Published: 27 June 2024
Contributors: Phill Powell, Ian Smalley
Flash memory is a form of nonvolatile memory with ongoing storage, even without a power source. It enables byte-level rewrites and deletions of data blocks.
The word “flash” is synonymous with speed. A flash is a brief glimmer of light—it happens quickly and then it’s over. Traditional “flash cards” are high-speed teaching aids designed to reinforce memorization techniques. The Flash, the fastest superhero of them all, can race Superman and leave him in a cloud of dust.
Flash memory devices are in wide use and store data for very specific purposes. They are commonly found in a range of portable devices, like USB flash drives, smartphones, digital cameras, video games, tablet computers, flash memory cards and SD cards.
In addition, flash memory now assumes some of the same functionality formerly reserved for computer hard disk drives. For example, when someone turns on a computer, that machine goes through a boot sequence known as Basic Input/Output System (BIOS). The firmware that first contained BIOS required the use of a read-only memory (ROM) chip. Later systems have seen a switch to flash memory for BIOS so that contents can be rewritten without having to extract the chip from the system board.
Take a virtual tour of IBM Flashsystem to learn more about this cyber-resilient primary storage system.
Flash memory stores data in flash memory cells that are based on floating-gate transistors. The computer memory cells of flash memory chips are made up of transistors, which serve as routing switches for the electrical current passing through that flash memory cell.
Flash memory chips are arranged in grids, almost like city blocks. The memory cells are distributed in rows, with these rows being known as bit lines. Similarly to city blocks, these chips contain intersections, and each intersection features a transistor. In turn, these transistors each have two gates.
One of these is the control gate, which is on the top layer of the transistor. The other gate is called the floating gate, so named because it effectively floats between the control gate and the top layer of the MOSFET transistor chip.
In addition, there is a thin separation layer between the control gate and the floating gate, referred to as the oxide layer, although it’s formulated with silicon dioxide (SiO2).
The particular amount of flash memory designates whether that use of flash memory falls into a low-density, medium-density or high-density classification. Higher recordings of density reflect larger amounts of flash memory.
Nearly all progress in computing has been through a cumulative process. First came the development of the early central processing units (CPUs). By 1960, the MOSFET transistor had been created, which would allow for the mass miniaturization of the electronics industry.
In 1967, two researchers from Bell Labs (Dawon Kahng and Simon Min Sze) suggested that a MOSFET’s floating gate might be repurposed as a source of reprogrammable read-only memory (ROM). By 1971, Intel engineer Dov Frohman had invented erasable programmable read-only memory (EPROM). EPROMs can be quickly identified visually because they all have a transparent window on the top of the chip.
The next incremental step involved the creation of electrically erasable programmable read-only memory (EEPROM), another form of an electrically erasable program. EEPROMs were developed during the late 1970s/early 1980s as an update to EPROMs.
EPROMs and EEPROMs differ most noticeably in how data erasure occurs in each. The data on an EPROM can be erased by the presence of ultraviolet (UV) rays while EEPROMs must be erased by using electrical signals.
Flash memory as we know it got its start during the 1980s because of the pioneering work of Dr. Fujio Masuoka, who invented flash memory during his tenure at Toshiba, the Japanese manufacturing giant.
A colleague of the inventor noticed how quickly all the data from a semiconductor chip might be erased—as if that process matched the speed of a camera’s flash unit. Flash memory had been born and now had its name.
There are two basic types of flash memory technology, each with its own architecture and algorithms. In addition, each storage medium offers its own advantages and disadvantages.
NAND flash memory gets its name from a combination of “NOT” and “AND.” This is reference to the logic gate that controls a NAND cell’s internal circuitry.
When a NAND cell is being programmed, an electrical current reaches the control gate and electrons flow onto the floating gate, creating a net positive charge that interrupts current flow. The oxide layer keeps the floating gate isolated so that any electrons on the floating gate are kept there, along with the data being stored. This is what gives flash memory the ability to both hold an electrical charge and retain data.
Erasing a NAND cell is quick since it’s designed to delete entire blocks of data. Again, an electrical charge is applied to the memory cell, and this causes the electrons (and data) that had been trapped within the floating gate to drain back into a bottom isolation layer in the chip. This effectively erases the memory cell.
Producing NAND flash memory chips is not simple or fast.1 It’s been estimated that over 800 distinct manufacturing processes are involved, as well as about one month to create one NAND “wafer,” which is typically about the size of a medium pizza with a 12-inch diameter. Individual NAND chips—roughly the size of a human fingernail—are cut from these wafers and graded according to their chip quality and overall utility.
NAND chips offer many advantages. For starters, NAND chips contain no moving parts, which makes them more rugged and capable of operation even when enduring mechanical shocks, excessive operating temperatures or high pressure. In this regard, NAND-chip operation compares favorably to hard disk drives (HDD) that are more subject to vibration.
On the other hand, NAND use also has drawbacks. Most notable among them is that this storage medium is not open-ended in allowing an infinite number of rewrites to the memory. NAND chips can only be rewritten a certain number of times, which limits their ongoing utility.
Further, NAND flash memory is subject to the same constraints as other systems or devices, which is to say that organizations are overflowing with data and NAND memory cells have had to keep pace by engineering new forms of memory cells. What began with single-level cell (SLC) memory and the storage of one bit for each cell and two levels of charge has ramped up over time, resulting in the creation of multilevel cells (MLCs), triple-level cells (TLCs) and even quadruple-level cells (QLC).
