The concept of flash memory

Flash memory is a long-life non-volatile memory that can retain stored data information even when the power is off. Data deletion is not in units of single bytes but in fixed blocks. The block size is generally 256KB to 20MB. Flash memory is a variant of electronically erasable read-only memory (EEPROM). The difference between EEPROM and flash memory is that it can be deleted and rewritten at the byte level instead of erasing the entire chip, so that flash memory can be updated faster than EEPROM. . Because it can still save data when powered off, flash memory is usually used to save setup information, such as saving data in a computer's BIOS (Basic Input Output Program), PDA (Personal Digital Assistant), digital camera, etc. On the other hand, flash memory does not rewrite data in bytes like RAM (random access memory), so it cannot replace RAM.

Flash Card is a memory that uses Flash Memory technology to store electronic information. It is generally used as a storage medium in digital cameras, handheld computers, MP3 and other small digital products, so it looks compact. , like a card, so it is called a flash card. According to different manufacturers and different applications, flash memory cards include SmartMedia (SM card), Compact Flash (CF card), MultiMediaCard (MMC card), Secure Digital (SD card), Memory Stick (memory stick), and XD-Picture Although flash memory cards such as XD card and MICRODRIVE have different appearances and specifications, the technical principles are the same.

Technology and Features

There is a big difference between NOR type and NAND type flash memory. For example, NOR type flash memory is more like memory, with independent address lines and data lines, but The price is more expensive and the capacity is smaller; while the NAND type is more like a hard disk, the address line and data line are the most commonly used I/O lines. All information similar to a hard disk is transmitted through a hard disk line, and the NAND type and the NOR type are generally Compared with flash memory, the cost is lower and the capacity is much larger. Therefore, NOR flash memory is more suitable for frequent random reading and writing. It is usually used to store program codes and run directly in flash memory. Mobile phones are the largest users of NOR flash memory, so the "memory" capacity of mobile phones is usually not large; NAND flash memory It is mainly used to store data. Our commonly used flash memory products, such as flash drives and digital memory cards, all use NAND flash memory.

Here we also need to correct a concept, that is, the speed of flash memory is actually very limited. Its operating speed and frequency are much lower than that of memory, and NAND flash memory operates similarly to a hard disk and is more efficient than memory. The direct access method is much slower. Therefore, do not think that the performance bottleneck of the flash drive is the interface, or even take it for granted that the flash drive will get a huge performance improvement after adopting the USB2.0 interface.

As mentioned earlier, the operating efficiency of NAND flash memory is low. This is related to its architecture design and interface design. It does operate like a hard disk (in fact, NAND flash memory did consider the operation efficiency at the beginning of its design). Hard drive compatibility), its performance characteristics are also very similar to those of a hard drive: small data block operations are very slow, while large data block operations are very fast. This difference is much greater than other storage media. This kind of performance characteristics is very worthy of our attention.

Flash memory access is relatively fast, noiseless, and has low heat dissipation. If you buy it, you don’t have to think so much and just buy flash memory with the same storage space. If the hard disk space is large, then buy a hard disk, which can also meet the needs of your application.

Classification of flash memory

·Common storage currently on the market can be divided into categories:

U disk

CF card

SM card

SD/MMC card

Memory stick

·Common brands in the domestic market are:

Kingston, Sony, Sandisk, Kingmax, Yingtai, Transcend.

NAND flash memory

The basic storage unit of memory and NOR flash memory is bit, and users can randomly access the information of any bit. The basic storage unit of NAND flash memory is page (as you can see, a page of NAND flash memory is similar to a sector of a hard disk, and a sector of a hard disk is also 512 bytes). The effective capacity of each page is a multiple of 512 bytes. The so-called effective capacity refers to the part used for data storage. In fact, 16 bytes of check information are added, so we can see the expression "(512+16) Byte" in the flash memory manufacturer's technical information. . At present, most NAND flash memories with a capacity below 2Gb have a page capacity of (512+16) bytes, while NAND flash memories with a capacity above 2Gb expand the page capacity to (2048+64) bytes.

NAND flash memory performs erase operations in block units.

The write operation of flash memory must be performed in a blank area. If there is already data in the target area, it must be erased first and then written. Therefore, the erase operation is a basic operation of flash memory. Generally, each block contains 32 512-byte pages, with a capacity of 16KB; when large-capacity flash memory uses 2KB pages, each block contains 64 pages, with a capacity of 128KB.

