beginners guides memory.doc

Upload: wy-teay

Post on 08-Aug-2018

215 views

Category:

Documents


0 download

TRANSCRIPT

  • 8/22/2019 Beginners Guides memory.doc

    1/13

    Beginners Guides: RAM, Memory and Upgrading

    What is memory? Well, let me think... - Version 1.0.2

    Bookmark this PCstats guide for future reference.

    Modern computer processors can perform several billion operations per second,

    creating and changing incredible amounts of data in a short period of time. To

    perform at this level, they have to be able to juggle the information they process, to

    have someplace to store it until it is needed again for modification or reference.

    As a metaphor, the more jobs a technician takes on at once, the more bench space

    they are going to need to place the components they are assembling, and the more

    shelf space they will need to place the finished products. Similarly, computers need

    space to store data while they are working on it, and space to store data that is not

    being worked on, but will be needed in the future. This is provided by RAM (Random

    Access Memory) and hard disk drives respectively.

    Computers have a memory structure which can be easily (if somewhat sloppily)

    compared to the human brain. The hard drive provides long-term memory storage

    similar to our long-term memories, a place where data is put to be permanently

    stored. RAM (Random Access Memory) provides a pallet that the computer can work

    from in normal operation, similar to our short-term memory. It holds information

    that is essential now but may or may not be transferred to long-term memory,

    depending on need.

    Modern processors also include a memory cache, a comparatively small amount of

    high-speed memory which stores the data that is currently being used most often.

    This could be compared to our awareness, the memory that connects one moment to

    the next and keeps us doing what we were doing a second ago.

    Random Access Memory (RAM)

    can be thought of as the short-

    term memory, in the sense that

    once the power is turned off, all

    information stored there is not

    saved. All modern computers

    have hard drives which storedata permanently as magnetic

    information, but even with the

    improved speed of today's hard

    drive technology. Hard drives

    are still too slow to keep up with

    the needs of the processor since

    Figure 1

    http://window.external.addfavorite%28%27http//www.pcstats.com/articleview.cfm?articleID=985%27,%27Beginners%20Guides:%20RAM%20Memory%20and%20Upgrading%27)http://www.pcstats.com/articleview.cfm?articleID=985http://window.external.addfavorite%28%27http//www.pcstats.com/articleview.cfm?articleID=985%27,%27Beginners%20Guides:%20RAM%20Memory%20and%20Upgrading%27)http://www.pcstats.com/articleview.cfm?articleID=985
  • 8/22/2019 Beginners Guides memory.doc

    2/13

    it can operate on considerably more information per second than can possibly be

    transferred to and from the hard drive.

    This is where the need for a fast, short-term memory solution comes in, a memory

    space that provides very fast access for the processor so data can be written and

    read as needed without slowing down the system appreciably.

    RAM fulfills this need, specifically DRAM (Dynamic RAM), the template for all modern

    memory types.

    DRAM consists ofsemiconductor chips arranged on a small circuit board, each

    containing a logical arrangement of cells laid out in rows and columns. These cells

    use a combination of a capacitor and a transistor to achieve one of two states, filled

    with electrons (1) or empty (0), thus allowing binary (digital) information to be

    stored.

    The dynamic aspect of this type of memory is that it needs to be constantly

    refreshed with an electric charge to keep its information stored. When the computer

    is turned off, all data in the DRAM is lost. In all modern desktop computers, DRAM

    can be added directly to the motherboard in the form of memory modules, a circuit

    board with mounted memory DRAM chips.

    Types of memory

    There are three main types of memory in common use today. SDRAM (Synchronous

    Dynamic RAM), DDR-SDRAM (Double Data Rate SDRAM) and RDRAM (RAMBUS

    Dynamic RAM). This article will detail all three, though it should be mentioned at thistime that DDR-SDRAM in its various forms is by far the dominant type in today's PC

    market.

    Beginners Guides: RAM, Memory and Upgrading

    SDRAM

    Synchronous Dynamic Random Access Memory

    SDRAM started life as an

    evolution of the EDO (Extended

    Data Output) DRAM memory

    type, as seen in 486 and older

    Pentium systems. The main

    drawback of these older memory

    Figure 2

    http://www.pcstats.com/articleview.cfm?articleID=985http://www.pcstats.com/articleview.cfm?articleID=985http://www.pcstats.com/articleview.cfm?articleID=985http://www.pcstats.com/articleview.cfm?articleid=985&page=2http://www.pcstats.com/articleview.cfm?articleID=985http://www.pcstats.com/articleview.cfm?articleID=985http://www.pcstats.com/articleview.cfm?articleID=985http://www.pcstats.com/articleview.cfm?articleid=985&page=2
  • 8/22/2019 Beginners Guides memory.doc

    3/13

    technologies was that they ran at a different speed than the rest of the system

    components (asynchronously).

