Types of Flash memory. What is Flash Memory

Today, manufacturers produce several types of flash memory drives: these are cards Compact Flash, SmartMedia, MultiMedia Card, SecureDigital Card, Memory Stick and USB keys.

ATAFlash. The first flash memory drives to hit the market were cards ATA Flash . These drives are manufactured in the form of standard cards PC Card . In addition to flash memory chips, an ATA controller is installed in them, and during operation they emulate a regular IDE -disk. The interface of these cards is parallel. Cards ATA Flash are not widely used and are currently used extremely rarely.

CompactFlash. Compact Flash Cards (CF ) were offered by the company SanDisk as a more compact and easy-to-use alternative to cards ATA Flash . Therefore, the developers of the standard CF provided for the possibility of operating these cards as devices PC Card or as IDE -devices. In the first case, the cards work like regular ones PC Card devices and their interface “turns” into a bus PC Card . In the second - how tough IDE -disks and their interface works like an ATA bus.

CF Cards first appeared in 1994. All cards of this type have a 50-pin parallel interface. By the way, there are maps CF two types - Ture I and Tour II . Toure type cards II two millimeters thicker and appeared only because Toure card bodies preceded them I did not allow large-capacity flash memory to be placed inside for the production of capacious media CF . Currently there is no such need and the Toure card II gradually leaving the market. Please note that the drives for Tour cards II You can install Tour maps I , while the opposite is not possible.

Among flash cards, the undisputed leader in performance was Transcend Ultra Performance CF Card 25 x CompactFlash 256 MB, which can rightfully be considered the standard for the speed of modern flash drives. The sequential/random write speed of this flash card reaches 3.6/0.8 MB/s, the read speed is 4.0/3.7 MB/s.

CF operating speed -cards slow down with increasing volume, which is clearly seen in the example of flash cards512 MB. A twofold increase in capacity leads to a 30% decrease in productivity. with the exception of the random write speed, which has increased by 2.5 times - this looks rather strange and unexpected.

Speed characteristics CF -cards also greatly depend on the manufacturer. U Kingston CompactFlash 256 MB - low write speed (sequential/random write - 1.4/0.3 MB/s), but in terms of read speed it was the leader (4.4/3.8 MB/s). Map PQI Hi - Speed Compact Flash 256 MB showed average performance in both cases: writing - 2.1/0.7 MB/s, reading - 3.8/3.3 MB/s. Cards SanDisk CompactFlash 256 MB and SanDisk CompactFlash 512 MB worked very slowly: writing - 1.1/0.2 and 0.9/0.5 MB/s, reading - 2.3/2.1 and 1.8/1.7 MB/s. And the map256 MB wrote and read data equally well.

If we compare CF cards with other types of drives, it turns out that flash memory is not at all as slow as is commonly believed! In terms of performance, the fastest flash memory samples (let's take the card as a standard Transcend Ultra Performance 25x CompactFlash 256 MB) are comparable to Iomega Zip 750 MB, and in terms of sequential write speed they even outpace this drive by more than 1.5 times! Flash memory outperforms disks in sequential write speed CD-RW 2 times, sequential reading speed - by 10%! Flash memory outperforms MO disks in sequential write speed - 2 times - and random read speed - by 10%, but lags behind in sequential read speed and random write speed - by 20%. Flash memory lags behind in sequential write speed DVD -disks (when “burning” in 4x mode) - 1.4 times.

Note that if CF - the card is used in a digital camera, then speed is primarily important for it consistent recording - the higher it is, the faster the camera will return to working condition after “capturing” the frame and “resetting” it to the flash card. However, the reading speed CF -cards in this case are also important, although not so critical - the faster the data is read, the faster the camera will work in the viewing mode of the footage.

SmartMedia . Card design SmartMedia (SM ) is extremely simple. On the map S.M. there is no built-in interface controller and, in fact, it is one or two flash memory chips “packed” in a plastic casing. Standard S.M. was developed by companies Toshiba and Samsung in 1995 Map Interface S.M. - parallel, 22-pin, but only eight lines are used for data transmission.

MultiMedia Card . Multi-Media Cards (MMC) ) have a 7-pin serial interface that can operate at frequencies up to 20 MHz. Inside the plastic case of the card there is a flash memory chip and an MMC interface controller. The MMC standard was proposed in 1997 by companies Hitachi, SanDisk and Siemens.

SecureDigital Card . SecureDigi-tal Card (SD ) is the youngest flash card standard: it was developed in 2000 by companies Matsushita, SanDisk and Toshiba. Actually SD - this is a further development of the MMC standard, so MMC cards can be installed in drives SD (the reverse will not be true). Interface SD - 9-pin, serial-parallel (data can be transmitted one at a time,two or four lines simultaneously), operates at frequencies up to 25 MHz. Cards SD are equipped with a switch to protect their contents from writing (the standard also provides for a modification without such a switch).

USB -flash memory. USB flash memory (USB -memory) is a completely new type of flash memory media that appeared on the market in 2001. By USB form - the memory resembles an oblong-shaped keychain, consisting of two halves - a protective cap and the actual drive with USB - connector (one or two flash memory chips are placed inside it and USB controller).

