Random-access memory: Difference between revisions

Revision as of 23:23, 20 October 2024

Latest revision as of 04:28, 28 August 2025

(34 intermediate revisions by 17 users not shown)

Line 1:

{{short description|Form of computer data storage}}

{{Redirect|RAM|other uses|RamRAM (disambiguation){{!}}Ram}}

{{Distinguish|Random Access Memories|Random-access machine}}

{{use dmy dates|date=August 2025}}

{{pp-protected|small=yes}}

[[File:Swissbit 2GB PC2-5300U-555.jpg|right|thumb|Example of [[read/write memory|writable]] [[volatile memory|volatile]] random-access memory: Synchronous [[Dynamicdynamic RAM]] [[DIMM|modules]], primarily used as main memory in [[personal computers]], [[workstation]]s, and [[Server (computing)|server]]s.]]

{{Memory types}}

[[File:Electronic Memory.jpg|thumb| A 64 bit memory chip die, the SP95 Phase 2 Bufferbuffer Memorymemory produced at IBM mid 60s-1960s, versus [[Magnetic-core memory|memory core iron rings]]]]

[[File:Random Access Memory HyperX.jpg|thumb|8GB [[DDR3]] [[RAM]] stick with a white [[heatsink]]]]

[[File:Swissbit 2GB PC2-5300U-555.jpg|right|thumb|Example of [[read/write memory|writable]] [[volatile memory|volatile]] random-access memory: Synchronous [[Dynamic RAM]] [[DIMM|modules]], primarily used as main memory in [[personal computers]], [[workstation]]s, and [[Server (computing)|server]]s.]]

[[File:Random Access Memory HyperX.jpg|thumb|8GB [[DDR3]] [[RAM]] stick with a white [[heatsink]]]]

'''Random-access memory''' ('''RAM'''; {{IPAc-en|r|æ|m}}) is a form of [[Computer memory|electronic computer memory]] that can be read and changed in any order, typically used to store working [[Data (computing)|data]] and [[machine code]].<ref>{{cite web |title=RAM |url=https://dictionary.cambridge.org/dictionary/english/ram |website=[[Cambridge English Dictionary]] |access-date=11 July 2019}}</ref><ref>{{cite web |title=RAM |url=https://www.oxfordlearnersdictionaries.com/definition/american_english/ram_2 |website=[[Oxford Advanced Learner's Dictionary]] |access-date=11 July 2019}}</ref> A [[random- access]] memory device allows data items to be [[read (computer)|read]] or written in almost the same amount of time irrespective of the physical ___location of data inside the memory, in contrast with other direct-access data storage media (such as [[hard disk]]s and [[Magnetic tape data storage|magnetic tape]]), where the time required to read and write data items varies significantly depending on their physical locations on the recording medium, due to mechanical limitations such as media rotation speeds and arm movement.

In today'smodern technology, random-access memory takes the form of [[integrated circuit]] (IC) chips with [[MOSFET|MOS]] (metal–oxide–semiconductor) [[Memory cell (computing)|memory cells]]. RAM is normally associated with [[Volatile memory|volatile]] types of memory where stored information is lost if power is removed. The two main types of volatile random-access [[semiconductor memory]] are [[static random-access memory]] (SRAM) and [[dynamic random-access memory]] (DRAM).

Non-volatile RAM has also been developed<ref>{{cite magazine|last=Gallagher|first=Sean|title=Memory that never forgets: non-volatile DIMMs hit the market|url=https://arstechnica.com/information-technology/2013/04/memory-that-never-forgets-non-volatile-dimms-hit-the-market/|magazine=[[Ars Technica]]|url-status=live|archive-url=https://web.archive.org/web/20170708073138/https://arstechnica.com/information-technology/2013/04/memory-that-never-forgets-non-volatile-dimms-hit-the-market/|archive-date=July 8, 2017|date=April 4, 2013}}</ref> and other types of [[Non-volatile memory|non-volatile memories]] allow random access for read operations, but either do not allow write operations or have other kinds of limitations. These include most types of [[ROM]] and [[NOR flash memory]].

Line 20 ⟶ 21:

[[File:Early SSA accounting operations.jpg|thumb|These IBM [[tabulating machine]]s from the mid-1930s used [[mechanical counter]]s to store information.]]

Early computers used [[relay]]s, [[mechanical counter]]s<ref>{{cite web|url=http://www-03.ibm.com/ibm/history/reference/faq_0000000011.html|title=IBM Archives -- FAQ's for Products and Services|work=ibm.com|url-status=dead|archive-url=https://web.archive.org/web/20121023184527/http://www-03.ibm.com/ibm/history/reference/faq_0000000011.html|archive-date=2012-10-23}}</ref> or [[Delay-line memory|delay lines]] for main memory functions. Ultrasonic delay lines were [[bit-serial architecture|serial devices]] which could only reproduce data in the order it was written. [[Drum memory]] could be expanded at relatively low cost but efficient retrieval of memory items requires knowledge of the physical layout of the drum to optimize speed. Latches built out of [[triode vacuum tube]]s, and later, out of [[discrete transistor]]s, were used for smaller and faster memories such as [[Hardware register|registers]]. Such registers were relatively large and too costly to use for large amounts of data; generally, only a few dozen or few hundred bits[[bit]]s of such memory could be provided.

