Comparison of new non-volatile memory technologies. Non-volatile RAM memory

Types of real memory and their main characteristics

One of critical devices The computer is the memory, or storage device (storage). According to the definition given in the book “Informatics in Concepts and Terms,” memory is “the functional part of a digital computer, intended for recording, storing and issuing information presented in digital form"At the same time, this definition includes both the memory itself and external storage devices (such as hard drives and floppy disks, magnetic tape, CD-ROM), which are better classified as information input/output devices. Thus under computer memory in the future we will only mean “the internal memory of the computer: RAM, ROM, cache memory and flash memory.” So, let's look at the classification internal memory computer.

The random access memory device is perhaps one of the very first devices of a computer. It was already present in the first generation of computer architecture, created in the early forties and fifties of the twentieth century. Over these fifty years, more than one generation has changed element base, on which the memory was built. For this reason, we present a classification of RAM by element base and design features. With some stretch, ROM can also be classified as RAM, if we consider it as a fast read-only memory.

The scheme of this classification is shown in the figure.

Rice. Classification of RAM.

As can be seen from the diagram, based on the safety of data when the power is turned off, RAM is divided into volatile and non-volatile.

Non-volatile chargers, first of all, include the class of all possible ferrite chargers. Further, UV erasable and electrically erasable PROMs (reprogrammable – flash memory) can be called conditionally non-volatile. The convention consists in a fairly long (tens of thousands of hours) but not infinite period of storage of recorded information in the memory data. The next class of non-volatile memory is made up of one-time programmable ROM. These ROMs can be supplied blank (all memory written with zeros or ones) and then electrically programmed once, or programmed during manufacturing (custom ROMs).

Volatile memory - all possible types of RAM for fast reading/writing. When the power is turned off, such a memory completely loses information, but has high performance. This class of real memories is divided into dynamic (with the utmost importance of information regeneration) and static (not requiring information regeneration). Now let's look at each class of memory in more detail.

The first generation computers were extremely unreliable in terms of their elemental base. Thus, the average time to failure for the ENIAC computer was 30 minutes. The counting speed was not comparable with the counting speed modern computers. For this reason, the requirements for storing data in computer memory in the event of a computer failure were stricter than the requirements for the speed of RAM. As a result, these computers used non-volatile memory.

Non-volatile memory made it possible to store data entered into it long time(up to one month) when the power is turned off. Most often, ferrite cores were used as non-volatile memory. They are made from special materials - ferrites. Ferrites are characterized by the fact that the hysteresis loop of the dependence of their magnetization on the external magnetic field is almost rectangular in nature.

Rice. B.1. Ferrite magnetization diagram.

As a result, the magnetization of this core changes abruptly (the position of the binary 0 or 1, see Figure B.1.) For this reason, by assembling the circuit shown in Figure B.2, it is practically assembled simplest element memory capacity of 1 bit. Memory on ferrite cores worked slowly and inefficiently: after all, reversing the magnetization of the core took time and spent a lot of time electrical energy. For this reason, with the improvement in the reliability of the computer element base, non-volatile memory began to be replaced by volatile memory - faster, more economical and cheaper. However, scientists different countries work is still underway to find fast volatile memory that could work in a computer for critical important applications, primarily the military.

Rice. B.2. Diagram of a memory element on ferrite cores.

Good afternoon friends!

Have you seen the mysterious abbreviation “NVRAM” that flashes on the monitor when you turn on the computer? NVRAM is one of the necessary for the computer“piece of hardware”, and we will now figure out what kind of animal this is and why it is needed.

We will also see how this thing developed and got smarter, and with it the whole computer got smarter. First, let's look at

What is non-volatile memory?

NVRAM (Non Volatile Random Access Memory) is the general name for non-volatile memory. Non-volatile memory is one in which the data is not erased when the power is turned off. In contrast, there is volatile memory, the data in which disappears when the power is turned off. Those. when power is supplied to the memory chip (or module), it “remembers” the data; when it stops being supplied, it “forgets” them.

The concept of “non-volatile” includes several types of memory. By the way, memory (both volatile and non-volatile) is available not only in the computer, but also in all computer-related and peripheral devices:

in printers - laser, and
in monitors,
in modems,
graphic cards, etc.

Even computers have both types of memory.

Both of them are packaged in a packageless microcircuit (“droplet”) coated with a compound.

This design - all “in one bottle” - is called a controller (from the English “control” - management) and is very widely used in electronics.