Each NAND flash memory generally has 8 I/O interfaces. Each data line transmits (512+16) bits of information each time. 8 I/O interfaces are (512+16) × 8bit, that is The 512 bytes mentioned earlier. However, larger-capacity NAND flash memories are increasingly adopting 16 I/O line designs. For example, Samsung’s chip number K9K1G16U0A is a 64M×16bit NAND flash memory with a capacity of 1Gb and the basic data unit is (256+8 )×16bit, or 512 bytes.

When addressing, NAND flash memory transmits address information packets through 8 I/O interface data lines, and each packet transmits 8-bit address information. Due to the relatively large capacity of the flash memory chip, a set of 8-bit addresses is only enough to address 256 pages, which is obviously not enough. Therefore, usually one address transfer needs to be divided into several groups and takes up several clock cycles. The address information of NAND includes the column address (the starting operation address in the page), the block address and the corresponding page address. They are grouped separately during transmission, which takes at least three times and takes up three cycles. As the capacity increases, more address information will take up more clock cycles to transmit. Therefore, an important feature of NAND flash memory is that the larger the capacity, the longer the addressing time. Moreover, because the transfer address cycle is longer than other storage media, NAND flash memory is less suitable for a large number of small-capacity read and write requests than other storage media.

What are the factors that determine NAND flash memory?

1． Number of pages

As mentioned earlier, the larger the capacity of the flash memory, the more pages, the larger the pages, and the longer the addressing time. But this extension of time is not a linear relationship, but changes step by step. For example, 128 and 256Mb chips require 3 cycles to transmit address signals, 512Mb and 1Gb chips require 4 cycles, and 2 and 4Gb chips require 5 cycles.

2. Page capacity

The capacity of each page determines the amount of data that can be transferred at one time, so pages with larger capacity have better performance. As mentioned earlier, large-capacity flash memory (4Gb) increases the page capacity from 512 bytes to 2KB. Increasing the page capacity not only easily increases the capacity, but also improves the transmission performance. We can give an example. Take Samsung K9K1G08U0M and K9K4G08U0M as examples. The former has a 1Gb, 512-byte page capacity, a random read (stable) time of 12μs, and a write time of 200μs; the latter has a 4Gb, 2KB page capacity, a random read (stable) time of 25μs, and a write time of 25μs. is 300μs. Assume they operate at 20MHz.

Reading performance: The reading steps of NAND flash memory are divided into: sending commands and addressing information → transferring data to the page register (random read stabilization time) → transferring data (8 bits per cycle, required Teleport 512+16 or 2K+64 times).

K9K1G08U0M needs to read a page: 5 commands, addressing cycle × 50ns + 12μs + (512 + 16) × 50ns = 38.7μs; K9K1G08U0M actual read transfer rate: 512 bytes ÷ 38.7μs = 13.2MB /s; K9K4G08U0M needs to read a page: 6 commands, addressing cycle × 50ns + 25μs + (2K + 64) × 50ns = 131.1μs; K9K4G08U0M actual read transfer rate: 2KB bytes ÷ 131.1μs = 15.6MB/s. Therefore, using a 2KB page capacity improves read performance by about 20% compared to a 512-byte page capacity.

Writing performance: The writing steps of NAND flash memory are divided into: sending addressing information → transferring data to the page register → sending command information → writing data from the register to the page. The command cycle is also one, and we will merge it with the addressing cycle below, but these two parts are not consecutive.

K9K1G08U0M needs to write a page: 5 commands, addressing cycle × 50ns + (512 + 16) × 50ns + 200μs = 226.7μs. K9K1G08U0M actual write transfer rate: 512 bytes ÷ 226.7μs = 2.2MB/s. K9K4G08U0M needs to write a page: 6 commands, addressing cycle × 50ns + (2K + 64) × 50ns + 300μs = 405.9μs. K9K4G08U0M actual write transfer rate: 2112 bytes/405.9μs=5MB/s. Therefore, using a 2KB page capacity improves write performance by more than twice that of a 512-byte page capacity.

3. Block capacity

Block is the basic unit of erase operation. Since the erase time of each block is almost the same (the erase operation generally takes 2ms, the time occupied by the commands and address information of the previous cycles can be ignored. Not counting), the capacity of the block will directly determine the erasure performance.

The page capacity of large-capacity NAND flash memory has increased, and the number of pages in each block has also increased. Generally, the block capacity of 4Gb chips is 2KB × 64 pages = 128KB, and that of 1Gb chips is 512 bytes × 32 pages = 16KB. It can be seen that within the same time, the wiping speed of the former is 8 times that of the latter!