The first practical form of random-access memory was the [[Williams tube]]. It stored data as electrically charged spots on the face of a [[cathode-ray tube]]. Since the electron beam of the CRT could read and write the spots on the tube in any order, memory was random access. The capacity of the Williams tube was a few hundred to around a thousand bits, but it was much smaller, faster, and more power-efficient than using individual vacuum tube latches. Developed at the [[Victoria University of Manchester|University of Manchester]] in England, the Williams tube provided the medium on which the first electronically stored program was implemented in the [[Manchester Baby]] computer, which first successfully ran a program on 21 June, 1948.<ref>{{Citation | last = Napper | first = Brian | title = Computer 50: The University of Manchester Celebrates the Birth of the Modern Computer | url = http://www.computer50.org/ | access-date = 26 May 2012 | url-status = dead | archive-url = https://web.archive.org/web/20120504133240/http://www.computer50.org/ | archive-date = 4 May 2012 }}</ref> In fact, rather than the Williams tube memory being designed for the Baby, the Baby was a [[testbed]] to demonstrate the reliability of the memory.<ref>{{Citation |last1=Williams |first1=F. C. |last2=Kilburn |first2=T. |title=Electronic Digital Computers |journal=Nature |volume=162 |pages=487 |date=Sep 1948 |doi=10.1038/162487a0 |issue=4117 |postscript=. |bibcode=1948Natur.162..487W |s2cid=4110351|doi-access=free }} Reprinted in ''The Origins of Digital Computers''.</ref><ref>{{Citation |last1=Williams |first1=F. C. |last2=Kilburn |first2=T. |last3=Tootill |first3=G. C. |title=Universal High-Speed Digital Computers: A Small-Scale Experimental Machine |url=http://www.computer50.org/kgill/mark1/ssem.html |journal=Proc. IEE |date=Feb 1951 |volume=98 |issue=61 |pages=13–28 |postscript=. |doi=10.1049/pi-2.1951.0004 |url-status=dead |archive-url=https://web.archive.org/web/20131117101730/http://www.computer50.org/kgill/mark1/ssem.html |archive-date=2013-11-17|url-access=subscription }}</ref>

[[Magnetic-core memory]] was invented in 1947 and developed up until the mid-1970s. It became a widespread form of random-access memory, relying on an array of magnetized rings. By changing the sense of each ring's magnetization, data could be stored with one bit stored per ring. Since every ring had a combination of address wires to select and read or write it, access to any memory ___location in any sequence was possible. Magnetic core memory was the standard form of [[computer memory]] until displaced by [[semiconductor memory]] in [[integrated circuit]]s (ICs) during the early 1970s.<ref name="computerhistory1970"/>

Line 28 ⟶ 29:

Prior to the development of integrated [[read-only memory]] (ROM) circuits, ''permanent'' (or ''read-only'') random-access memory was often constructed using [[Diode matrix|diode matrices]] driven by [[address decoder]]s, or specially wound [[core rope memory]] planes.{{Citation needed|date=December 2016}}

[[Semiconductor memory]] appeared in the 1960s with bipolar memory, which used [[bipolar transistor]]s. Although it was faster, it could not compete with the lower price of magnetic core memory.<ref name="computerhistory1966"/><!--[[User:Kvng/RTH]]-->

===MOS RAM===

In 1957, Frosch and Derick were able to manufacturemanufactured the first silicon dioxide field -effect transistors at Bell Labs, the first transistors in which drain and source were adjacent at the surface.<ref>{{Cite journal |last1=Frosch |first1=C. J. |last2=Derick |first2=L |date=1957 |title=Surface Protection and Selective Masking during Diffusion in Silicon |url=https://iopscience.iop.org/article/10.1149/1.2428650 |journal=Journal of the Electrochemical Society |language=en |volume=104 |issue=9 |pages=547 |doi=10.1149/1.2428650|url-access=subscription}}</ref> Subsequently, in 1960, a team demonstrated a working [[MOSFET]] at Bell Labs 1960.<ref>{{Cite journal |last=KAHNG |first=D. |orig-date=1961 |title=Silicon-Silicon Dioxide Surface Device |url=https://doi.org/10.1142/9789814503464_0076 |journal=Technical Memorandum of Bell Laboratories |pages=583–596 |doi=10.1142/9789814503464_0076 |isbn=978-981-02-0209-5|url-access=subscription}}</ref><ref>{{Cite book |last=Lojek |first=Bo |title=History of Semiconductor Engineering |date=2007 |publisher=Springer-Verlag Berlin Heidelberg |isbn=978-3-540-34258-8 |___location=Berlin, Heidelberg |page=321}}</ref> This led to the development of [[metal–oxide–semiconductor]] (MOS) memory by John Schmidt at [[Fairchild Semiconductor]] in 1964.<ref name="computerhistory1970" /><ref>{{Cite book |url=https://books.google.com/books?id=kG4rAQAAIAAJ&q=John+Schmidt |title=Solid State Design – Vol. 6 |date=1965 |publisher=Horizon House}}</ref> In addition to higher speeds, MOS [[semiconductor memory]] was cheaper and consumed less power than magnetic core memory.<ref name="computerhistory1970"/> The development of [[silicon-gate]] [[MOS integrated circuit]] (MOS IC) technology by [[Federico Faggin]] at Fairchild in 1968 enabled the production of MOS [[memory chip]]s.<ref>{{cite web |title=1968: Silicon Gate Technology Developed for ICs |url=https://www.computerhistory.org/siliconengine/silicon-gate-technology-developed-for-ics/ |website=[[Computer History Museum]] |access-date=10 August 2019}}</ref> MOS memory overtook magnetic core memory as the dominant memory technology in the early 1970s.<ref name="computerhistory1970" />