    This resulted in occasional delays, or wait-states while the processor was waiting for

    the RAM to be available to receive data, which in turn reduced the overall speed of

    the system. SDRAM, initially available at the 66Mhz specification, was synchronizedwith the system clock, eliminating any unnecessary wait times. Ideally, as long as

    the SDRAM used was fast enough to keep up with the system clock, it would perform

    the requested data storage or retrieval action within one clock cycle, and be ready to

    receive or transmit data again on the next cycle.

    To give an example, a 66Mhz system clock, as seen in older Intel Pentium II and

    Celeron systems, performed 66.6 million cycles per second, each cycle taking 15

    nanoseconds to complete. The SDRAM would theoretically perform one read or write

    action every cycle. We say theoretically since the actual memory chips used to make

    SDRAM are not generally significantly faster than the older DRAM types, and still

    take considerably more than one clock cycle (generally 5 cycles for SDRAM) to begin

    a read/write action by locating the correct row and column to begin reading or

    writing from. After this first cell is located, subsequent read/write actions on the

    adjacent memory cells are much faster, happening one per clock cycle. This is known

    as burst mode.

    Tying the memory to the system clock enabled memory access to keep pace with the

    increasing speed of modern computers, while also putting pressure on manufacturers

    to increase the quality of the memory to cope with the increasing demands put on it.

    SDRAM memory is commonly available in 66, 100 and 133Mhz speeds, also calledPC66, PC100 and PC133 respectively. It should be noted though, that these values

    do not refer to the speed of the memory itself, but rather the bus and system clock

    speed of the systems it is rated to be used with.

    The memory itself works the same way, and is generally backwards compatible, in

    that higher rated memory (PC133 for example) will work in a lower speed system

    (say 100Mhz) running at the lower speed. You gain no performance benefit from

    using the higher rated memory in this scenario though, and be aware that older

    SDRAM systems are likely to be incompatible with newer SDRAM memory due to the

    memory speed settings that the motherboards may default to.

    Manufacturers have been known to put ratings on their SDRAM of up to 166Mhz even

    though there are no motherboard/processor combinations using SDRAM which run at

    that clock speed by default. They do this to advertise their memory's ability to run at

    higher clock speeds for users who wish to overclock their computers. Since if you

    increase your computer's clock speed, the frequency at which it will attempt to

  • 8/22/2019 Beginners Guides memory.doc

    4/13

    access the memory will also increase, leading to stability problems if the memory can

    not keep up.

    Most SDRAM memory chips are capable of transferring and receiving data slightly

    faster than their rated speed to give some margin for error. SDRAM is typically

    available in a 168-pin DIMM (Dual Inline Memory Module) in capacities from 16MB upto 1 GB.

    Currently, aside from a few motherboards that support both SDRAM and DDR-

    SDRAM, you cannot purchase new systems that use SDRAM. The huge volume of

    older computers on the market ensure that SDRAM modules will be manufactured for

    at least a few more years, however.

    DDR-SDRAM

    DDR-SDRAM

    Double Data Rate SDRAM

    DDR-SDRAM is again an

    evolution, this time of the

    SDRAM specification. As

    the speed of computer

    processors has increased

    in leaps and bounds,

    the amount of data they

    are able to process in a set amount of time has also increased vastly. The recent

    families of processors from Intel and AMD such as the Pentium 4 and Athlon XP are

    capable of several billion operations per second. This is wonderful from a

    performance standpoint if you are looking at the speed of the chip alone, but it

    presents somewhat of a problem for the system as a whole, since it is limited by the

    bandwidth of the memory.

    The bandwidth of the memory is how much data it can potentially transfer in a set

    period of time. Essentially, the faster a processor can go, the faster the memory

    system supporting it needs to be able to carry data.

    To increase its bandwidth, DDR-SDRAM transfers data twice on each clock cycle,

    achieving twice the theoretical maximum bandwidth of SDRAM running at the same

    speed. This does not translate to twice the memory or system performance, since

    the efficiency of the memory (expressed as a percentage where 100% efficiency

    equals one data transfer performed every clock cycle) suffers as the speed it

    attempts to perform operations in increases.