4． I/O bit width

In the past, NAND flash memory usually had 8 data lines, but starting from 256Mb products, products with 16 data lines have appeared. However, due to reasons such as controllers, x16 chips are relatively rarely used in practice, but the number will still show an upward trend in the future. Although the x16 chip still uses 8-bit groups when transmitting data and address information, and the occupied cycle remains the same, it transmits data in 16-bit groups, doubling the bandwidth. K9K4G16U0M is a typical 64M×16 chip. Each page is still 2KB, but the structure is (1K+32)×16bit.

Imitating the above calculation, we get the following. K9K4G16U0M needs to read a page: 6 commands, addressing cycle × 50ns + 25μs + (1K + 32) × 50ns = 78.1μs. K9K4G16U0M actual read transfer rate: 2KB bytes ÷ 78.1μs = 26.2MB/s. K9K4G16U0M needs to write a page: 6 commands, addressing cycle ×50ns+(1K+32)×50ns+300μs=353.1μs. The actual write transfer rate of K9K4G16U0M: 2KB bytes ÷ 353.1μs = 5.8MB/s

It can be seen that for a chip with the same capacity, after increasing the number of data lines to 16, the read performance is improved by nearly 70%, and the write performance is improved by nearly 70%. Performance is also improved by 16%.

5． Frequency

The impact of operating frequency is easy to understand. The operating frequency of NAND flash memory is between 20 and 33MHz. The higher the frequency, the better the performance. When taking K9K4G08U0M as an example earlier, we assumed that the frequency is 20MHz. If we double the frequency to 40MHz, K9K4G08U0M needs to read a page: 6 commands, addressing cycle × 25ns + 25μs + (2K + 64) × 25ns = 78μs. K9K4G08U0M actual read transfer rate: 2KB bytes ÷ 78μs = 26.3MB/s. It can be seen that if the operating frequency of K9K4G08U0M is increased from 20MHz to 40MHz, the read performance can be improved by nearly 70%! Of course, the above example is just for convenience of calculation. In Samsung's actual product line, the one that can work at higher frequencies should be K9XXG08UXM, not K9XXG08U0M. The former's frequency can currently reach 33MHz.

6． Manufacturing process

Manufacturing process can affect the density of transistors and also have an impact on the time of some operations. For example, the write stability and read stability times mentioned earlier account for an important part of the time in our calculations, especially when writing. If you can reduce these times, you can further improve performance. Can 90nm manufacturing process improve performance? The answer is probably no! The current actual situation is that as storage density increases, the required read and write stabilization time shows an upward trend. This trend is reflected in the examples given in the previous calculations, otherwise the performance improvement of 4Gb chips will be even more obvious.

Generally speaking, although the addressing and operation time of large-capacity NAND flash memory chips will be slightly longer, as the page capacity increases, the effective transmission rate will still be larger. Large-capacity chips are in line with the market Demand trends for capacity, cost and performance. Increasing data lines and increasing frequency are the most effective ways to improve performance. However, due to the influence of process and physical factors such as command and address information occupying the operating cycle, as well as some fixed operating times (such as signal stabilization time, etc.), they will not Bringing year-on-year performance improvements.

1Page=（2K+64）Bytes; 1Block=（2K+64）B×64Pages=（128K+4K）Bytes; 1Device=（2K+64）B×64Pages×4096Blocks=4224Mbits

Among them: A0~11 address within the page, which can be understood as "column address".

A12～29 address the page and can be understood as "row address". For convenience, "column address" and "row address" are transmitted in two groups instead of directly combining them into one large group. Therefore, each group will have several data lines with no information transmission in the last cycle. Unused data lines remain low. The so-called "row address" and "column address" of NAND flash memory are not the definitions we are familiar with in DRAM and SRAM, but are just a relatively convenient way of expression. In order to facilitate understanding, we can make a cross-section of the above three-dimensional NAND flash memory chip architecture diagram in the vertical direction. It is more intuitive to apply the two-dimensional "row" and "column" concepts to this cross-section.

Applications and prospects

"USB flash drive" is the most obvious portrayal of flash memory entering daily life. In fact, flash memory has appeared in many electronic products long before USB flash drive. The traditional way of storing data is to use volatile storage of RAM, and the data will be lost when the battery is exhausted. Products using flash memory overcome this problem and make data storage more reliable. In addition to flash drives, flash memory is also used in BIOS, PDAs, digital cameras, voice recorders, mobile phones, digital TVs, game consoles and other electronic products in computers.