This led to the development of [[metal–oxide–semiconductor]] (MOS) memory by John Schmidt at [[Fairchild Semiconductor]] in 1964.<ref name="computerhistory1970" /><ref>{{Cite book |url=https://books.google.com/books?id=kG4rAQAAIAAJ&q=John+Schmidt |title=Solid State Design – Vol. 6 |date=1965 |publisher=Horizon House}}</ref> In addition to higher speeds, MOS [[semiconductor memory]] was cheaper and consumed less power than magnetic core memory.<ref name="computerhistory1970" /> The development of [[silicon-gate]] [[MOS integrated circuit]] (MOS IC) technology by [[Federico Faggin]] at Fairchild in 1968 enabled the production of MOS [[memory chip]]s.<ref>{{cite web |title=1968: Silicon Gate Technology Developed for ICs |url=https://www.computerhistory.org/siliconengine/silicon-gate-technology-developed-for-ics/ |website=[[Computer History Museum]] |access-date=10 August 2019}}</ref> MOS memory overtook magnetic core memory as the dominant memory technology in the early 1970s.<ref name="computerhistory1970" />

Integrated bipolar [[static random-access memory]] (SRAM) was invented by Robert H. Norman at [[Fairchild Semiconductor]] in 1963.<ref>{{cite patent

Line 45 ⟶ 44:

| inventor = Robert H. Norman

| invent1 = Fairchild Camera and Instrument Corporation

}}</ref> It was followed by the development of MOS SRAM by John Schmidt at Fairchild in 1964.<ref name="computerhistory1970"/> SRAM became an alternative to magnetic-core memory, but required six MOS transistors for each [[bit]] of data.<ref name="ibm100">{{cite web |title=DRAM |url=https://www.ibm.com/ibm/history/ibm100/us/en/icons/dram/ |website=IBM100 |publisher=[[IBM]] |access-date=20 September 2019 |date=9 August 2017}}</ref> Commercial use of SRAM began in 1965, when [[IBM]] introduced the SP95 memory chip for the [[IBM System/360|System/360 Model 95]].<ref name="computerhistory1966"/>

[[Dynamic random-access memory]] (DRAM) allowed replacement of a 4 or 6-transistor latch circuit by a single transistor for each memory bit, greatly increasing memory density at the cost of volatility. Data was stored in the tiny capacitance of each transistor, and had to be periodically [[Memory refresh|refreshed]] every few milliseconds before the charge could leak away.

[[Toshiba]]'s Toscal BC-1411 [[electronic calculator]], which was introduced in 1965,<ref>[http://collection.sciencemuseum.org.uk/objects/co8406093/toscal-bc-1411-calculator-with-electronic-calculator Toscal BC-1411 calculator]. {{webarchive|url=https://web.archive.org/web/20170729145228/http://collection.sciencemuseum.org.uk/objects/co8406093/toscal-bc-1411-calculator-with-electronic-calculator |date=2017-07-29 }}, [[Science Museum, London]].</ref><ref name="bc-spec" /><ref name="bc" /> used a form of capacitor- bipolar DRAM, storing 180-bit data on discrete [[Memory cell (computing)|memory cells]], consisting of [[germanium]] bipolar transistors and capacitors.<ref name="bc-spec" /><ref name="bc" /> Capacitors had also been used for earlier memory schemes, such as the drum of the [[Atanasoff–Berry Computer]], the [[Williams tube]] and the [[Selectron tube]]. While it offered higher speeds than magnetic-core memory, bipolar DRAM could not compete with the lower price of the then -dominant magnetic-core memory.<ref>{{cite web |title=1966: Semiconductor RAMs Serve High-speed Storage Needs |url=https://www.computerhistory.org/siliconengine/semiconductor-rams-serve-high-speed-storage-needs/ |website=Computer History Museum}}</ref> Capacitors had also been used for earlier memory schemes, such as the drum of the [[Atanasoff–Berry Computer]], the [[Williams tube]] and the [[Selectron tube]].

