Accelerated Graphics Port (AGP)
A high-speed graphical interface. The AGP is found on the motherboard and allows for direct graphical data transfer between the AGP card and the computer memory with no caching, resulting in increased graphical performance. AGP graphics cards must be from a compatible generation of cards to work in a motherboard. PCI Express technology is a new alternative to AGP technology.
The average time, measured in nanoseconds, that it takes the memory between receiving a request to read or write data at a specific location and completing the read or write operation.
The storage layout of computer memory. The array is a grid of rows and columns, with a single bit of data stored in cells at the intersections of the rows and columns. An array can be pictured as a spreadsheet, with a single bit of data stored in each cell.
Memory that runs at a different bus speed than the system clock. Includes EDO and FPM memory.
Ball Grid Array (BGA)
The amount of data moved between two points in a given time. Although it can be measured in several ways, the most common measurement is cycles per second (often expressed as MHz in the case of CPUs and memory). The width in bits of data multiplied by the cycles give a total data per time measure. Higher bandwidth can be achieved by increasing cycles per time unit or by increasing bits per data cycle.
- A slot or set of memory slots on the motherboard that must be filled with the same amount of memory for proper data transfer width, which is determined by the CPU (i.e., a 32-bit CPU requires a 32-bit data width). Depending on the type of memory, a bank can only require a single module to work properly (such as with DRAM) or can require a pair or 4 sticks of memory (such as with Rambus).
- A sector of memory on a module. Also often referred to as a row. All modules are either single or dual banked. Some motherboards will only accept a certain number of memory banks, regardless of the number of memory sockets on the board.
Ball Grid Array. A type of memory chip (package type) that is directly mounted to the module by solder balls found on the underside of the chip. This direct mounting allows for a smaller chip size. The smaller chip size in turn allows for greater memory densities because of the larger amount of chips that can be mounted on the module, as well as better heat dissipation than with traditionally packaged chips (SOJ and TSOP packages).
A numbering system used in computing that is based on two numbers: 0 and 1. Combinations of these two digits are used to represent data. Also referred to as Base 2.
Basic Input/Output System. An interface routine contained in a chip on the motherboard that controls how the system's hardware is accessed. Can be reconfigured by the user to change the operation (such as hardware boot sequence) of the system. Plays an integral role in overclocking and may allow users to change memory speeds and latency timings. The BIOS is stored in non-volatile memory and does not loose information when the system power is off.
A binary digit. This is the smallest amount of data that is processed by the computer (represented by a 1 or a 0).
A group of records accessed on a storage device, such as a disk drive or USB Flash drive. Blocks are typically groups of bytes. (see byte)
Memory modules that utilize temporary storage areas (known as buffers) to help manage the data sent to the memory by the memory controller. Buffers act as current regulators keeping the amount of current flowing to and from the chips at optimal levels. The purpose of this is to allow for more memory chips on modules by keeping the memory from being overloaded by the chipset. It can also be used to allow more modules to be used in the system. Not all systems can use buffered memory; the type of memory required is dependent upon the computer's memory controller.
The practice of running a system at increased voltage and speed in the hopes of identifying weak components early on. The thought is that any sub-par components will fail early on under these conditions, preventing them from failing later on during critical operation. Burn-ins are usually done when first purchasing or building a system, or when adding a new component. While the value of burning in components is questionable, since it can lead to failures that might otherwise not occur, it has become relatively popular in the overclocking and high-end memory markets.
A common memory-related BIOS (link to BIOS) setting, this controls how many read or write operations are performed in a single burst transaction (high-speed transmission of data blocks). The most common options for this setting are 4 and 8. This is the number of read or write operations allowed per burst. Setting this to 8 will give better performance, as the system will perform more operations per transaction.
The circuitry that connects various parts of a computer and along which data is transferred. For example, the internal bus of a PC connects all of the components to the CPU and memory, allowing data to be transferred among them. The bus consists of two parts, the address bus and the data bus. The address bus sends information about where the data is being transferred, and the data bus actually transmits the information. An important feature of any system bus is its width, which determines how much data can be transferred at one time. A 16-bit bus transmits 16 bits of data simultaneously, whereas a 32-bit bus can transfer double that amount. The higher the bus width, the higher the performance of the system since it is constantly transferring more data. Along with the bus width, the bus clock speed is another important feature. The clock speed is measured in MHz, and the higher the speed, the more quickly data can be transferred. This will also lead to better system performance. Computer memory uses a separate bus speed and bus width from the CPU, which are very important factors in memory performance. An example of memory bus speed is PC133, which has a 133MHz bus speed. Check product documentation for bus speed and width.