Types of non-volatile memory

One type of non-volatile memory is called ROM (Read Only Memory). In Russian-language literature, such memory is called ROM (read-only memory). Data is written into the microcircuit, which is also called the English term “chip” (chip, crystal), during manufacturing. They cannot be changed later.

Another type of non-volatile memory is PROM (Programmable ROM). The equivalent Russian term is PPZU (Programmable ROM). In such a microcircuit original condition all memory cells contain the same information (zeros or ones). By using special procedure programming, the necessary information is written into the cells.

This happened by burning out fusible links.

After recording, it was impossible to change the data in the cells.

Programmability provides flexibility in production and use. To write modified information into the microcircuit, you do not need to rebuild technological process production. The user (more precisely, the manufacturer electronic technology) writes down the information he needs.

But once-programmable memory is also not always good. You cannot modify the information “stitched” into the microcircuit; you need to change the microcircuit. This is not always convenient or possible. Therefore, multiple programmable microcircuits appeared. In the first products, information was erased by ultraviolet radiation, for which a special lamp was used.

Such microcircuits had a window covered with quartz glass, which transmitted UV radiation. But it was still inconvenient, and then they learned to erase and record information with an electrical signal. Such memory began to be called EEPROM (Electric Erasable PROM, EEPROM, electrically erasable programmable ROM).

Then a variety of it appeared - Flash (flash) memory, which received last years very widespread.

This is a microcircuit in a computer.

These are the now well-known “flash drives” (portable data storage devices), solid-state SSD drives (Solid State Drive), an alternative to electromechanical hard drives, memory cards used in cameras, etc.

Note that information in such drives can be overwritten a limited (albeit large) number of times.

Computer time problem

The first computers did not have an RTS (Real Time Clock) chip.

It was inconvenient, and then they started installing it.

The problem that arose with RTC at the very beginning was that the computer does not work 24 hours a day. It is turned on by the user at the beginning of the working day and turned off at the end of the day. While the computer was turned on, it “remembered” the time; as soon as it was turned off, it “forgot” the time.

Setting the time again every time would be very inconvenient. It would be inconvenient to renew other system settings(hard drive type, download source, etc.). Therefore, they came up with the idea of building an RTC chip into the common case, which remembered not only the time, but also all the settings BIOS Setup, and a power source - a battery of galvanic cells.

RTC memory cells were essentially (RAM). Such memory was also classified as non-volatile, since it did not depend on an external voltage source. It was non-volatile until the built-in battery ran out. Such memory was made on the basis of CMOS structures, therefore it consumed a very small current in static mode (storage mode), on the order of several microamps.

Therefore, the built-in battery lasted for several years. After which the entire module had to be replaced. There were designs of motherboards with a connector for such a module. And it was easy to replace it. But then technological progress continued its inexorable run. The number of chips on the motherboard decreased, and the degree of their integration increased.

In the end, we came to a chipset (chip set) consisting of 1-2 cases, which included almost all subsystems motherboard.

It was considered inappropriate to integrate a voltage source into the same housing (where a lot of things are already crammed).

This case has many pins. Installing it in the connector would complicate the design, increase its cost and reduce reliability.

Therefore, the power source (3 V lithium cell) began to be installed separately. This simplified and made the board cheaper, since now only the element needs to be changed, and not all at once. It should be noted that initially, as a source backup power Nickel-cadmium batteries were used.

After prolonged use they could leak. And the leaked electrolyte could damage the conductors of the motherboard. Modern lithium cells do not leak even when deeply discharged.

The technology has changed, but the name of the structure storing BIOS settings Setup remains the same - NVRAM. But now, in the strict sense, it is not energy independent. After all, its “energy independence” is ensured external source voltage.

Let us recall that the first sign that element 2032 has exhausted its resource is that the time and date are reset when the computer is turned on. The voltage of a fresh element is about 3.3 V. As it is depleted, the emf drops. And, as soon as it drops (approximately) less than 2.8 V, the structure storing the settings will “forget” them. Lithium cells cannot be charged.

What do the numbers on the lithium cell label mean?

In conclusion, we note that the first two digits of the element marking (20) determine its diameter in millimeters.

The second two are its capacity (the ability to deliver a certain amount of energy).

How higher figure, the larger the capacity and the thicker the element. The typical capacity value of element 2032 is 225 mAh (milliamp-hours), element 2025 is 160 mAh.

It should be noted that these are maximum values. Real numbers depend on load resistance and ambient temperature. The greater the load resistance and the higher the temperature (of course, up to certain limits), the greater the equivalent capacitance. Those. the longer the element will supply energy to the load. At lower ambient temperatures, the element “shrinks” faster.