Work with USB -memory is very convenient - no additional devices are required. It is enough to have a PC at hand running Windows with unused USB -port to “get” to the contents of this drive in a couple of minutes. Worst case scenario you will have to install drivers USB -memory, at best - new USB -the device and logical drive will appear in the system automatically. It is possible that in the future USB -memory will become the main type of device for storing and transferring small amounts of data.

What about USB? -flash memory, then this is undoubtedly a more convenient solution for transferring data than flash cards - no additional flash drive is required. However, the performance of tested drives of this type is Transcend JetFlash 256 MB and Transcend JetFlashA 256 MB - limited by low interface bandwidth USB 1.1. Therefore, their performance in speed tests was rather modest. If USB -flash memory equipped with a fast interface USB 2.0, then in terms of “rate of fire” these drives will, of course, not be inferior to the best flash cards.

It is interesting to note that in terms of sequential write speed, flash memory exceeds Iomega Zip 750, CD - RW and MO carriers and is second only to DVD -disks. This once again emphasizes that flash memory developers primarily sought to increase speed consistent recording, since flash memory was originally intended for use in digital cameras, where this indicator is primarily important.

As a result, we can conclude that flash memory is the undisputed leader in reliability, mobility and power consumption among small and medium-capacity drives, which also has good performance and sufficient capacity (flash cards with a capacity of up to 2 GB are already available on the market today). Undoubtedly, this is a very promising type, but their widespread use is still limited by high prices.

It has become indispensable in mobile devices (PDAs, tablets, smartphones, players). USB flash drives and memory cards for electronic devices (SD, MMC, miniSD, etc.) have been developed based on flash memory.

Definition 1

Flash memory(Flash Memory) – solid-state semiconductor non-volatile and rewritable memory.

Information can be read from flash memory a large number of times within the life of the drive (from $10$ years), but the number of write processes is limited (about $100\000$ rewrite cycles).

Flash memory is considered a more reliable type of storage medium, because... does not contain moving mechanical parts (such as a hard drive).

Advantages of flash memory:

high speed data access;
low power consumption;
vibration resistance;
ease of connection to a PC;
compact dimensions;
cheapness.

Disadvantages of flash memory:

limited number of write cycles;
sensitivity to electrostatic discharge.

History of Flash Memory

Flash memory was first invented in $1984.

The name “flash” comes from the English “flash”, because the process of erasing data resembled a photographic flash.

In $1988, the first commercial NOR flash processor was released. The following year, NAND flash memory architecture was developed, which featured faster write speeds and a smaller circuit area.

Principle of operation

The elementary data storage cell is a floating gate transistor that can hold electrons (charge) is the elementary data storage cell in flash memory. Based on the transistor, the main types of NAND and NOR flash memory have been developed. The operating principle is based on changing and recording the electric charge in an isolated region (“pocket”) of the semiconductor structure.

Figure 1. NOR memory architecture

Figure 2. NAND memory architecture

Flash memory manufacturers use 2 types of memory cells:

MLC(Multi-Level Cell - multi-level memory cells) - more capacious cells and cheaper, but characterized by long access times and a small number of write/erase cycles (about $10\000$);
SLC(Single-Level Cell - single-level memory cells) - cells with shorter access times and a maximum number of write/erase cycles ($100\000$).

Figure 3. Main elements of a USB flash drive: $1$ – USB connector, $2$ – controller, $3$ – PCB board, $4$ – NAND memory module, $5$ – crystal oscillator, $6$ – LED indicator , $7$ – write protection switch, $8$ – space for an additional memory chip.

Application

Exists There are two main uses for flash memory:

as a mobile information carrier;
as a software repository for digital devices.

Often both methods are combined in one device.

The use of NOR memory, which has a relatively small volume, is to provide fast access to random addresses and guarantee the absence of faulty elements (standard ROM chips for working with a microprocessor, computer boot chips (POST and BIOS), medium-sized data storage chips, for example ,DataFlash). Typical volumes range from $100$ KB to $256$ MB. NAND memory is used in mobile devices and storage media that require large amounts of storage. Basically, these are USB keys and memory cards of all types, as well as mobile devices (phones, cameras, players). NAND memory is built into household appliances: cell phones and TVs, network routers, access points, game consoles, photo frames and navigators.

Figure 4. Different types of flash cards

Types and types of memory cards and flash drives

Note 1

CF(Compact Flash) is the oldest memory type standard. It has high reliability, a fairly large volume ($128 GB or more) and high data transfer speed ($120 MB/s). Due to its large size, it is used in professional video and photographic equipment.

MMC (Multimedia Card) is small in size, highly compatible with various devices and contains a memory controller. SD Card (Secure Digital Card) is the result of the development of the MMC standard. The card has cryptographic protection against unauthorized copying, increased protection of information from accidental erasure or destruction, and a mechanical write-protection switch. Maximum capacity up to $4$ GB. SDHC (SD High Capacity) has a maximum capacity of $32$ GB.

There are also miniSD and microSD cards.

Note 2

The main manufacturers of NAND flash memory are Micron/Intel, SK Hynix, Toshiba/SanDisk, Samsung. The main manufacturers of NAND flash memory controllers are Marvell, LSI-SandForce and NAND memory manufacturers.