A series of 8 bits of data. Bytes are the most common unit of measurement in computing, particularly in the areas of memory and storage. 1,000 bytes equals a kilobyte (KB), approximately 1 million bytes equals a megabyte (MB) and approximately 1 billion bytes equals a gigabyte (GB).
High-speed static RAM (SRAM) directly connected to the CPU, usually between 256KB-2MB. Cache memory operates like a "frequently used data" file for your CPU. The cache dynamically stores and accesses the data your CPU and applications access most often so that it can be more quickly retrieved. The larger the cache, the better the performance of your system, since more data can be retrieved with high speed. There are two types of cache memory:
- Level 1 (closest to the CPU and a very small amount of memory)
- Level 2 (secondary cache, located on the motherboard and much larger than the L1). Cache memory resides in between the CPU and the system's main memory.
Column Address Strobe (CAS)
A signal that directs the memory to access a particular column address on the row-column grid used to store data. Works alongside the RAS (Row Address Strobe) to access bits of data.
CAS latency is determined by dividing the column access time by the memory clock cycle time. Simply put, the CAS latency is the number of clock cycles it takes the memory to respond to an operation request. Lower CAS latency, therefore, will normally give slightly better performance than higher CAS latency. The default CAS latency for SDRAM is 3 and for DDR is 2.5. The lowest possible CAS latency is 2. Low-latency memory is generally only recommended for memory-intensive applications like overclocking, gaming, video, and sound editing. In normal operation it is unlikely that lower latency will show significant improvement.
A BGA (see BGA) chip type used in Rambus and Flash memory.
A set of chips attached to the motherboard that control how some system hardware interacts with the CPU and motherboard. Memory is one component controlled by the chipset.
The rate at which a microprocessor executes operations. The faster the clock, the more operations it can perform. Both the CPU and memory have internal clocks. Clock speed is measured in MHz or GHz.
One half of the memory grid array that stores data. A memory grid is laid out like a spreadsheet, using columns and rows.
A small, removable data storage device comprised of memory chips encased in plastic. Commonly used in digital cameras, cell phones, audio players, printers, and handheld computers. Retains stored data when removed from device and can be transferred to a different device.
Continuity RIMM (C-RIMM)
A blank Rambus memory module that contains no chips. It is used to complete Rambus memory channels in unused memory sockets. In a Rambus system, all empty slots must be filled with C-RIMMs.
CPU (Central Processing Unit)
One of the central components of a system, the CPU carries out the vast majority of the calculations performed by a computer. Can be thought of as the brain of a computer.
DDR (Double Data Rate)
A type of SDRAM that doubles the amount of data transferred per clock cycle by transmitting data on both the rising and falling edge of the clock cycle, which is a wave. This allows the memory to operate at double the effective speed without actually requiring an increase in the clock cycle speed.
DIMM (Dual Inline Memory Module)
A memory module comprised of a printed circuit board (PCB), memory chips, and gold leads. Sends a 64-bit path of data, so DIMMs can be installed individually. This is in contrast to earlier SIMMs, which only sent a 32-bit path of data to a 64-bit CPU, requiring SIMMs to be installed in pairs.
DRAM (Dynamic Random Access Memory)
Computer memory that stores information to be used by the CPU. Dynamic memory can only store data for a short period of time, at which point it needs to be refreshed with a new pulse of electric charge, or the data is erased from the memory. Information is not stored in DRAM when the system is powered down.
A memory module with two banks. (See Bank definition #2).
A development in motherboard and chipset design that allows for increased bandwidth transfer when using two or more memory modules in separate transfer channels. These separate channels allow each module access to the memory controller, increasing throughput bandwidth. It is a good idea to use matched pairs of modules to ensure the best compatibility for dual channel operation.
ECC (Error Correcting Code)
A method of error-detection in memory which checks and corrects single and multiple-bit errors. Memory with ECC capabilities has an extra chip in each bank. The ECC function is performed by the chipset, not by the module.
EEPROM (Electrically Erasable Programmable Read-Only Memory)
A small memory chip that retains data even without power, unlike DRAM, and that can be erased and rewritten many times. Commonly used interchangeably with SPD.