[[File:Bundesarchiv Bild 183-1989-0406-022, VEB Carl Zeiss Jena, 1-Megabit-Chip.jpg|thumb|right|CMOS 1-[[megabit]] (Mbit) DRAM chip, one of the last models developed by [[VEB Carl Zeiss Jena]], in 1989]]

In 1966, Dr. [[Robert Dennard]], invented invented modern DRAM architecture for which there's a single MOS transistor per capacitor.<ref name="ibm100" /> Whilewhile examining the characteristics of MOS technology, he found it was capable of building [[capacitor]]s, and that storing a charge or no charge on the MOS capacitor could represent the 1 and 0 of a bit, whileand the MOS transistor could control writing the charge to the capacitor. This led to his development of modern DRAM architecture for which there is a single-transistor DRAMMOS memorytransistor cellper capacitor.<ref name="ibm100"/> In 1967, Dennard filed a patent under IBM for a single-transistor DRAM memory cell, based on MOS technology.<ref name="ibm100" /><ref name="Robert Dennard"/> The first commercial DRAM IC chip was the [[Intel 1103]], which was [[Semiconductor manufacturing process|manufactured]] on an [[10 μm process|8{{nbsp}}μm]] MOS process with a capacity of 1{{nbsp}}[[Kilobit|kbit]], and was released in 1970.<ref name="computerhistory1970"/><ref name="Lojek-1103"/><ref>{{cite web |first=Mary |last=Bellis |url=http://inventors.about.com/library/weekly/aa100898.htm |title=TheWho Invention ofInvented the Intel 1103 DRAM Chip? |access-date=20152025-0703-1103 |archive-date=2020-03-14 |archive-url=https://web.archive.org/web/20200314061801/http://inventors.about.com/library/weekly/aa100898.htm |url-status=dead }}</ref>

The earliest DRAMs were often synchronized with the CPU clock (clocked) and were used with early microprocessors. In the mid-1970s, DRAMs moved to the asynchronous design, but in the 1990s returned to synchronous operation.<ref>{{cite book |author=P. Darche |url=https://books.google.com/books?id=rLC9zQEACAAJ |title=Microprocessor: Prolegomenes - Calculation and Storage Functions - Calculation Models and Computer |year=2020 |isbn=9781786305633 |page=59| publisher=John Wiley & Sons }}</ref><ref>{{cite book |author1=B. Jacob |url=https://books.google.com/books?id=SrP3aWed-esC |title=Memory Systems: Cache, DRAM, Disk |author2=S. W. Ng |author3=D. T. Wang |publisher=Morgan Kaufmann |year=2008 |isbn=9780080553849 |page=324}}</ref> In 1992 Samsung released KM48SL2000, which had a capacity of 16{{nbsp}}[[Mbit]].<ref name="electronic-design">{{cite journal |title=Electronic Design |journal=[[Electronic Design]] |date=1993 |volume=41 |issue=15–21 |url=https://books.google.com/books?id=QmpJAQAAIAAJ |publisher=Hayden Publishing Company |quote=The first commercial synchronous DRAM, the Samsung 16-Mbit KM48SL2000, employs a single-bank architecture that lets system designers easily transition from asynchronous to synchronous systems.}}</ref><ref>{{cite web |title=KM48SL2000-7 Datasheet |url=https://www.datasheetarchive.com/KM48SL2000-7-datasheet.html |publisher=[[Samsung]] |access-date=19 June 2019 |date=August 1992}}</ref> and mass-produced in 1993.<ref name="electronic-design"/> The first commercial [[DDR SDRAM]] ([[double data rate]] SDRAM) memory chip was Samsung's 64{{nbsp}}Mbit [[DDR SDRAM chip]], released in June 1998.<ref>{{cite news |title=Samsung Electronics Develops First 128Mb SDRAM with DDR/SDR Manufacturing Option |url=https://www.samsung.com/semiconductor/insights/news-events/samsung-electronics-develops-first-128mb-sdram-with-ddr-sdr-manufacturing-option/ |access-date=23 June 2019 |work=[[Samsung Electronics]] |publisher=[[Samsung]] |date=10 February 1999}}</ref> [[GDDR]] (graphics DDR) is a form of DDR [[SGRAM]] (synchronous graphics RAM), which was first released by Samsung as a 16{{nbsp}}Mbit memory chip in 1998.<ref>{{cite news |title=Samsung Electronics Comes Out with Super-Fast 16M DDR SGRAMs |url=https://www.samsung.com/semiconductor/insights/news-events/samsung-electronics-comes-out-with-super-fast-16m-ddr-sgrams/ |access-date=23 June 2019 |work=[[Samsung Electronics]] |publisher=[[Samsung]] |date=17 September 1998}}</ref>

==Types==

In general, the term ''RAM'' refers solely to solid-state memory devices, and more specifically the main memory in most computers. The two widely used forms of modern RAM are [[static RAM]] (SRAM) and [[dynamic RAM]] (DRAM). In SRAM, a [[Bit|bit of data]] is stored using the state of a six-[[transistor]] [[Memory cell (computing)|memory cell]], typically using six MOSFETs. This form of RAM is more expensive to produce, but is generally faster and requires less dynamicstatic power than DRAM. In modern computers, SRAM is often used as [[CPU cache|cache memory for the CPU]]. DRAM stores a bit of data using a transistor and [[capacitor]] pair (typically a MOSFET and [[MOS capacitor]], respectively),<ref>{{cite book |last1=Sze |first1=Simon M. |author1-link=Simon Sze |title=Semiconductor Devices: Physics and Technology |date=2002 |publisher=[[Wiley (publisher)|Wiley]] |isbn=0-471-33372-7 |page=214 |edition=2nd |url=http://www.fulviofrisone.com/attachments/article/453/Semiconductor.Devices_Physics.Technology_Sze.2ndEd_Wiley_2002.pdf}}</ref> which together comprise a DRAM cell. The capacitor holds a high or low charge (1 or 0, respectively), and the transistor acts as a switch that lets the control circuitry on the chip read the capacitor's state of charge or change it. As this form of memory is less expensive to produce than static RAM, it is the predominant form of computer memory used in modern computers.