Lithium cells are very good sources energy.

They have high specific energy values, i.e. high energy/weight ratio and very low self-discharge (less than one percent per year). For lead, for example, these indicators are much worse.

Victor Geronda was with you.

See you on the blog!

Introduction

Computer memory(information storage device, memory device) - part of a computer, physical device or a medium for storing data used in computation for a specified period of time. Memory, like CPU, has been a fixture of the computer since the 1940s. Memory in computing devices has a hierarchical structure and usually involves the use of several storage devices with different characteristics.

Any information can be measured in bits and therefore, regardless of what physical principles and in what number system it operates digital computer(binary, ternary, decimal, etc.), numbers, text information, images, sound, video and other types of data can be represented as sequences of bit strings or binary numbers. This allows the computer to manipulate data as long as the storage capacity is sufficient (for example, about one megabyte is needed to store the text of an average-sized novel).

It is necessary to distinguish between the classification of memory and the classification of storage devices (storage devices). The first classifies memory by functionality, the second - by technical implementation. The first one is considered here - thus, it includes both hardware types of memory (implemented in memory) and data structures, implemented in most cases in software.

Based on recording stability and rewriting capabilities, memories are divided into:

· Read-only memories (ROMs), the contents of which cannot be changed by the end user (for example, BIOS). ROM in operating mode allows only reading information.

· Writeable memory (PROM), in which end user can only write information once (eg CD-R).

· Re-writable memory (PROM) (for example, CD-RW).

· Random access memory (RAM) provides a mode for recording, storing and reading information during its processing. Fast but expensive RAM (SRAM) is built on flip-flops, slower but cheaper types of RAM - dynamic memory (DRAM) are built on capacitors. In both types of memory, the information disappears after disconnection from the current source.

According to the type of access, storage devices are divided into:

Devices with sequential access(for example, magnetic tapes).

· Random access (RAM) devices (eg, random access memory).

Devices with direct access (e.g. hard magnetic disks).

Devices with associative access(special devices to improve productivity.)

Theoretical part

Non-volatile memory

Non-volatile memory(eng. NVRAM, from Non Volatile Random Access Memory) - rewritable or random access memory in electronic device, which retains its contents regardless of the supply of main power to the device.

In more in a general sense, non-volatile memory-- any device or part thereof that stores data regardless of the supply voltage. However, storage media that fall under this definition, ROM, PROM, devices with movable storage media (disks, tapes) and others have their own, more precise names.

Therefore, the term “non-volatile memory” is most often used more narrowly, in relation to such electronic memory, which usually runs volatile, and the contents of which usually disappear when turned off.

Non-volatile device- any device as part of a complex, device, computer system, which does not require connection to a common power source in this complex for its operation. For example:

· autonomous lamps emergency lighting;

· Clock (CMOS Clock) on system board personal computer;

Classification by device

Read Only Memory (ROM, English ROM- Read-Only Memory) - non-volatile memory, used to store an array of immutable data.

RAM(Also random access memory, RAM) - part of the computer memory system, which the processor can access in one operation (jump, move, etc.). Designed for temporary storage of data and commands, required by the processor to perform operations. RAM transfers data to the processor directly or through cache memory. Each RAM cell has its own individual address.

RAM can be manufactured as a separate unit, or included in the design of a single-chip computer or microcontroller.

Ferroelectric (FRAM)

Ferroelectric memory FRAM (Ferroelectric RAM) is a static random access memory, the cells of which store information using the ferroelectric effect (“ferroelectric” is translated as “ferroelectric, ferroelectric”, and not “ferroelectric”, as you might think). The memory cell consists of two conductive plates and a film of ferroelectric material. At the center of a ferroelectric crystal there is a mobile atom.

The application of an electric field causes it to move. If the field “tries” to move the atom to a position, for example, corresponding to a logical zero, and it is already there, less charge passes through the ferroelectric capacitor than in the case of switching the cell. Reading is based on measuring the charge passing through the cell.

During this process, the cells are overwritten and information is lost (regeneration is required). Research in this direction is carried out by Hitachi together with Ramtron, Matsushita and Symetrix. Compared to flash memory, FRAM cells practically do not degrade - up to 10 10 rewrite cycles are guaranteed.

Memory implemented by a memory device, the records in which are erased when the power supply is removed. This type of memory includes memory implemented in RAM and cache memory.