Electrostatic Discharge (ESD)
Potentially harmful discharge of static electricity. Often referred to as "shocking" a component. Even a small amount of ESD can severely damage or destroy computer components. Precautions such as static straps are recommended when working with ESD-sensitive devices such as memory to prevent damage.
A type of memory that retains data after power is removed. Operates both like memory and like a storage device. Examples include Compact Flash, SmartMedia, and PCMCIA, among others. Used in devices like digital cameras, audio players, handheld computers, cell phones, USB Drives and printers.
The physical size and shape of computer components. Some examples in the memory industry are: SIMM, DIMM, 184-pin, 232-pin, etc.
See Fast Page Mode
Front Side Bus (FSB)
The main path for data transfer in a computer, it connects all of the major components, such as the CPU, memory, chipset, and AGP socket. See Bus for more information on how the FSB operates.
Equal to approximately 1 billion bits of data. Abbreviated as Gb (note the lowercase "b").
Equal to approximately 1 billion bytes of data. Abbreviated as GB (note the uppercase "B").
A piece of metal, usually made of a zinc alloy, that attaches to components and is intended to dissipate heat. CPUs require heatsinks, and they are becoming very common on graphics cards and certain motherboard components. Heatsinks intended for higher-performance use and overclocking are made of copper, which dissipates heat more effectively than zinc.
An external casing, commonly seen on memory intended for overclocking and gaming, which is intended to dissipate heat. Heat spreaders are usually made from either aluminum or copper.
HPM (Hyper Page Mode)
The same as EDO.
IC (Integrated Circuit)
A connection of electronic circuits contained in a silicon wafer. Includes components and connectors, and has external connection pins attached to it.
A method of allowing increased memory performance by allowing the CPU to access one bank of memory while the other is recharging. This way the CPU never has to wait while the memory refreshes, because it is always accessing available memory. The system can only take advantage of interleaving if the data accessed are in opposite banks. If the system needs to access data in a bank that is refreshing, the interleaving will have no effect.
JEDEC (Joint Electron Device Engineering Council)
A group that determines the official specifications for memory design, features and operation.
Notches cut into the underside of a memory module that ensure correct installation. Keys prevent the module from being installed backwards in a memory socket, and from being installed in an incompatible system.
Equal to 1,024 bits, although commonly referred to as approximately 1,000 bits. Abbreviated as Kb (Note the lowercase "b").
Equal to 1,024 bytes, although commonly referred to as approximately 1,000 bytes. Abbreviated as KB (Note the uppercase "B").
L1 Cache (Level 1)
Also referred to as primary cache, L1 Cache is a very small amount of memory that is installed directly onto the CPU. This provides easy and extremely fast access to the most commonly accessed data.
L2 Cache (Level 2)
Also referred to as secondary cache, L2 Cache is a small amount of high-speed memory located close to, and sometimes on, the CPU. L2 Cache allows for high-speed access to the system's most commonly accessed data. On newer CPU's the L2 Cache is most often located directly on the CPU, allowing for higher performance.
See CAS Latency.
The metal wires attaching an IC (or memory chip) to the module. Allows for data transfer from the IC to the rest of the system. The process of putting leads on a chip is often done in a separate factory from the semiconductor production.
Equal to 1,048,576 bits, though commonly referred to as one million bits. Abbreviated as Mb (Note the lowercase "b").
Equal to 1,048,576 bytes, though commonly referred to as one million bytes. The most common unit of measurement for memory modules. Abbreviated as MB (note the uppercase "B").
A unit of measuring clock cycles per second. A MHz is equal to one million clock cycles per second.
A device used by a computer for temporarily storing data for access by the CPU. Commonly referred to as RAM, although there are several different types of memory.
See Bank, definition 2.
The data path that runs directly from the CPU to the memory sockets on a motherboard.
The amount of time required for memory to complete an operation such as a read, write, or combination of the two.
A chip on the motherboard that manages the flow of data to and from the system memory.
A type of memory smaller than an SODIMM, these are primarily used in notebooks and sub-notebooks. Available in 144-pin SDRAM, 172-pin DDR, and 214-pin DDR2.
One millionth of a meter. Abbreviated as mm.
One millionth of a second.
One billionth of a meter.
One billionth of a second. Memory data transfer times are measured in nanoseconds.
Equal to 4 bits (half of a byte).