I constantly encounter confusion in the term flash drive, often becomes the cause of misunderstanding between the buyer and the seller when choosing the required storage medium. So, “in the broad masses” there are the following basic interpretations of the word flash drive: USB flash drive(USB flash drive), memory card microSD(read micro-ES-Di), in general any memory card, in general any flash storage medium. Here under the word flash(read flash) I mean flash memory technology and use the English term to avoid confusion. Moreover, I sometimes see that people in everyday life can simultaneously call any of these devices a flash drive, relying on the fact that their interlocutor, by context or with the help of telepathy, will understand what they are talking about!

I won’t argue about which term is more correct, much less will I skip the question of whether “flash” or “flash” is correct (in fact, both spellings are used at least the same way, and nothing can be done about it). Instead of unnecessary debate, I will simply describe all the devices called by this word, and all the words by which they are called, and then you will definitely be able to buy exactly what you need!

So let's start with USB flash drive. It is this device, which is a universal storage device containing flash memory and connected directly to a USB connector, that has given rise to the word flash drive in the Russian language. However, the word flashdrive or flashdrive, derived from the English Flash Drive, is also popular, as is the more official flash drive (or flash drive). Since it is impossible to come up with a reasonable translation for this phrase (well, don’t call flash drive a “flickering driver”!), the words flash drive or flash drive should be recognized as the best term. Here are typical examples of flash drives:

Flash drives are mainly used to transfer information between computers. Or to store information that you always want to have with you. Since we are talking about typology, I note that recently flash drives with connection USB3.0. What does it mean? This means that if your computer has a USB3.0 interface (its most noticeable external difference is the blue color), a USB3.0 flash drive will be able to work faster. If you connect it to traditional USB2.0 (what is on every computer), then its speed will be comparable to the speed of a regular flash drive. This is what USB3.0 and USB2.0 look like: Now the second category of devices called flash drives: microSD memory cards(or microSDHC, their immediate heirs)
According to my observation, they are called flash drives either by those who have never held any other flash media in their hands (and this is no wonder, since microSD/microSDHC are used in almost all phones, players and all sorts of gadgets), or by those who have other names for all these “ little things” doesn’t know. They also contain flash memory, which means they have the right to be called flash drives. But for understanding between people, it is desirable to somehow differentiate the concepts, so “memory card” will sound preferable, especially if you need to explain to the seller what you need. It is also important to know that memory cards are different! Therefore, it’s nice to add: “such a small memory card,” but even here you can get into trouble: there are M2 memory cards that are very similar in size. Fortunately, they are used only in Sony products. We will mention them below. But it’s still better to remember the magic words microSD and microSDHC (read microESDe and microESDeHaTse). In colloquial speech, by the way, most often the first word (microSD) is used to refer to both types of cards (microSD and microSDHC). There's nothing wrong with that.

What do you need to know about microSD and microSDHC memory cards? Firstly, how are they different? microSDHC is a newer standard that supports memory capacity of more than 4 GB. All memory cards over 4 GB can only be microSDHC, and less than 4 only microSD. But 4 GB is out of luck: they can be this or that! However, 4 GB microSD is very rare. Now the most important question: how to choose the one that suits your device? There are two rules: first, you need to determine the maximum capacity of the memory card that your device can work with (to do this, open the instructions for it, or use an Internet search). Secondly, you need to buy a card that is the same or less than the maximum capacity. Moreover, all devices that support microSDHC will work with any microSD card of any size. There is only one nuance here: if your device states that it supports a card of no more than 4 GB, then this may mean that it does not support any microSDHC cards and supports any microSD cards, including 4 GB. Or it could mean that it supports any 4GB cards, both microSD and microSDHC, but does not support microSDHC cards 8GB and above. This is the arithmetic. And if the instructions do not clarify this, then you will have to use the good old “scientific poking method”.

Now there is another important characteristic that buyers are often interested in: what is it? Class indicated for microSDHC cards? It is designated by a number inside the English letter C.
I must say right away that this is not a variety like, say, tomatoes. Memory card class- this is its ability to record information at a certain minimum guaranteed speed. The higher the class, the higher the speed. Moreover, this is precisely the guaranteed lowest speed, while the maximum and average speeds can be significantly higher. Two cards of different classes can often have almost the same average and maximum write speeds, but if one of them has “dips” in speed, that is, sometimes writes more slowly, then it will have a lower class. In other words: the class guarantees that the speed of the card in any part of the recording will not fall below a certain threshold. Why is it needed? The class is needed for devices that quickly record information and cannot wait. These are mainly video cameras that need to record video, because if the memory card does not have time to record a frame during its shooting, then “the train will leave”: the next frame will need to be recorded, followed by the next one, and the camera will have to “throw out” some part of the information ", which will have a bad effect on the quality of shooting. So, again, take the instructions and look at what is written in it about the class of the memory card. If nothing, you can save money; if the class is specified, take the one specified or higher.

Finally, the last thing you need to decide on when buying a microSD/microSDHC memory card is an adapter or adapter to SD. This is a thing 4 times larger than the card itself, with the help of which your micro card turns into a “large” SD/SDHC card (see about them below). Some cards are sold with an adapter, some without. Assess whether you need such an adapter, taking into account the devices you have: cameras, old e-books, etc. And also don’t forget about your card reader: maybe it doesn’t read micro cards directly and then the adapter won’t hurt you at all. In general, the adapter expands your capabilities in case of emergency. On the other hand: will you find it in your desk when you need it? The choice is yours.