Classification of RAM

Types of real memory and their main characteristics

One of the most important devices of a computer is memory, or storage device (storage). According to the definition given in the book “Computer Science in Concepts and Terms,” a memory is “a functional part of a digital computer designed for recording, storing and issuing information presented in digital form.” However, this definition includes both memory itself and external storage devices (such as hard and floppy drives, magnetic tape, CD-ROM), which are better classified as input/output devices. Thus, in the future, computer memory will only be understood as “the internal memory of the computer: RAM, ROM, cache memory and flash memory.” So, let's look at the classification of the internal memory of a computer.

The random access memory device is perhaps one of the very first devices of a computer. It was already present in the first generation of computer architecture, created in the early forties and fifties of the twentieth century. Over these fifty years, more than one generation of the elemental base on which the memory was built has changed. Therefore, we present a classification of RAM according to its element base and design features. With some stretch, ROM can also be classified as RAM if we consider it as a fast read-only memory.

The scheme of this classification is shown in the figure.

Rice. Classification of RAM.

As can be seen from the diagram, depending on the safety of data when the power is turned off, RAM is divided into volatile and non-volatile.

Non-volatile chargers, first of all, include the class of all kinds of ferrite chargers. Further, UV erasable and electrically erasable PROMs (reprogrammable – flash memory) can be called conditionally non-volatile. The convention lies in the fairly long (tens of thousands of hours) but not infinite period of storage of recorded information in the memory data. The next class of non-volatile memory is made up of one-time programmable ROM. These ROMs can be supplied blank (all memory written with zeros or ones) and then electrically programmed once, or programmed during manufacturing (custom ROMs).

Volatile memory is all kinds of RAM for fast read/write. When the power is turned off, such a memory completely loses information, but has high performance. This class of real memories is divided into dynamic (with the need for information regeneration) and static (without requiring information regeneration). Now let's look at each class of memory in more detail.

The first generation computers were extremely unreliable in terms of their elemental base. Thus, the average time to failure for the ENIAC computer was 30 minutes. The counting speed was not comparable to the counting speed of modern computers. Therefore, the requirements for storing data in computer memory in the event of a computer failure were stricter than the requirements for the speed of RAM. As a result, these computers used non-volatile memory.

Non-volatile memory made it possible to store data entered into it for a long time (up to one month) when the power was turned off. Ferrite cores were most often used as non-volatile memory. They are a torus made of special materials - ferrites. Ferrites are characterized by the fact that the hysteresis loop of the dependence of their magnetization on the external magnetic field is almost rectangular in nature.

Rice. B.1. Ferrite magnetization diagram.

As a result, the magnetization of this core changes abruptly (the position of binary 0 or 1, see Figure B.1.) Therefore, by assembling the circuit shown in Figure B.2, the simplest memory element with a capacity of 1 bit has been practically assembled. Memory on ferrite cores worked slowly and inefficiently: reversing the magnetization of the core required time and consumed a lot of electrical energy. Therefore, with the improvement in the reliability of the computer element base, non-volatile memory began to be replaced by volatile memory - faster, more economical and cheaper. However, scientists from different countries are still working to find fast, volatile memory that could work in computers for critical applications, primarily military ones.

Rice. B.2. Diagram of a memory element on ferrite cores.

Flash memory is a special type of non-volatile rewritable semiconductor memory.

Non-volatile - does not require additional energy to store data (energy is required only for recording).

Rewritable - allowing the data stored in it to be changed (overwritten).

Semiconductor (solid-state) - does not contain mechanically moving parts (like conventional hard disks or CD), built on the basis integrated circuits(IC-Chip).

Unlike many other types of semiconductor memory, a flash memory cell does not contain capacitors - a typical flash memory cell consists of just one transistor of a special architecture. A flash memory cell is highly scalable, which is achieved not only due to advances in miniaturization of transistor sizes, but also thanks to design innovations that allow several bits of information to be stored in one flash memory cell.

Flash memory historically comes from ROM (Read Only Memory) memory, and functions similar to RAM (Random Access Memory). Flash stores data in memory cells similar to cells in DRAM. Unlike DRAM, data in flash memory is not lost when the power is turned off.

Replacement of SRAM and DRAM memory with flash memory does not occur due to two features of flash memory: flash is significantly slower and has a limit on the number of rewrite cycles (from 10,000 to 1,000,000 for different types).