Similar to its counterpart NAND, NOR flash memory’s name is a combination of two words: “NOT” and “OR”—a reference to the type of logic gate that controls the NOR cell’s internal circuitry.
In NOR flash memory, memory cells are connected in parallel to bit lines. This allows them to be both read and programmed individually. One end of each memory cell is attached to the ground, with the other end attached to a bit line.
NOR’s main advantages are its reading speed, a high number of possible rewrites and its ability to accommodate random-access data. This makes NOR gates perfect for use in municipal traffic-light systems, industrial automation, alarm systems, digital circuit design and electronic devices. Another key advantage of NOR flash is the fact that NOR devices can handle both data storage and code execution with one device when using NOR flash.
In terms of its disadvantages, NOR flash memory employs a larger cell size. This results in slower write and erase speeds than NAND flash memory,
Read on to learn more differences between the two types of flash memory.
A main design difference between NAND flash technology and NOR flash technology is the way that memory cells are distributed within a semiconductor. In NAND chips, these cells are aligned vertically. In NOR chips, memory cells are arranged horizontally. This design difference makes these memory systems function differently, with different rates of speed and performance.
NAND technologies usually exhibit latencies somewhere in the range of 80 microseconds to 120 microseconds, while it’s commonly considered that NOR latency rates (link resides outside ibm.com) vary between 160 nanoseconds to 210 nanoseconds—showing that a smaller amount of latency tends to be experienced by NOR flash memory.
It’s often estimated that the typical lifespan of NAND flash memory is somewhere between three and five years. In stark contrast, estimates about the lifespan of NOR flash memory can range anywhere from 20 years up to 100 years (or longer).
Another area of difference between NAND and NOR technologies involves the amount of electricity each requires. However, the power consumption used by each involves a tradeoff. For example, NAND uses less power during its start-up procedures but more current when in standby mode. This differs completely with NOR, which uses more electrical current when first powered on, but less energy when standing by.
The amount of power they use during the “work” each performs is roughly comparable, although this measurement is subject to the rate of memory used by each, and this depends on the activities undertaken by each technology. NOR specializes in quick data reads and consumes less power when doing so. When writing and erasing data, NAND uses less power than NOR.
It should be noted here that neither NAND flash memory or NOR flash memory can approach the processing speeds routinely achieved by other forms of memory. Cache memory is often thought to be the fastest memory of all, by virtue of its position between a computer’s random access memory (RAM) and its central processing unit (CPU).
Further, there’s not a cut-and-dried answer for whether NAND is faster than NOR, or vice versa. It depends on the immediate application that they’re engaged in. If the comparison is based on quick reads, then NOR is faster. If the comparison is about executing tasks and data management, then NAND is faster.
Neither can NAND or NOR keep up with Dynamic Random Access Memory (DRAM), a unique form of RAM that achieves high-performance speeds up to 100 times faster than NAND and offers temporary file storage during the operation of apps or programs. (It’s also worth noting, however, that DRAM is a volatile form of memory, which means its greatest utility is in aiding processing that is occurring in the moment, since once its supporting power is switched off or lost, the DRAM memory loses any data it was working with.)
In another key differentiator, NAND flash memory offers substantially greater storage capacity than NOR, which is typically available in memory increments of 64 Mb to 2 Gb, while NAND storage solutions range in capacity from 1 Gb to 16 Gb—making NAND’s top storage capacity 8 times larger than NOR’s top capacity.
There are other key differences between NAND and NOR, based on the purposes each is used for. It is often suggested that NAND is better-suited to performing “in-depth” processes like rewrites and data-block erasures, while NOR excels at quick data searches that are less involved.
Experience improved performance and cyber-resilience with the new IBM FlashSystem 5300.
Select the ideal data storage technology for your organization, whether it needs data storage hardware, software-defined storage or storage management software. IBM Storage solutions defend data with top cybersecurity capabilities to protect the vital activities of operating systems.
Power all your apps, streamline data administration and reduce silos and complexity to optimize your data center. Storage virtualization options help keep server and desktop environments secure, reliable and efficient.
Discover how to shuttle data between on-premises and cloud data centers or between cloud data centers. Either way, you’ll benefit from consistent data management between on-premises storage and the public cloud. In addition, learn how to implement disaster recover strategies.
Learn what flash storage is and the main types of flash storage used in business. Read use-case stories and how flash storage capably meets business demands, and also revisit its history and see predictions on coming trends.
Want to store data and ensure that the data won’t ever be changed? That’s a job for immutable storage, the storage protocol that’s dedicated to keeping data safe and free from any threat of alteration. Discover its ins and outs here.
Data storage—the process of using magnetic, optical or mechanical media to record and preserve digital information—takes many forms, types and devices. Learn about each of them, in addition to what data storage does for business.
Check how strong your organization’s defenses against cybersecurity threats are. Cyber resilience measures that effectiveness, in addition to the ability of companies to return to normal operations, should cybersecurity incidents occur.
Put the superior speeds of NAND flash memory to work with solid-state drives (SSDs), which use a semiconductor-based design to save persistent data. Explore the advantages and disadvantages of solid-state storage systems and see how they compare to hard disk drives.
Discover how nonvolatile memory express (NVMe) expands the number of parallel requests possible on a single connection into the thousands by accessing flash storage over a peripheral component interconnect express (PCIe) bus.
1 “Understanding NAND Flash Technology” (link resides outside ibm.com), Simms.