Now let's move on to SD/SDHC maps
I won’t talk much about them: these are the older brothers of microSD/microSDHC cards. Everything that was said about those is also true for these overgrown ones (although it’s more likely that the “micro” cards are undersized, because at first, on the contrary, there were large ones, and then their smaller counterparts appeared). The only thing is that they don’t have adapters, since you don’t need to adapt them to yourself, and they are used in larger devices - these are, first of all, point-and-shoot cameras and all sorts of electronic books (although the latter are increasingly equipped with microSDHC cards ).

M2. Full name Memory Stick micro M2- these are cards very similar to microSD/microSDHC. They differ in that they are used in phones and players of the company Sony, It would be more correct to say “used”, because Sony finally realized that “one man in the field is not a warrior” and began to use SD line formats. If you are a happy owner of a Sony, be careful and check what card you have! These cards do not have any classes.

The last card we'll look at is Compact Flash(in Russian it is pronounced “compact flash”, but is almost always written in English, probably because writing “compact” about the largest card on the market today is somehow not literary :-).
Due to their decent size, these cards have their undoubted advantages: the capacity is many times greater than that of other cards and the speed is still unattainable for SDHC memory cards. Therefore, they are used in large “advanced” cameras and other demanding devices. It remains to add that the speed (this time without any “tricks” with a guaranteed minimum) is indicated by a number and the letter X. For example: 133x, 266x, 300x. The number represents how many times the card is faster than a certain minimum standard CD reading speed.

If you didn’t see your favorite prehistoric map in this review, don’t be upset! You will definitely find it on Wikipedia. I deliberately limited myself to only the types of flash media that are common today, so as not to fill anyone’s head with unnecessary information and not turn the article into an archivist. So, now you are armed with knowledge, and choosing the right flash drive will not be a problem for you. Enjoy the shopping!

Good day everyone!
Today’s article will mark the beginning of a new, small series of articles devoted to information storage, different types of memory, methods of writing/reading information and everything connected with it 😉 And we will start with the device of the well-known Flash memory.

What exactly is Flash memory? Yes, just an ordinary microcircuit, no different in appearance from any other. Therefore, a reasonable question may arise - what’s inside and how the processes of storing/reading information generally occur.

So, the heart of many memory devices is the floating gate field effect transistor. A most brilliant invention of the 70s of the 20th century. Its difference from conventional field-effect transistors is that between the gate and the channel, right in the dielectric, there is another conductor - which is called a floating gate. Here's what it all looks like:

In the figure we see the usual drain-source-gate, as well as an additional conductor located in the dielectric. Let's figure out how this device works.

Let's create a potential difference between the drain and source and apply a positive potential to the gate. What will happen then? That's right, current will flow through the field-effect transistor from drain to source. Moreover, the current is large enough to “pierce” the dielectric. As a result of this breakdown, some of the electrons will fall on the floating gate. A negatively charged floating gate creates an electric field that begins to impede the flow of current in the channel, causing the transistor to turn off. And if you turn off the power to the transistor, the electrons from the floating gate will not go anywhere and its charge will remain unchanged for many years.

But of course there is a way to discharge the floating bolt. To do this, you just need to apply a voltage of the opposite sign to the “main” gate, which will “drive” all the electrons, as a result of which the floating gate will remain uncharged.

This is actually how information is stored - if there is a negative charge on the gate, then this state is considered a logical one, and if there is no charge, then it is a logical zero.

We have sorted out the storage of information, all that remains is to understand how we can read information from a floating-gate transistor. And everything is very simple. When there is a charge on a floating gate, its electric field prevents drain current from flowing. Suppose, in the absence of charge, we could apply a voltage of +5V to the “main” gate, and at the same time current began to flow in the drain circuit. When the floating gate is charged, such a voltage will not be able to cause current to flow, since the electric field of the floating gate will interfere with it. In this case, the current will flow only at a voltage of +10V (for example =)). This gives us two voltage thresholds. And, by applying, for example, +7.5V, we can, based on the presence or absence of drain current, draw a conclusion about the presence or absence of charge on the floating gate. This is how the stored information is read.

How does all this relate to Flash memory? And it’s very simple - a field-effect transistor with a floating gate is the minimum memory cell capable of storing one bit of information. And any memory chip consists of a huge number of transistors arranged in a certain way. And now it’s time to look at the main types of Flash memory. Namely, I would like to discuss NOR and NAND memory.

Both of these types of memory are built on the basis of floating-gate transistors, which we spent a lot of time on today) And the fundamental difference is how these transistors are connected.

The NOR design uses a two-dimensional conductor table. The conductors are called bit line and word line. All transistor drains are connected to the bit line, and all gates are connected to the word line. Let's look at an example for better understanding.

Suppose we need to read information from a specific cell. This cell, or rather this particular transistor, is connected with the gate to one of the word lines, and the drain to one of the bit lines. Then we simply apply a threshold voltage to the word line corresponding to the gate of our transistor and read its state as in the example that we looked at just above for one cell.