ROM (Read Only Memory) - read-only memory. The Russian equivalent is ROM (Read Only Memory). To be completely precise, this type memory is called Mask-ROM (Mask ROM). The data on the ROM was written during production by applying a mask (hence the name) of aluminum connecting tracks using a lithographic method. The presence or absence of such a track in the corresponding place was coded “0” or “1”. Mask-ROM is characterized by the difficulty of modifying the contents (only by manufacturing new chips), as well as the length of the production cycle (4-8 weeks). Therefore, and also due to the fact that modern software often has many shortcomings and often requires updating; this type of memory is not widely used.

Advantages:

1. Low cost of a finished programmed microcircuit (for large production volumes).

2. High speed access to a memory cell.

3. High reliability finished microcircuit and resistance to electromagnetic fields.

Flaws:

1. Inability to record and modify data after production.

2. Complex production cycle.

PROM

PROM - (Programmable ROM), or one-time programmable ROM. As memory cells in this type memory fusible jumpers were used. Unlike Mask-ROM, PROM now has the ability to encode ("burn through") cells if there are special device for recording (programmer). Programming the cell in PROM is carried out by destroying ("burning") the fusible jumper by applying current high voltage. Opportunity self-recording information in them made them suitable for piece and small-scale production. PROM almost completely fell out of use in the late 80s.

Advantages:

1. High reliability of the finished microcircuit and resistance to electromagnetic fields.

2. The ability to program a finished microcircuit, which is convenient for piece and small-scale production.

3. High speed of access to memory cells.

Flaws:

1. Inability to rewrite

2. Big percentage marriage

3. The need for special long-term thermal training, without which the reliability of data storage would be low

EPROM

Different sources decipher the abbreviation EPROM differently - as Erasable Programmable ROM or Electrically Programmable ROM (erasable programmable ROM or electrically programmable ROM). In EPROM, before writing, it is necessary to erase it (accordingly, it became possible to overwrite the contents of the memory). Erasing EPROM cells is performed on the entire chip at once by irradiating the chip with ultraviolet or x-rays for several minutes. Microcircuits, which are erased by exposure to ultraviolet light, were developed by Intel in 1971, and are called UV-EPROM (prefix UV (Ultraviolet) - ultraviolet). They contain quartz glass windows, which are sealed after the erasing process is completed.

In EPROM, erasing brings all the bits of the erased area to the same state (usually all ones, less often all zeros). Writing to EPROM, as well as to PROM, is also carried out using programmers (however different from programmers for PROM). Currently, EPROM has been almost completely displaced from the EEPROM and Flash market.

Advantage: Ability to overwrite the contents of the chip

Flaws:

1. Not a large number of rewrite cycles.

2. Impossibility of modifying part of the stored data.

3. High probability of “under-rubbing” (which will ultimately lead to failures) or overexposing the microcircuit under UV light (the so-called overerase - the effect of excessive removal, “burning”), which can reduce the service life of the microcircuit and even lead to its complete disrepair.

EEPROM (E?PROM or Electronically EPROM) - electrically erasable PROMs. Home distinctive feature EEPROM (including Flash) from the types of non-volatile memory we previously considered is the possibility of reprogramming when connected to a standard system bus microprocessor device. EEPROM now has the ability to erase a single cell using electric current. For EEPROM, each cell is erased automatically when writing to it new information, i.e. you can change the data in any cell without affecting the others. The erasing procedure usually takes much longer than the writing procedure.

Advantages of EEPROM over EPROM:

1. Increased service life.

2. Easier to use.

Disadvantage: High cost

Flash (full historical name Flash Erase EEPROM)

Flash (full historical name Flash Erase EEPROM):

The invention of flash memory is often unfairly attributed to Intel, citing 1988. In fact, memory was first developed by Toshiba in 1984, and the following year it began production of 256Kbit flash memory chips in industrial scale. In 1988, Intel developed own version flash memory.

Flash memory uses a slightly different type of transistor cell than EEPROM. Technologically, flash memory is related to both EPROM and EEPROM. The main difference between flash memory and EEPROM is that erasing the contents of cells is performed either for the entire chip or for a specific block (cluster, frame or page). Regular size such a block is 256 or 512 bytes, but in some types of flash memory the block size can reach 256KB. It should be noted that there are microcircuits that allow you to work with blocks different sizes(to optimize performance). You can erase both the block and the contents of the entire microcircuit at once. Thus, in general case, in order to change one byte, first the entire block containing the byte to be changed is read into the buffer, the contents of the block are erased, the value of the byte in the buffer is changed, and then the block changed in the buffer is written. This scheme significantly reduces the write speed small volumes data into arbitrary memory areas, however, it significantly increases performance when writing data sequentially in large portions.