    Figure 3

    http://www.pcstats.com/articleview.cfm?articleid=985&page=3http://www.pcstats.com/articleview.cfm?articleid=985&page=3
  • 8/22/2019 Beginners Guides memory.doc

    5/13

    Despite this, it is still capable of feeding and receiving considerably more data than

    SDRAM, and is a suitable memory platform for modern processors like the AMD

    Athlon XP and the Intel Pentium 4, both of which rely chiefly on various speeds of

    DDR-SDRAM to provide memory support.

    DDR-SDRAM has been in development since the late 90's, and was first introduced tothe desktop PC market in the Geforce video card by Nvidia, followed by AMD in late

    2000 with their 760 chipset for the Athlon processor. It has since completely

    supplanted SDRAM as the memory of choice for the home and small business PCs

    using both Intel and AMD processors.

    In the form of 184-pin DIMM modules, DDR-SDRAM is currently available in a few

    speeds: PC1600 (200Mhz) PC2100 (266Mhz), PC2700 (333Mhz), PC3200 (400Mhz),

    PC3500 (433MHz), PC3700 (466MHz), PC4000 (500MHz), PC4200 (533MHz) and

    PC4400 (566MHz). The first number, for example 'PC2100' represents the maximum

    memory bandwidth the module can provide in Megabytes per second. The Mhz value

    is the clock speed it is certified to operate at. DDR-SDRAM is commonly available in

    64MB-2GB sizes.

    Note that some newer chipsets such as Nvidia's nForce and nForce 2 and the Intel

    I865 use dual-channel memory, essentially accessing two separate DDR memory

    modules at the same time to double the maximum bandwidth. Dual-channel requires

    that memory modules be added in identical pairs to the board. Regular DDR-SDRAM

    can still be used in this case, just be sure to purchase identical modules.

    RDRAM

    RAMBUS Dynamic RAM

    Currently all but extinct in the

    standard desktop PC market,

    RDRAM is a proprietary memory

    standard, developed by the

    RAMBUS company. RDRAM

    originally made a big splash in

    1998 with its adoption by Intel to provide memory support for their high-end

    Pentium III boards and the early Pentium 4 models.

    Sadly for the company, this was closely followed by a protracted series of court

    battles with major memory manufacturing companies such as Infineon and Hyundai

    over alleged patent violations among other things. The comparatively high price of

    RDRAM, its early stranglehold on the Pentium 4 processor market, and the

    perception that the RAMBUS company's series of lawsuits might well drive up

    Figure 4

  • 8/22/2019 Beginners Guides memory.doc

    6/13

    conventional SDRAM and DDR-SDRAM prices if they succeeded combined to make

    RDRAM rather unpopular with the home user and PC enthusiast markets.

    RDRAM failed to decrease appreciably in price, so Pentium 4 chipsets supporting

    SDRAM were introduced by Intel in 2001 to attract the lower-end market. SDRAM did

    not deliver the necessary performance for the Pentium 4, so Intel introducedPentium 4 motherboard chipsets supporting DDR-SDRAM in 2002, all but eliminating

    RAMBUS memory from the home and small business PC market. There have been

    some signs of a resurgence in RAMBUS fortunes recently, most notably the January

    overturning of a court ruling against them in favour of Infineon, so who knows what

    the future holds.

    In design and operation, RDRAM differs considerably from SDRAM and DDR-SDRAM

    memory in several ways. First of all, RDRAM transfers only 16 bits, or 2 bytes of data

    at a time, as compared to SDRAM/DDR-SDRAM's 64-bit data channel, but it

    transmits those 16 bits at a considerably higher frequency, 400Mhz for basic PC800

    RDRAM. Also, RDRAM transmits data twice per clock cycle just like DDR-SDRAM, so

    the effective data transfer rate starts at 800Mhz. Using the formula

    (Memory frequency) * (# of bits in data channel) / 8 800,000,000 * 16 / 8

    To determine maximum memory bandwidth gives RDRAM a theoretical maximum

    bandwidth of 1.6 GB per second. This gave it a considerable on-paper advantage to

    SDRAM at the time, and was one of the main reasons why Intel decided to use

    RDRAM to launch its Pentium 4 processor line.