Dating back to 1998, USB flash drives entered the market. The interface has developed from USB1.0 to 2.0, and the speed has gradually increased. The popularity of U disks has also indirectly promoted the promotion of USB interfaces. Why are USB flash drives so popular among people?

Flash drives can be used to exchange data between computers. In terms of capacity, the capacity of the flash disk is optional from 16MB to 2GB, breaking through the limitations of the floppy drive of 1.44MB. In terms of reading and writing speed, the flash disk uses a USB interface, and the reading and writing speed is much higher than that of a floppy disk. In terms of stability, the flash drive does not have a mechanical reading and writing device, which prevents the mobile hard drive from being easily damaged by bumps, drops, etc. Some models of flash drives have encryption and other functions to make user use more personalized. Flash drives are compact and easier to carry. And it uses a hot-swappable USB interface, which is very convenient to use.

Currently, flash memory is developing in the direction of large capacity, low power consumption, and low cost. Compared with traditional hard drives, flash memory has high read and write speeds and low power consumption. Flash hard drives have already appeared on the market. As manufacturing processes improve and costs decrease, flash memory will appear more in daily life.

Differences from hard drives

In terms of storage media alone, flash memory is better than hard drives. But it does not mean that the sound quality is good, it refers to the data transmission speed and shock resistance (flash memory does not have shock resistance). It is not difficult to compare the advantages and disadvantages between the two. First, understand what digital is. After knowing what digital signals are, you should understand that digital signals are usually not interfered by storage media. (Ignore the bit errors of audio stream files. Hard disks and flash memory can be ignored in this aspect. Optical discs are different.) The data accuracy of hard disk and flash memory is very high. Under the same test conditions (same decoding and same output), the sound quality of both is definitely It's the same. For the Walkman, I agree with the flash memory type.

Advantages:

1. The flash memory of the Walkman is small. This does not mean that the integration level of flash memory will necessarily be high. The main reason why micro hard drives are so big is that micro hard drives cannot be made smaller than flash memory. This does not mean that micro hard drives are not highly integrated. Besides, high integration does not necessarily mean that the sound quality will decline. MD is an example.

2. Compared with hard disks, the flash memory structure is not afraid of shocks and is more resistant to drops. What the hard drive fears most is strong vibrations. Although we can be very careful when using it, tigers also take a nap sometimes.

3. Flash memory can provide faster data reading speed, while hard disk is limited by rotation speed.

4. Light weight.

The development process of flash memory

·The development history of flash memory

In 1984, Fujio Masuoka, the inventor of Toshiba Corporation, first proposed a fast flash memory (herein referred to as flash memory) concept. Unlike traditional computer memory, flash memory is characterized by non-volatility (that is, the stored data will not be lost after the host is powered off), and its recording speed is also very fast.

Intel was the first company in the world to produce flash memory and put it on the market. In 1988, the company launched a 256K bit flash memory chip. It was about the size of a shoe box and was embedded in a tape recorder. Later, the type of flash memory invented by Intel was collectively called NOR flash memory. It combines EPROM (erasable programmable read-only memory) and EEPROM (electrically erasable programmable read-only memory) technologies and has an SRAM interface.

The second type of flash memory is called NAND flash memory. It was developed by Hitachi in 1989 and is considered an ideal replacement for NOR flash memory. The write cycle of NAND flash memory is ten times shorter than that of NOR flash memory, and its save and delete processing speeds are also relatively fast. The storage unit of NAND is only half of that of NOR, and NAND achieves better performance in a smaller storage space. In view of the excellent performance of NAND, it is often used in memory cards such as CompactFlash, SmartMedia, SD, MMC, xD, and PC cards, USB sticks, etc.

·Analysis of the current situation of the flash memory market

The current flash memory market is still in the final stage of competition between the heroes. Samsung, Hitachi, Spansion and Intel are the four major manufacturers in this market.

Due to some strategic mistakes, Intel gave up its top spot for the first time, falling behind Samsung, Hitachi and Spansion.

Spansion, AMD's flash memory business unit, produces both NAND and NOR flash memory.

Its NOR flash memory output in the first half of the year was almost the same as Intel's, making it the largest manufacturer of NOR flash memory. The company earned $1.3 billion in the first half, nearly half of its overall corporate profit of $2.5 billion.

Overall, Intel and AMD performed well in the first half of the year, but Samsung and Hitachi suffered setbacks.