With NAND everything is somewhat more complicated. If we return to the array analogy, NAND memory cells are a three-dimensional array. That is, not one, but several transistors are connected to each bit line at once, which ultimately leads to a reduction in the number of conductors and an increase in compactness. This is precisely one of the main advantages of NAND memory. But how can we calculate the state of a certain transistor with such a structure? To understand the process, consider the diagram:

As can be seen from the diagram, one bit line corresponds to several cells. And an important feature is the following: if at least one of the transistors is closed, then there will be a high voltage on the bit line. Look here:

Indeed, a low level on the bit line will only occur when the entire chain of transistors is open (remember the course on field-effect transistors 😉).

With this seemingly clear, we return to our question - how to calculate the state of a specific transistor? And to do this, it is not enough to simply apply a threshold voltage to the word line (to the transistor gate) and monitor the signal on the bit line. It is also necessary that all other transistors are in the open state. And this is done this way: a threshold voltage is applied to the gate of our transistor, the state of which we need to read (as in the case of NOR memory), and an increased voltage is applied to the gates of all other transistors in this chain, such that, regardless of the state of the floating gate the transistor opened. And then, by reading the signal from the bit line, we find out in what state the transistor we are interested in is in (after all, all the others are absolutely open). That's all)

This is how the article turned out today) We figured out the operating principle and the main types of Flash, as well as the structure and operating principle of NAND and NOR memory. I hope that the article will be useful and understandable, see you soon!

Perhaps many people noticed when viewing the characteristics of their drive that its capacity does not reach that specified by the manufacturer. This applies not only to the capacity of flash drives, but to all digital media: hard drives and others in which the capacity is measured in Megabytes, Gigabytes and, in the latest devices, Terabytes.

What is the matter here and is there a deception hidden in this? It so happened that drive manufacturers, in general, like manufacturers of other products, want to sell “candy” with a beautiful inscription (capacity) for less money. To win the competition. But the capacity indicated on the drive is true, but on the one hand.

So why does a 2 GB flash drive actually only have 1.86 GB, and a 4 GB only 3.72 GB.

The answer to this question follows from the basics of computer technology, namely: 1 kilobyte contains 1024 bytes and so on with megabytes, gigabytes...

real capacity ( http://www.ixbt.com/storage/flashdrives/svodka/size.shtml) is slightly different.

As a result, making a simple calculation: 4,000,000,0000/1024/1024/1024 = 3.72; we get the figure 3.72 GB.

For larger capacity drives, the absolute deviation will be greater. For example, for a 1 Terabyte hard drive, the actual capacity will be 931 GB.

In addition, the usable capacity of the drive depends on the selected file system: FAT16, FAT32, NTFS. Media formatted on different systems will have different usable capacity. This is due to the fact that when a disk is formatted, system information about it is written to it and it is different for different file systems.

Well, one last thing. There is such a phenomenon as a Chinese flash drive: this is when information is deliberately entered into the system section of a small-capacity flash drive that its capacity is large. For example, from 1 GB you can make 32 GB. In practice, if you insert this flash drive into your computer, it will show that its capacity is 32 GB. When the user writes data to it in a volume greater than its actual volume, the copying will complete without errors. But it will be possible to read data from such a medium in an amount commensurate with the actual volume, i.e. no more than 1 GB for our example.

Flash memory belongs to the EEPROM class, but uses a special technology for constructing storage cells. Erasing in flash memory is performed immediately for an entire area of cells (in blocks or the entire chip). This made it possible to significantly increase productivity in recording (programming) mode. Flash memory has a combination of high packaging density (its cells are 30% smaller than DRAM cells), non-volatile storage, electrical erase and write, low consumption, high reliability and low cost... These are reprogrammable memories.

Like RAM, Flash memory is electrically modified in-system, but like ROM, flash is non-volatile and stores data even after power is turned off. However, unlike RAM, Flash cannot be rewritten byte by byte. Flash memory is read and written byte by byte and has a new requirement: it must be erased before writing new data.

Flash memory is semiconductor memory, and a special type. Her unit cell, which stores one bit of information, is not a capacitor, but field-effect transistor with a special electrically isolated region called a "floating gate". An electrical charge placed in this area can persist for many years. When writing one bit of data, the cell is charged - the charge is placed on the floating gate, when erasing - the charge is removed from the floating gate and the cell is discharged.

Among such devices, circuits with specialized blocks (asymmetric block structures) are distinguished. By the name of the so-called Boot blocks in which information is reliably protected from accidental erasure, memories are called Boot Block Flash Memory.

Flash memory Boot block type serves to store updated programs and data in a wide variety of systems, including cell phones, modems, BIOS, car engine management systems and much more. By using flash memory instead of EEPROM to store parametric data, designers can reduce the cost and improve the reliability of their systems.

Advantages of flash memory compared to EEPROM:
1.

Higher write speed for sequential access due to the fact that erasing information in flash is done in blocks.
2. The production cost of flash memory is lower due to its simpler organization.
Flaw: Slow writing to random memory locations.

Memory with sequential access Used where data can be queued.