Both static and dynamic RAM are considered ''volatile'', as their state is lost or reset when power is removed from the system. By contrast, [[read-only memory]] (ROM) stores data by permanently enabling or disabling selected transistors, such that the memory cannot be altered. Writable variants of ROM (such as [[EEPROM]] and [[NOR flash]]) share properties of both ROM and RAM, enabling data to [[Persistence (computer science)|persist]] without power and to be updated without requiring special equipment. [[ECC memory]] (which can be either SRAM or DRAM) includes special circuitry to detect and/or correct random faults (memory errors) in the stored data, using [[parity bit]]s or [[Error detection and correction#Error-correcting code|error correction codes]].

[[ECC memory]] (which can be either SRAM or DRAM) includes special circuitry to detect and/or correct random faults (memory errors) in the stored data, using [[parity bit]]s or [[Error detection and correction#Error-correcting code|error correction codes]].

In general, the term ''RAM'' refers solely to solid-state memory devices (either DRAM or SRAM), and more specifically the main memory in most computers. In optical storage, the term [[DVD-RAM]] is somewhat of a misnomer since, it is not random access; it behaves much like a hard disc drive if somewhat slower. Aside, unlike [[CD-RW]] or [[DVD-RW]], DVD-RAM does not need to be erased before reuse.

==Memory cell==

{{main|Memory cell (computing)}}

The memory cell is the fundamental building block of [[computer memory]]. The memory cell is an [[electronic circuit]] that stores one [[bit]] of binary information. andThe itcell mustcan be set to store a logic 1 (high voltage level) and reset to store a logic 0 (low voltage level). Its value is maintained/stored until it is changed by the set/reset process. The value in the memory cell can be accessed by reading it.

In SRAM, the memory cell is a type of [[flip-flop (electronics)|flip-flop]] circuit, usually implemented using [[FET]]s. This means that SRAM requires very low power when not being accessed, but it is complex, expensive and has low storage density.

A second type, DRAM, is based around a capacitor. Charging and discharging this capacitor can store a "1" or a "0" in the cell. However, the charge in this capacitor slowly leaks away, and must be refreshed periodically. Because of this refresh process, DRAM uses more power, but it can achieve greater storage densities and lower unit costs compared to SRAM.

{| style="text-align:center; margin: 1em auto 1em auto"

|[[File:SRAM Cell (6 Transistors).svg|thumb|class=skin-invert-image|SRAM Cellcell (6 Transistorstransistors)]]||[[File:DRAM Cell Structure (Model of Single Circuit Cell).PNG|thumb|DRAM Cellcell (1 Transistortransistor and one capacitor)]]

|}<!--[[User:Kvng/RTH]]-->

|}

==Addressing==

To be useful, memory cells must be readable and writable. Within the RAM device, multiplexing and demultiplexing circuitry is used to select memory cells. Typically, a RAM device has a set of address lines <math>A_0, A_1,...A_n</math>, and for each combination of bits that may be applied to these lines, a set of memory cells are activated. Due to this addressing, RAM devices virtually always have a memory capacity that is a power of two.

Usually several memory cells share the same address. For example, a 4 bit '"wide'" RAM chip has 4four memory cells for each address. Often the width of the memory and that of the microprocessor are different, for a 32 bit microprocessor, eight 4 bit RAM chips would be needed.

Often more addresses are needed than can be provided by a device. In that case, external multiplexors to the device are used to activate the correct device that is being accessed. RAM is often byte addressable, although it is also possible to make RAM that is word-addressable.<ref>{{cite book | url=https://books.google.com/books?id=QGPHAl9GE-IC&dq=size+of+a+memory+address&pg=PA321 | isbn=978-0-7637-3769-6 | title=The Essentials of Computer Organization and Architecture | date=2006 | publisher=Jones & Bartlett Learning }}</ref><ref>{{cite book | url=https://books.google.com/books?id=-vQCEAAAQBAJ | title=Foundations of Computer Technology | isbn=978-1-000-15371-2 | last1=Anderson | first1=Alexander John | date=25 October 2020 | publisher=CRC Press }}</ref>

==Memory hierarchy==

Line 105 ⟶ 104:

==Memory wall==

The ''''memory wall''' is the growing disparity of speed between CPU and the response time of memory (known as [[memory latency]]) outside the CPU chip. An important reason for this disparity is the limited communication bandwidth beyond chip boundaries, which is also referred to as ''bandwidth wall''. From 1986 to 2000, [[CPU]] speed improved at an annual rate of 55% while off-chip memory response time only improved at 10%. Given these trends, it was expected that memory latency would become an overwhelming [[bottleneck (engineering)|bottleneck]] in computer performance.<ref>The term was coined in {{cite web |url=http://www.eecs.ucf.edu/~lboloni/Teaching/EEL5708_2006/slides/wulf94.pdf |title=Archived copy |access-date=2011-12-14 |url-status=live |archive-url=https://web.archive.org/web/20120406111104/http://www.eecs.ucf.edu/~lboloni/Teaching/EEL5708_2006/slides/wulf94.pdf |archive-date=2012-04-06 }}.</ref>

Another reason for the disparity is the enormous increase in the size of memory since the start of the PC revolution in the 1980s. Originally, PCs contained less than 1 mebibyte of RAM, which often had a response time of 1 CPU clock cycle, meaning that it required 0 wait states. Larger memory units are inherently slower than smaller ones of the same type, simply because it takes longer for signals to traverse a larger circuit. Constructing a memory unit of many gibibytes with a response time of one clock cycle is difficult or impossible. Today'sModern CPUs often still have a mebibyte of 0 wait state cache memory, but it resides on the same chip as the CPU cores due to the bandwidth limitations of chip-to-chip communication. It must also be constructed from static RAM, which is far more expensive than the dynamic RAM used for larger memories. Static RAM also consumes far more power.

CPU speed improvements slowed significantly partly due to major physical barriers and partly because current CPU designs have already hit the memory wall in some sense. [[Intel]] summarized these causes in a 2005 document.<ref>{{Cite web |title= Platform 2015: Intel Processor and Platform Evolution for the Next Decade |date= March 2, 2005 |url= http://epic.hpi.uni-potsdam.de/pub/Home/TrendsAndConceptsII2010/HW_Trends_borkar_2015.pdf |url-status= live |archive-url= https://web.archive.org/web/20110427072037/http://epic.hpi.uni-potsdam.de/pub/Home/TrendsAndConceptsII2010/HW_Trends_borkar_2015.pdf |archive-date= April 27, 2011 }}</ref>

<blockquote>First of all, as chip geometries shrink and clock frequencies rise, the transistor [[leakage current]] increases, leading to excess power consumption and heat... Secondly, the advantages of higher clock speeds are in part negated by memory latency, since memory access times have not been able to keep pace with increasing clock frequencies. Third, for certain applications, traditional serial architectures are becoming less efficient as processors get faster (due to the so-called [[von Neumann bottleneck]]), further undercutting any gains that frequency increases might otherwise buy. In addition, partly due to limitations in the means of producing inductance within solid state devices, [[RC time constant#Delay|resistance-capacitance]] (RC) delays in signal transmission are growing as feature sizes shrink, imposing an additional bottleneck that frequency increases don't address.</blockquote>

Line 117 ⟶ 116:

A different concept is the processor-memory performance gap, which can be addressed by [[Three-dimensional integrated circuit|3D integrated circuits]] that reduce the distance between the logic and memory aspects that are further apart in a 2D chip.<ref>{{cite book |page=790 |url=https://books.google.com/books?id=1PgYS7zDCM8C&q=processor-memory+performance+gap&pg=PA790 |access-date=March 31, 2014 |title=Nanoelectronics and Information Technology |author=Rainer Waser |publisher=John Wiley & Sons |year=2012 |url-status=live |archive-url=https://web.archive.org/web/20160801114150/https://books.google.com/books?id=1PgYS7zDCM8C&pg=PA790&dq=processor-memory+performance+gap&hl=en&sa=X&ei=jeM5U93YAqTr2QWc74A4&ved=0CDYQ6AEwAg#v=onepage&q=processor-memory%20performance%20gap&f=false |archive-date=August 1, 2016 |isbn = 9783527409273|author-link = Rainer Waser}}</ref> Memory subsystem design requires a focus on the gap, which is widening over time.<ref>{{cite book |url=https://books.google.com/books?id=0IY7LW5J4JgC&q=processor-memory+performance+gap&pg=PA109 |page=109 |access-date=March 31, 2014 |title=Advances in Computer Systems Architecture: 11th Asia-Pacific Conference, ACSAC 2006, Shanghai, China, September 6-8, 2006, Proceedings |author=Chris Jesshope and Colin Egan |publisher=Springer |date=2006 |url-status=live |archive-url=https://web.archive.org/web/20160801135254/https://books.google.com/books?id=0IY7LW5J4JgC&pg=PA109&dq=processor-memory+performance+gap&hl=en&sa=X&ei=jeM5U93YAqTr2QWc74A4&ved=0CEkQ6AEwBg#v=onepage&q=processor-memory%20performance%20gap&f=false |archive-date=August 1, 2016 |isbn=9783540400561 }}</ref> The main method of bridging the gap is the use of [[Cache (computing)|caches]]; small amounts of high-speed memory that houses recent operations and instructions nearby the processor, speeding up the execution of those operations or instructions in cases where they are called upon frequently. Multiple levels of caching have been developed to deal with the widening gap, and the performance of high-speed modern computers relies on evolving caching techniques.<ref>{{cite book |url=https://books.google.com/books?id=7i9Z69lrYBoC&q=processor-memory+performance+gap&pg=PA90 |pages=90–91 |access-date=March 31, 2014 |title=Multiprocessor Systems-on-chips |author=Ahmed Amine Jerraya and Wayne Wolf |publisher=Morgan Kaufmann |year=2005 |url-status=live |archive-url=https://web.archive.org/web/20160801105357/https://books.google.com/books?id=7i9Z69lrYBoC&pg=PA90&dq=processor-memory+performance+gap&hl=en&sa=X&ei=jeM5U93YAqTr2QWc74A4&ved=0CFMQ6AEwCA#v=onepage&q=processor-memory%20performance%20gap&f=false |archive-date=August 1, 2016 |isbn=9780123852519 }}</ref> There can be up to a 53% difference between the growth in speed of processor and the lagging speed of main memory access.<ref>{{cite book |url=https://books.google.com/books?id=f0pJYJQMlmoC&q=processor-memory+performance+gap&pg=PA529 |page=529 |access-date=March 31, 2014 |title=Experimental and Efficient Algorithms: Third International Workshop, WEA 2004, Angra Dos Reis, Brazil, May 25-28, 2004, Proceedings, Volume 3 |author=Celso C. Ribeiro and Simone L. Martins |publisher=Springer |year=2004 |url-status=live |archive-url=https://web.archive.org/web/20160801092734/https://books.google.com/books?id=f0pJYJQMlmoC&pg=PA529&dq=processor-memory+performance+gap&hl=en&sa=X&ei=1eM5U7veEaTx2QXM2oDYCw&ved=0CCwQ6AEwADgU#v=onepage&q=processor-memory%20performance%20gap&f=false |archive-date=August 1, 2016 |isbn=9783540220671 }}</ref>