According to estimates made by market research company iSuppli, global flash memory revenue will reach US$16.6 billion this year, an increase of 46% from 2003 (US$11.64 billion). According to sources, the demand for memory in digital cameras, USB sticks and compressed MP3 players will significantly boost the sales of flash memory. It is predicted that flash memory sales will reach US$17.5 billion in 2005. However, iSuppli estimates that flash memory profit growth will slow down from 2005 to 2008, reaching a maximum of $22.4 billion.

·Are new alternatives possible?

Compared with many short-lived information technologies, flash memory has fully demonstrated its "old-timer" style with its 16-year development history. In the early 1990s, flash memory first entered the market; by 2000, the profit had exceeded one billion US dollars. Peter, director of the flash memory department of Infineon Technologies, once said: "As far as the life cycle of flash memory is concerned, we are still in an upward stage." Infineon believes that there is still room for growth in the sales of flash memory and is planning to join the market. investment. Infineon announced earlier this year that its 200mm DRAM factory in Dresden has begun producing 512Mb NAND-compatible flash memory chips. By the end of 2004, Infineon plans to manufacture more than 10,000 wafers per month using the 170-nanometer manufacturing process. In 2007, the company hopes to become the top three in the NAND market.

In addition, Stefan Lai, vice president of Intel Technology and Manufacturing Group, believes that flash memory will be irreplaceable before 2008. In 2006, Intel will first adopt 65-nanometer technology; by 2008, the new generation of 45-nanometer technology currently under development is expected to be put on the market. Stefan Lai feels that the current predictions are still relatively superficial, and perhaps 32nm and 22nm technologies are entirely possible. But Stefan Lai also admitted that from 2008 to 2010, new technologies may take its place.

Despite the growing discussion about alternatives to flash memory, flash memory is still being taken seriously by the market. The future replacement must not only be non-volatile memory like flash memory, but also be slightly better in speed and write cycles. In addition, production costs should also be relatively low. Since manufacturing technology is not yet mature, new alternatives will not pose an absolute threat to flash memory. Let's take a look at several possible alternative products:

·Nanocrystals

Motorola's semiconductor unit Freescale is developing a product that increases the life cycle of flash memory. This product uses silicon nanocrystals as the medium and replaces the solid layer inside the semiconductor with a grid of silicon atoms. Nanocrystals are not an entirely new storage technology. It's just an improvement over flash memory that makes it more scalable. Its production cost can be about 10-15% lower than the original, and the production process is simpler. Its performance and reliability are comparable to current flash memory.

Motorola has spent ten years developing this technology and plans to mass-produce such products. The company had successfully launched one such chip using this technology last June. Silicon nanocrystal chips are expected to be fully launched on the market in 2006.

→For more related content, please see Nanocrystals

·MRAM (Magnetic RAM magnetic charge random access memory)

MRAM magnetic charge random access memory is developed by Infineon A storage medium developed by two companies with Freescale that uses magnetic charges to store data. MRAM has a high number of writes and its access speed is greatly enhanced compared to flash memory. According to calculations, the time to write 1 bit on an MRAM chip is one million times shorter than the time to write to flash memory.

·Magnetic Charge Random Access Memory

Both companies believe that MRAM will not only be an ideal substitute for flash memory, but also a strong competitor to DRAM and SRAM. In June of this year, Infineon launched its first product on the market. At the same time, Freescale is also stepping up research and development, striving to launch a 4M bit chip next year.

However, some commentators are concerned about whether MRAM can achieve the size of flash memory storage cells. According to Infineon's report, the current size of flash memory storage cells is 0.1?m², while the 16M bit MRAM chip only reaches 1.42?m?. In addition, the production cost of MRAM is also a big problem.

For more information, see Magnetic Charge Random Access Memory.

·OUM (Ovonic Unified Memory Ovonyx standardized memory)

OUM was developed by Intel and is a thin film made of compounds such as Ge, Sb and Te. OUM.

The writing, deleting and reading functions of OUM are similar to those of CD-RW and DVD-RW. However, CD/DVD uses lasers to heat and change materials called chalcogenides; while OUM controls the power supply through transistors to cause phase changes to store data.

The number of erases and writes of OUM is 10 to the 12th power, and the average data access time for 100 times is 200 nanoseconds. OUM is faster than flash memory. Although the data access time of OUM is slower than that of MRAM, low cost is the magic weapon of OUM.

Unlike MRAM, the development of OUM is still in its early stages. Although test chips have been produced, they serve only to confirm the concept rather than demonstrate the feasibility of the technology. Intel has been committed to the research and development of OUM for the past four years and is working hard to expand this market.