Addressable Flash Memory. Storing rarely changed data. Recording and erasing is carried out by the processor of the computing device in normal operating mode. For this purpose, Flash memory has additional command word control , written by the processor to a special register on the chip. When a special programming voltage is applied, the circuit provides recording and erasing of information. Before programming, the processor reads a code from the microcircuit - an identifier containing the code of the manufacturer and the microcircuit to coordinate erase and write algorithms automatically.

All bytes of memory or the selected block are erased, after which they are all checked, erased and checked again.

Memory programming is carried out byte by byte, the recorded information is checked. The processor reads the written byte from the memory and compares it with the original one.

One of the blocks is designed to store BIOS software and is hardware protected from accidental erasure.

Operating principle and design of flash memory

The memory also contains parameter blocks and main blocks that are not protected from accidental erasure. Main blocks store the main control programs, and parameter blocks store relatively frequently changed system parameters.

File Flash Memory used to replace hard drives. Reduces power consumption, increases memory reliability, reduces their size and weight, and increases performance when reading data. The program can be read by the processor directly from the file Flash memory, and the results are also written there.

Compact removable external storage devices are created based on file Flash memory.

ZE – MNOP.

2 threshold voltages. Upor1 – has a small value, 1-2 V. When Upor is applied, the drain-source m/d channel is initiated. If the m/d of nitride and silicon dioxide have charges, then Upore has increased to 7V.

Writing (programming) flash memory– the process of replacing 1 with 0. Erasing– replacing 0 with 1.

3.RS architecture. Computer processors. The structure of processors and their main characteristics. System buses and their characteristics. Local buses. Chipsets.
Architecture is a multi-level hierarchy of hardware and software, each level allows for multiple construction and application.

Structure is a collection of elements and their connections.

A computer is a complex of hardware and software designed to automate the preparation and solution of user tasks.

Computer architecture- this is a general description of the structure and functions of a computer at a level sufficient to understand the principles of operation and the computer command system, which does not include details of the technical and physical structure of the computer.

The architecture includes the following principles of computer construction:

1. computer memory structure;
2. methods of accessing memory and external devices;
3. ability to change configuration;
4. command system;
5. data formats;
6. interface organization.

The architecture of modern personal computers is based on backbone-modular principle. Information communication between computer devices is carried out through system bus(another name is system highway).

A bus is a cable consisting of many conductors. One group of conductors each - data bus processed information is transmitted, on the other - address bus— addresses of memory or external devices accessed by the processor. The third part of the highway - control bus, control signals are transmitted through it (for example, a signal that the device is ready for operation, a signal to start operation of the device, etc.).

The system bus is characterized clock frequency and bit depth. The number of bits simultaneously transmitted on the bus is called bus width. Clock frequency characterizes the number of elementary data transfer operations in 1 second. The bus width is measured in bits, the clock frequency is measured in megahertz.
System buses

Transfer of information between MP and other elements. Devices are also addressed and special service signals are exchanged. The transmission of information over the bus is controlled by one of the devices connected to it or a node specially dedicated for this purpose, called a bus arbiter.

ISA bus(Industry Standard Architecture) there is a 36-pin connector for expansion cards. Due to this, the number of address lines is 4, and the number of data is 8. It is possible to transmit 16 bits of data in parallel, and thanks to 24 address lines, directly access 16 MB of memory systems. Number of hardware interrupt lines - 15.

EISA bus(Extended ISA). provides the largest possible amount of addressable memory, 32-bit data transfer, an improved interrupt system, automatic configuration of the system and expansion cards. The EISA connector on the computer's system board is ISA compatible. The EISA bus allows you to address 4GB of address space. Theoretically, the maximum speed is 33 MB/s. The bus is clocked at a frequency of about 8-10 MHz.

Local buses are designed to increase the speed of the computer, allowing peripheral devices (video adapters, storage controllers) to operate at clock frequencies of up to 33 MHz and higher. The connector is MCA type.

PCI buses. Between the local processor bus and PCI itself there is a special matching circuit

According to the PCI specification, up to 10 devices can be connected to the bus. The PCI bus operates at a fixed clock frequency of 33 MHz and provides both 5 and 3.3 V supply voltages for controllers, plug and play mode.

PCI-X bus – high performance PCI. is synchronous, i.e. all data is processed simultaneously upon receipt of a control signal. The bus width is 32-bit. At 33 MHz, the theoretical throughput is 132 MB/s.

Any information transmitted from the processor to other devices via the data bus is accompanied by address transmitted over the address bus. This can be the address of a memory cell or the address of a peripheral device. It is necessary that the bus width allows the address of the memory cell to be transmitted. Thus, in words, the bus width limits the amount of computer RAM; it cannot be greater than , where n is the bus width.

diagram of a computer built on the backbone principle

Chipset- from English "chip set" is a set of microcircuits designed to work together to perform a set of functions. Thus, in computers, the chipset acts as a connecting component that ensures the joint functioning of the memory, CPU, input-output and other subsystems. Chipsets are also found in other devices, for example, in radio units of cell phones.

The chipset of computer motherboards consists of two main chips (sometimes they are combined into one chip):

MCH - Memory Controller Hub - northbridge - ensures interaction between the central processing unit (CPU) and the memory and video adapter. New chipsets often have an integrated video subsystem.
The memory controller can be integrated into the processor (eg Opteron, Nehalem, UltraSPARC T1).
ICH - I/O Controller Hub - southbridge - provides interaction between the CPU and hard drive, PCI cards, IDE, SATA, USB interfaces, etc.