[[Solid-state drive|Solid-state hard drives]] have continued to increase in speed, from ~400&nbsp;Mbit/s via [[Serial ATA|SATA3]] in 2012 up to ~7&nbsp;GB/s via [[NVMe]]/[[PCIe]] in 2024, closing the gap between RAM and hard disk speeds, although RAM continues to be an order of magnitude faster, with single-lane [[DDR5 SDRAM|DDR5]] 8000MHz capable of 128&nbsp;GB/s, and modern [[GDDR]] even faster. Fast, cheap, [[non-volatile]] solid state drives have replaced some functions formerly performed by RAM, such as holding certain data for immediate availability in [[server farm]]s - 1 [[terabyte]] of SSD storage can be had for $200, while 1&nbsp;TB of RAM would cost thousands of dollars.<ref>{{Cite web|url=https://www.minitool.com/news/ssd-prices-fall.html|title=SSD Prices Continue to Fall, Now Upgrade Your Hard Drive!|date=2018-09-03|website=MiniTool|language=en-us|access-date=2019-03-28}}</ref><ref>{{Cite web|url=https://www.digitaltrends.com/computing/ram-prices-are-increasing-until-third-quarter-2017/|title=If you're buying or upgrading your PC, expect to pay more for RAM|last=Coppock|first=Mark|date=31 January 2017|website=www.digitaltrends.com|access-date=2019-03-28}}</ref>

==Timeline==

Line 162 ⟶ 161:

|{{?}}

|{{n/a}}

|<ref>{{cite book |title=IBM first in IC memory |url=https://www.computerhistory.org/collections/catalog/102770626 |websitevia=[[Computer History Museum]] |year=1965 |publisher=IBM Corporation |access-date=19 June 2019}}</ref>

|-

|{{?}}

Line 234 ⟶ 233:

|12,000 [[Nanometre|nm]]

|PMOS

|<ref name="Intel-Product-Timeline"/><ref name="shmj-1970s-sram">{{cite web |title=1970s: SRAM evolution |url=http://www.shmj.or.jp/english/pdf/ic/exhibi724E.pdf |website=Semiconductor History Museum of Japan |access-date=27 June 2019}}</ref><ref name="Pimbley">{{cite book |last1=Pimbley |first1=J. |title=Advanced CMOS Process Technology |date=2012 |publisher=[[Elsevier]] |isbn=9780323156806 |page=7 |url=https://books.google.com/books?id=8EUWHSqevQoC&pg=PA7}}</ref><ref>{{Cite web|url=https://www.intel-vintage.info/intelmemory.htm|title=Intel Memory|website=Intel Vintage|access-date=2019-07-06|ref=intel-memory|archive-date=2022-03-19|archive-url=https://web.archive.org/web/20220319073833/https://www.intel-vintage.info/intelmemory.htm|url-status=deadusurped}}</ref>

|-

|1972

Line 331 ⟶ 330:

|rowspan="2" | {{?}}

|rowspan="2" | CMOS

|rowspan="2" | <ref name="stolstl"/>

|-

|64 kbit

Line 346 ⟶ 345:

|2,500&nbsp;nm

|NMOS

|<ref name="stolstl"/>

|-

|{{dts|1981|10}}

Line 366 ⟶ 365:

|[[1.5 μm process|1,500 nm]]

|NMOS (HMOS)