    In reality, while RDRAM's high (at that point in time) bandwidth gave it an advantageover SDRAM in Pentium 4 chipsets, the longer time required for RDRAM to initially

    locate memory cells to be written or read from, as compared to SDRAM or DDR-

    SDRAM resulted in it actually performing worse than SDRAM when used with the

    Pentium 3.

    RDRAM's advantage is the speed of the burst or sequential memory transfers after

    the initial delay due to the higher frequency of the memory. Newer RDRAM modules

    can run at 533/1066Mhz and 600/1200Mhz with a corresponding increase in

    bandwidth. RDRAM numbers still trail those of the fastest DDR-SDRAM however.

    Just like DDR-SDRAM, some RDRAM chipsets may be dual-channel, requiring two

    identical chips to function. RDRAM memory chips can generate considerable heat,

    and require a metal heat spreader to help dissipate the excess. RDRAM is available in

    PC800, PC1066 and PC1200 types (where 'PC800' indicates the speed in Mhz after

    taking into account the doubled data transfer rate), at sizes from 64MB to 512MB.

    RDRAM modules are 184-pin packages called RIMMs (RAMBUS inline memory

    modules).

  • 8/22/2019 Beginners Guides memory.doc

    7/13

    What type of memory should you use?

    The answer to the above question is governed by a few factors. First of all, if you

    already have a computer that you do not intend to significantly upgrade, you are

    limited to the type of memory that your current motherboard supports. The chipsetof a motherboard is the collection of chips and circuits that allow the components of

    your computer to work together, and is tied to a specific kind of memory, SDRAM,

    DDR-SDRAM or RDRAM.

    As a general rule, to add x amount of memory to your system, you need a single

    memory module of that size, of the same type and speed that your system currently

    uses (note that type does not mean the brand of the memory, but rather whether it

    is SDRAM, DDR or RDRAM). This information should be easily attained from your

    motherboard manual. There are of course exceptions to this rule.

    Here are some common ones....

    In theory, any system that uses SDRAM can use memory that is rated faster than

    the requirements of the motherboard, as well as memory of the correct speed. For

    example, an older Intel Celeron system would use a 66Mhz clock speed, and thus

    would require PC66 SDRAM, but could also make use of PC100 and PC133 SDRAM,

    which it would simply access at 66Mhz. Note that this is not necessarily possible in

    the real world, as voltages and other factors have changed since SDRAM was

    introduced, and it is a good idea to stick with the recommended memory, but it is

    possible.

    Some recent chipsets support more than one type of memory, generally SDRAM and

    DDR-SDRAM. Consult your motherboard manual for information on this one, but note

    that if your board does support this, you cannot mix both types. Some DDR and

    RDRAM chipsets use dual-channel memory, meaning that two separate memory

    modules on the motherboard are accessed at the same time, doubling the maximum

    memory bandwidth. This requires that the memory be installed in identical pairs on

    the board, rather than single modules as is generally the case. Again, consult your

    manual.

    Also, you will need to verify that you have space on your motherboard to install more

    memory. Since we are going to tell you how to actually install it later in the article,

    the easiest way to do this is to open up the computer and physically check how many

    open memory slots you have.

    Regardless of the type of memory you use, they will look more or less like this.

  • 8/22/2019 Beginners Guides memory.doc

    8/13

    Figure 5

    If you are planning on purchasing a new system, your choice of memory is going to

    be governed by the type of processor you select. PCStats has several excellentarticles on the newest processor and memory combinations to help you research.

    Keep in mind that, as with all other computer components, the price is governed by

    the relative newness and availability of the memory.

    Right now, DDR2100 and PC133 SDRAM are the cheapest and easiest to find, but since

    both Intel and AMD are using higher rated DDR-SDRAM for their newest processors,

    this will change soon. It does seem that DDR-SDRAM of PC3200 speed and above is

    going to be the memory of choice for the near future anyhow, as both Intel and AMD are

    using it for their next generation 64bit processors.

    The Advantage of more memory

    The relation of memory to the actual perceived speed of a computer is always going

    to be a bit nebulous, as it is governed by so many other factors. To make it easier to

    think about, try this. Not having enough memory will slow your system down.

    That's about the easiest way to express it. Think of memory as enabling your system

    to reach its performance potential and you will have the right idea. The processor

    governs the overall speed of the system, but the memory provides it with aworkspace to store information it is using.