Also sometimes chipsets include a Super I/O chip, which connects to the south bridge and is responsible for low-speed RS232, LPT, PS/2 ports.

Currently, the main manufacturers of chipsets for desktop computers are companies Intel, nVidia, AMD(which acquired ATI and currently produces chipsets under its own name), VIA And SIS.

Firm Intel produces chipsets only for its own processors. For company processors AMD the most common are chipsets nVidia(usually produced under the brand name nForce) and AMD.

Chipsets of companies VIA And SIS They are popular mainly in the low-end sector, as well as in office systems, although their integrated graphics are significantly inferior to nVidia and AMD in terms of 3D capabilities.

⇐ Previous12345678910Next ⇒

Studopedia.org - Studopedia.Org - 2014-2018 (0.004 s)…

Comparison of performance of different types of server drives (HDD, SSD, SATA DOM, eUSB)

In this article we will look at modern server drive models from the point of view of performance and optimal areas of application.

At the moment, servers mainly use two types of data storage devices - hard magnetic disks (HDD, hard disk drive) and solid-state drives (SSD, solid-state drive). In addition, devices such as eUSB Flash Module and SATA DOM are also used. Let's look at all these types in more detail.

Modern magnetic hard drives can use one of two interfaces - SATA (Serial Advanced Technology Attachment) and SAS (Serial Attached SCSI). The current version of the SATA interface provides 6 Gbps throughput. Disks with this interface are primarily used in the segment of desktop personal computers, but can also be used in servers. In the server segment, such drives have a spindle speed of 7,200 rpm. The models that will take part in our testing of this type of drive are Seagate Constellation.2 ST91000640NS (SATA 7’200, 2.5″) and Seagate Constellation ES ST1000NM0011 (SATA 7’200, 3.5″).

A more reliable and productive SAS disk interface is designed for server solutions and workstations. It also has a throughput of up to 6 Gbps, but in Full Duplex mode, which means it can simultaneously transmit data in both directions at a speed of 6 Gbps. Disks with this interface have a higher MTBF (Mean Time Between Failures). Moreover, the SAS interface, unlike SATA, uses a different set of commands with support for a larger request queue depth (64 versus 32, the greater the queue depth, the better the optimization of request execution queue) and a dual-port connection for possible fault tolerance. An important feature of SAS is a more tailored connection of disks with the SAS interface to various backplanes, baskets, expanders, RAID and HBA controllers, storage systems and other devices, both via internal and external ports. Currently, servers use SAS disks with spindle speeds of 7’200, 10’000 and 15’000 rpm.

Speed 7’200 rpm. at first it was atypical for the server segment, but hard drive manufacturers at some point decided to produce drives with a rotation speed of 7,200 rpm not only with the SATA interface, but also with the SAS interface. In their “mechanical” part, these drives are exactly the same, they differ only in the connection method. This move increased the affordability of SAS drives and provided the server segment with larger capacity SAS drives. The main area of application for such drives is low-budget workstations and entry-level servers. The tested drives of this type are Seagate Constellation.2 ST91000640NS (SAS 7’200, 2.5″) and Seagate Constellation ES.3 ST1000NM0023 (SAS 7’200, 3.5″).

SAS disks with a spindle speed of 10,000 rpm are a good solution for powerful workstations and low-cost enterprise-class server solutions. The tested drive is Seagate Savvio 10K5 ST9900805SS (SAS 10000 2.5″).

SAS drives with a spindle speed of 15,000 rpm are the best choice for corporate servers, data centers (DPCs) and data storage systems (SDS). The tested drive is Seagate Cheetah 15K7 ST3300657SS (SAS 15000 3.5″).

The performance of the above drives on sequential and random read/write operations is shown in the following diagram.

At the same spindle speed and physical platter size, SAS drives are faster than SATA drives, which is explained by the higher linear data density of SAS drives compared to SATA drives.

On the other hand, SAS 7’200, 3.5” and SAS 10’000, 2.5” drives show almost identical results. This is explained by the fact that the advantage in rotation speed is compensated by the smaller physical size of the 2.5” disk platters, as a result of which, with the same linear data density, the linear speed of the heads relative to the platters is approximately the same.

In the random read test, which measures input/output operations per second (IOPS), 2.5" 7'200 RPM drives perform better than 3.5" drives of the same speed because the "small" drives have There is less movement of the head to the desired sector. SAS drives here again show better results compared to SATA drives, now due to better optimization of the order of execution of random requests thanks to support for a greater queue depth (64 for SAS versus 32 for SATA). The advantage of SAS 10,000 and 15,000 rpm drives is provided not only by the high spindle speed, but also by the fact that they have a more advanced head positioning mechanism with shorter access times.

SAS drives have the same advantage over SATA drives in random write operations as they do in read operations.