|<ref name="stolstl"/>

|-

|{{dts|1983|2}}

Line 386 ⟶ 385:

|1,200&nbsp;nm

|CMOS

|<ref name="stolstl"/><ref name="Pimbley"/>

|-

|1987

Line 396 ⟶ 395:

|{{?}}

|CMOS

|<ref name="stolstl"/>

|-

|{{dts|1987|12}}

Line 416 ⟶ 415:

|{{?}}

|rowspan="2" | CMOS

|rowspan="2" | <ref name="stolstl"/>

|-

|1992

Line 452 ⟶ 451:

|{{?}}

|CMOS

|<ref name="hynix90s-skhynix.com">{{cite web |title=History: 1990s |url=https://www.skhynix.com/eng/about/history1990.jsp |website=[[SK Hynix]] |access-date=6 July 2019 |archive-date=5 February 2021 |archive-url=https://web.archive.org/web/20210205032928/https://www.skhynix.com/eng/about/history1990.jsp |url-status=dead }}</ref>

|}

Line 526 ⟶ 525:

|PMOS

|10&nbsp;mm²

|<ref name="Intel2003">{{cite web |title=Intel: 35 Years of Innovation (1968–2003) |url=https://www.intel.com/Assets/PDF/General/35yrs.pdf |publisher=Intel |year=2003 |access-date=26 June 2019}}</ref><ref name="HC">[http://history-computer.com/ModernComputer/Basis/dram.html ''The DRAM memory of Robert Dennard''] {{Webarchive|url=https://web.archive.org/web/20200801004808/https://history-computer.com/ModernComputer/Basis/dram.html |date=2020-08-01 }} history-computer.com</ref><ref name="Lojek-1103">{{cite book |last1=Lojek |first1=Bo |title=History of Semiconductor Engineering |date=2007 |publisher=[[Springer Science & Business Media]] |isbn=9783540342588 |pages=362–363 |url=https://books.google.com/books?id=2cu1Oh_COv8C&pg=PA362 |quote=The i1103 was manufactured on a 6-mask silicon-gate P-MOS process with 8 μm minimum features. The resulting product had a 2,400 μm² memory cell size, a die size just under 10 mm², and sold for around $21.}}</ref>

|-

| rowspan="2" |1971

Line 742 ⟶ 741:

|CMOS

|{{?}}

| rowspan="2" |<ref name="stolstl">{{cite web|url=http://maltiel-consulting.com/Semiconductor_technology_memory.html|title=Memory|website=STOL (Semiconductor Technology Online)|access-date=25 June 2019|archive-date=2 November 2023|archive-url=https://web.archive.org/web/20231102131915/http://maltiel-consulting.com/Semiconductor_technology_memory.html|url-status=dead}}</ref>

|-

|1993

Line 771 ⟶ 770:

|CMOS

|{{?}}

|rowspan="2" |<ref name="HB19950109">{{usurped|1=[https://web.archive.org/web/20140827092848/http://business.highbeam.com/3591/article-1G1-16482653/breaking-gigabit-barrier-drams-isscc-portend-major ''Breaking the gigabit barrier, DRAMs at ISSCC portend major system-design impact. (dynamic random access memory; International Solid-State Circuits Conference; Hitachi Ltd. and NEC Corp. research and development)'']}}, January 9, 1995</ref><ref name="smithsonian-japan"/>

|-

|Hitachi

Line 796 ⟶ 795:

|CMOS

|{{?}}

|<ref name="stolstl"/>

|-

|1998

Line 806 ⟶ 805:

|CMOS

|{{?}}

|<ref name="hynix90s-skhynix.com"/>

|}-

|{{sort|2001|February 2001}}

|{{?}}

|4 Gbit

|DRAM

|Samsung

|[[100 nm]]

|CMOS

|{{?}}

|<ref name="stolstl"/><ref>{{cite web |title=A Study of the DRAM industry |url=https://dspace.mit.edu/bitstream/handle/1721.1/59138/659514510-MIT.pdf |publisher=[[MIT]] |date=8 June 2010 |access-date=29 June 2019}}</ref>

|-

|{{sort|2001|June 2001}}

Line 817 ⟶ 826:

|{{?}}

|<ref>{{cite news |title=Toshiba's new 32 Mb Pseudo-SRAM is no fake |url=https://www.theengineer.co.uk/toshibas-new-32-mb-pseudo-sram-is-no-fake/ |access-date=29 June 2019 |work=The Engineer |date=24 June 2001 |language=en-UK |archive-date=29 June 2019 |archive-url=https://web.archive.org/web/20190629232051/https://www.theengineer.co.uk/toshibas-new-32-mb-pseudo-sram-is-no-fake/ |url-status=dead }}</ref>

|-

|{{sort|2001|February 2001}}

|{{?}}

|4 Gbit

|DRAM

|Samsung

|[[100 nm]]

|CMOS

|{{?}}

|<ref name="stol"/><ref>{{cite web |title=A Study of the DRAM industry |url=https://dspace.mit.edu/bitstream/handle/1721.1/59138/659514510-MIT.pdf |publisher=[[MIT]] |date=8 June 2010 |access-date=29 June 2019}}</ref>

|}