    If you have multiple applications running at the same time, demands on the memory

    increase drastically, and if all available memory is used, the system will resort to

    virtual memory, which entails using a portion of your hard drive (the swap file) as

    extra RAM space. As you can imagine, as soon as virtual memory must be accessed,

    http://www.pcstats.com/articlesearch.cfm?SearchValue=&Search=Search&Category=162&CategorySearch=Get+Listinghttp://www.pcstats.com/articlesearch.cfm?SearchValue=&Search=Search&Category=167&CategorySearch=Get+Listinghttp://www.pcstats.com/articlesearch.cfm?SearchValue=&Search=Search&Category=162&CategorySearch=Get+Listinghttp://www.pcstats.com/articlesearch.cfm?SearchValue=&Search=Search&Category=167&CategorySearch=Get+Listing
  • 8/22/2019 Beginners Guides memory.doc

    9/13

    system performance slows down considerably, since hard drives are vastly slower

    than RAM in transferring data.

    Thus, adding more memory is not so much about speeding up your system as it is

    about avoiding slowdowns. Memory upgrades work on the law of diminishing returns

    though. You will see a much bigger increase in performance going from say 64MB ofRAM to 128MB on a Windows 2000 system than you would going from 512MB to 1GB

    of RAM on the same system.

    This is also dependent, of course, on the amount and type of applications you

    commonly run, as well as the operating system you use. Newer versions of Windows

    like 2000 and XP take better advantage of large amounts of system memory than do

    older operating systems like Windows 9x/ME.

    For example, we ran through a couple tests with Bapco sysmark 2002 on a 2.4GHz

    Intel Pentium 4 computer with 256MB and 512MB DDR. In the first round of tests,

    with 256MB, the Internet Content Benchmark came in at 425, Office productivity at

    219. By increasing the memory to 512MB, the Internet Content benchmark increased

    to 452, and Office productivity to 246.

    While not showing a massive increase in performance, doubling the memory on our

    test system gave an appreciable increase in performance when the system is under

    heavy load, especially in the office application portion of the Sysmark test. This is

    consistent with the real life performance benefits you will see by upgrading your

    system's memory.

    Installing Memory modules.

    Before proceeding with this section, please ensure that you have purchased the

    correct type and speed of memory for your computer as specified in your

    motherboard manual.

    Power off and open up your PC.

    All modern RAM is keyed so it can only fit into the RAM slots a certain way. With

    modern motherboards, it should not matter which slot you use, though if they are

    numbered in the manual or on the board, it is always a good idea to go with slot #1

    first. Hold the RAM chip next to the slot so that the indentation(s) on the chip line up

    with the bumps in the slot.

  • 8/22/2019 Beginners Guides memory.doc

    10/13

    Once you are certain of the orientation of your RAM, open the levers on either side of

    the RAM slot and push the RAM chip straight down into the slot until both levers snap

    closed on either side of the chip. This will require some force. If it does not seem to

    be going in with a moderate amount of force, remove the chip and re-insert it,

    making sure that it is exactly lined up with the slot.

    Now power on the computer. Check on the boot up screen and on the properties of'my computer' in your OS to verify the RAM was installed correctly. Everything

    should be good to go once the operating system boots up now!

    Memory Bandwidth vs. Latency Timings

    When Intel released the i865PE/i875P alongside the Intel Pentium 4C processors, the

    DDR memory game changed forever. With a DDR memory controller now capable of

    running dual channel, the Pentium 4 was no longer to be bandwidth limited as it had

    been with the i845 series. Those single channel DDR chipsets, like the i845PE for

    instance, could only provide half the bandwidth required by the Pentium 4 processor

    due to its single channel memory controller.

  • 8/22/2019 Beginners Guides memory.doc

    11/13

    As the new 800 MHz FSB Pentium 4 processors

    allowed users to hit never before seen highs in

    terms of bus speed, many memory

    manufacturers were trying to capitalize on the

    situation by releasing every increasing degrees

    of"high speed"memory.

    Unfortunately, to run the memory frequency at

    the same speed as the FSB (or a 1:1 ratio)

    almost all the high speed DIMM's (Dual Inline

    Memory Module) have to have very lax timings.

    Think about it this way, a car built for drag

    racing can go dead straight super fast, but

    cannot maneuver as well as an F1 race car.

    Likewise, the F1 racer is good in the corners but

    will be left in the dust on the drag strip. In otherwords, today's high speedmemory modules are

    built for one thing only, and that's top speed,

    where timings really aren't considered all that

    much.

    Confused about memory timings?

    When one talks about memory timings they're basically talking about how long the

    system has to wait for the memory to be in a ready state before data is fetched or

    delivered.