SSDs using non-volatile NAND-Flash memory have hundreds of times faster random read and write speeds than HDDs because SSDs do not need to move the magnetic head. In addition, SSDs have lower power consumption and no operating noise. But they also have disadvantages, namely: high cost and, compared to HDDs, relatively small volume. In the desktop PC segment, such drives are used in conjunction with HDDs according to a scheme where the operating system and the most necessary programs are installed on the SSD, and all other data is stored on the HDD. This approach significantly increases the speed of the computer without greatly increasing its cost. For testing, we chose an Intel 520 Series 240GB drive. This drive is recommended for use in desktop computers, laptops and workstations.

In the server segment, the situation with SSDs is significantly different. Placing large amounts of data on an SSD is quite expensive. But they can be successfully used for caching, when the SSD cache is used to store “hot” data, that is, data that is accessed most often. This gives a huge increase in the performance of the server's disk subsystem, especially for random access operations. The tested server SSD drive is Intel DC S3700 100GB.

When reading sequentially, desktop and server drives show almost identical results, but when writing sequentially, the server type SSD noticeably loses. This is due to the fact that the server drive uses memory that allows an order of magnitude greater number of rewrite cycles, but the write operations themselves are performed more slowly.

In random write operations, the lag is also significant, but this is caused by the need to provide a much larger write resource for server drives.

eUSB drives, like SSD drives, also use Flash modules to store data, but they are installed directly into the USB connector on the server motherboard. Such drives have a number of functional and other limitations due to the use of a USB port as an interface. A full version of Windows OS cannot be loaded from such a drive, and the interface speed (480 Mbit/s) is significantly lower than that of SATA (6 Gbit/s). The most optimal area for their use in servers is to use a small operating system as a bootloader, for example, the VMware ESXi hypervisor.

In thin clients, such drives are used to store an image of the Windows Embedded operating system. The tested drive is eUSB Transcend 4GB.

SATA DOM drives are more functional than eUSB drives. They are connected in the same way as SSD drives, to a SATA connector, but at the same time they “look” more like a USB drive than a hard drive.

Design and principle of operation of a flash drive

They are installed directly into SATA connectors on the motherboard of a computer or server. It is convenient when such a connector has built-in power, otherwise it has to be provided through an additional cable. Considering that these drives are connected to standard SATA connectors, the motherboard BIOS works with them as with regular HDD or SSD drives, which makes it possible to install a full bootable version of the Windows operating system on the SATA DOM. In a server, this frees up space in the disk subsystem basket, allowing it to be used for a RAID array disk. In addition, the SATA DOM drive is located inside the server platform, which prevents accidental removal of the disk with the installed OS. Such drives can be used in desktop and server segments, as well as in thin clients, installing any operating system or hypervisor for virtualization. The tested drive is SATA DOM Innodisk 8 GB.

The test results for eUSB-Flash and SATA DOM drives correspond to the performance of their interfaces. According to the USB 2.0 specification, the speed is regulated at 25 - 480 Mbit/s, and for SATA 3.0 - 6’000 Mbit/s, which already inclines the choice in favor of devices with the SATA interface. In the graph we see a 2.5 times superiority in sequential read and write operations of SATA DOM Innodisk over eUSB-Flash.

In the random read operations test, the situation does not change; SATA DOM is also in the lead. Random writing on both drives is equally at a very low level, but they are not intended for these operations.

The performance data for the best representatives of each drive type from our testing is shown in the following diagrams. The clear leader is the SSD from Intel.

We hope that our article will help you decide on the choice of a particular drive. And there really is plenty to choose from. A very large variety of drives are offered by manufacturers, but to achieve the best results you need to properly plan your needs and expectations from storage subsystems.

Measurements for HDD and SSD were carried out on the same Intel RS25DB080 controller. Testing was performed using the IOmeter program with the following parameters: controller and disk cache disabled, command queue depth - 256, Strip Size parameter - 256KB, data block size - 256KB for sequential operations and 4KB for random operations. The speed of sequential operations was measured in MB/s, random - in IOPS (input/output operations per second).

Server hardware department engineer Andrey Leontyev
03.06.13

The Taiwanese company Mach Xtreme Technology, specializing in high-performance components for computers and closely involved in the production of solid-state drives, has begun retail sales of a promising data storage solution called PCIe SSD MX-EXPRESS.

Flash memory. Past, present and future

The new product has a low-profile design, is characterized by the following overall dimensions: 152.5 x 19 x 69 mm, weighs 125 grams, connects to the computer via a PCI-Express 2.0 x2 slot, uses an as yet unnamed dual controller and is available in four versions in terms of volume: 128 GB, 256 GB, 512 GB and 1 TB.

The drive supports ROHS, CE and FCC certificates and does not require any drivers for installation in the system. Data transfer speeds vary depending on the capacity of the drives. Thus, for 512 GB and 1 TB solutions, the sequential read speed is 850 MB/s, and the write speed is 800 MB/s, the performance level is in the region of 100,000 IOPS, and the access time is 0.1 ms.

The MX-Express series drives have a huge service life of 2.5 million hours, can operate in ambient temperatures from zero to 70 degrees Celsius, and support TRIM, DuraClass, DuraWrite, RAISE and Garbage Collector. In addition, the new product comes with a low-profile PCI blank.

The 128 GB model will cost everyone 309.90 euros, 256 GB - 379.90 euros, 512 GB - 669.90 euros and 1 TB - 1449.90 euros. The manufacturer's quality guarantee for the devices is 2 years.