz: "Windows 98 for the user"

K .: BHV Trade and Publishing Bureau, 1998;

6. "PC Computing": "IDE vs SCSI"

7. "PC Magazine": "Interface IDE"

Tutoring

Need help learning a topic?

Our experts will advise or provide tutoring services on topics of interest to you.
Submit an application indicating the topic right now to find out about the possibility of obtaining a consultation.