    You could think about memory timings as people working at a drive through

    restaurant; you place your order then wait for the food to be ready. The lower the

    timings are, the faster the computer (and quicker your order comes) is able to get

    data from the memory, and the faster the rest of the PC will ultimately be.

    This rule of thumb applies whether you're on an Intel or AMD based system. As for

    why there aren't lower timings then 2-2-2-5, JEDEC (the memory governing body)

    does not think it's possible for current dynamic memory technology to run at 0 or 1.

    When we refer to timings it is common to quote a four digit number separated by

    dashes (ie. 2-2-2-5) The first number always represents CAS (Column Address

    Strobe) Latency as it's usually the most important.

    Next in line is RAS-to-CAS Delay (Row Address Strobe), RAS Precharge and Act-to-

    Precharge Delay (which is always the final, and largest number).

    Figure 6

  • 8/22/2019 Beginners Guides memory.doc

    12/13

    In the picture to the left here, we can see the the timing

    diagram from some Crucial DDR333 memory. If we take this as an example

    for all subsequent memory speeds, I think we should be able to illustrate

    just what all these 'timing' numbers really represent.

    The diagram shows timings of CAS2, CAS2.5 and CAS3 timings (marked asCL=2 for example). Note the vertical dashed lines which indicate a rise or

    fall of the clock signal, since this is double data RAM, there are two such

    points per 'Time unit'.

    CAS latency is the delay between the registration of a read command and

    the availability of the first piece of output data. CAS latency is measured in

    clock cycles. In the last of the three examples, a read command which is registered

    at T0 (Time=0) is not valid until T3 (Time=3).

    With all things equal, a stick of DDR memory capable of running 2-2-2-5 will make

    the computer operating experience seem faster than a DIMM which may only run at

    3-4-4-8. This is because the delay from when the memory receives an instruction,

    retrieves the data, and sends it back out is less.

    Where it starts to get confusing is when you has the choice of buying high speed

    memorywith slow timings. Just about ever PC3700+ rated memory module we've

    seen uses conservative timings after all. If your answer would be to buy fast

    memorywith tight timings, I'm afraid you're going to be disappointed as there are

    no such modules available yet. So, why are we still interested in fast memory with

    slow timings then? Well, the answer goes something like this....

    DDR memory with slow timings

    In highly competitive markets, once a major manufacturer releases a new and innovative

    product, the rest will surely follow close behind. If one manufacturer doesn't follow suit, their

    products are considered 'old tech'.

    As always, everything always boils down to money and that's why we have this dilemma; to

    run faster memory with slower access times, or run slower memory with faster access times.

    There are two trains of thought on this, the first is that high speed DIMM's (like PC4000 DDR)

    can make up for running slower timings by the amount of bandwidth provide the processor.

    Specifically, bandwidth is the amount of data that can be moved from one given device to

    another.

    Figure 7

  • 8/22/2019 Beginners Guides memory.doc

    13/13

    Most DIMM's that run tight timings, such as certain

    PC3200 & PC3500 modules, have to run the memory at

    lower MHz than the FSB. However, when overclocking

    to extreme speeds these DIMM's are bandwidth limiting

    the processor. What I mean by this, is that when the processor requires a great deal of

    bandwidth, the CPU will have to wait for another clock cycle before being filled, as the memory

    is just not fast enough to keep up at the same pace. Having a large pool of bandwidth is great

    when you're working with applications that process a lot of raw data, such as Photoshop or

    databases for example.

    The other point of view is that CAS2-rated PC3200 & 3500 memory can make up for the lack

    of bandwidth because the memory has a lower latency that in effect moves data between the

    CPU and memory faster. Programs that do not require a large amount of bandwidth tend to

    benefit more from quicker data transfers between the memory and the rest of the computer;

    such as games or 3D applications.

    While bandwidth is still very important to the Intel Pentium 4, it's not as important as it once

    was in the i845PE days of single channel memory controllers. Thanks to the i865PE/i875P's

    dual channel memory controller things are much brighter. On average, a system with the

    memory running at 400 MHz (5:4 memory divider enabled) with aggressive memory timings

    will perform 2-3% faster than the system using high speed memory with loose timings.

    While that may not seem like a lot to most people, it can make a world of a difference to the

    enthusiast, especially if you're gunning for that high score in a clan match where every FPS

    counts. Many enthusiasts I know, tend to favour slower memory which allows them to run

    aggressive timings for just this reason.