Memory that stores its data even when it has no power supplied to it. Examples of nonvolatile memory include SRAM and flash memory. DRAM is volatile memory, meaning it loses its data when powered down.
A broad software package that controls the operation of a computer, including input and output from devices, use of memory for software applications, organizing files and directories and many other important functions. Examples of operating systems include Windowsâ, Linuxâ, and the Macâ OS. The operating system has a significant effect on memory performance, as well as the amount of memory that can be used.
A type of error detecting that adds a single bit to each byte of data. The parity bit examines the previous 8 bits and determines whether the sum of the byte is even or odd. The system then compares the parity bit to the byte. If they are both even or odd, the system registers the data as correct and continues operating. If they do not match, an error is registered and the system fails.
An SDRAM speed used with 100MHz FSB systems. Runs at a clock speed of 100MHz.
An SDRAM speed used with 133MHz FSB systems. PC133 memory can be used in PC100 systems, as memory is backwards-compatible with regard to speed, but it will only operate as PC100. There are some exceptions to this backwards compatibility rule due to system or BIOS restrictions.
PCB (Printed Circuit Board)
A board made of fiberglass and electrical traces, to which chips and sockets can be attached. PCBs are used with memory, motherboard, graphics cards and many other types of devices.
- A name for the contacts on a memory module that fit into the memory socket. These can be either tin or gold, depending on the type of module.
- Another name for a lead.
To fill a memory socket, either with a true module or with something that completes the circuit, such as a C-RIMM. An empty socket is referred to as unpopulated.
Turning off the computer.
Turning on the computer.
A piece of circuitry on a memory module that sends information about the module to the system, such as speed, size and bank information.
Memory designed for a specific system or motherboard. Proprietary memory may be patented.
RAM (Random Access Memory)
A physical device for storing data for access by a system's CPU. Called random access because the CPU can access data anywhere on the memory at any time, rather than having to store and retrieve data sequentially. Data stored in RAM is usually only stored temporarily for use by the CPU, although some types of RAM will store data indefinitely. Common types of RAM include EDO, FPM, SDRAM, DDR and DDR2, among others.
A type of memory specification once championed by Intel that uses a 16-18 bit bus width to transfer data up to speeds of 1066MHz, making it much faster than SDRAM. Despite its inherent performance advantages it has declined in popularity in recent years and is far less prevalent than its major competitor, DDR.
RAS (Row Address Strobe)
A signal that directs the memory to access a particular row address on the row-column grid used to store data. Works alongside the CAS (Column Address Strobe) to access bits of data.
Another name for Rambus memory.
The process of sending an electrical charge through a piece of DRAM memory to retain the data stored in the chips. If the memory is not refreshed regularly the data is lost. The process is similar to recharging batteries.
The amount of rows (expressed in thousands) refreshed at a time in a DRAM module. Common refresh rates are 1K, 2K, 4K and 8K.
Memory that uses extra chips (known as registers) to delay data passing through it to allow more time for proper transfer. This allows for more memory chips on the module and can protect data integrity. Registered and unregistered memory cannot be mixed in the same system, and many systems cannot use registered memory. System owners should consult their manuals to determine if they can use registered memory.
A piece of Rambus memory.
SDRAM (Synchronous DRAM)
A type of memory that transmits data at the same clock speed as the CPU's internal clock, allowing for faster and more efficient data transfer. SDRAM performs approximately 25% faster than the previous memory technology, EDO.
SIMM (Single Inline Memory Module)
An older type of memory module that typically holds up to 24 memory chips. SIMMs use a 32-bit data transfer bus, so in use with modern 64-bit processors such as Pentiums, users usually must install SIMMs in pairs. SIMMs have since been replaced by DIMMs, which can usually be installed individually.
A memory module that has only one bank. See Bank, definition 2.
SODIMM (Small Outline Dual Inline Memory Module)
A memory module about half the size of a normal DIMM, used primarily in laptops and notebook computers. Comes in form factors of 72-pin, 144-pin (SDRAM) and 200-pin (DDR).
SORIMM (Small Outline Rambus Inline Memory Module)
A Rambus version of the SODIMM. Uses BGA chips and a heatspreader like larger Rambus modules. Used in laptop and notebook computers.
See Serial Presence Detect.
SRAM (Static Random Access Memory)
High-performance memory that does not need to be refreshed and is much faster than DRAM. SRAM retains data even when power to it is turned off. However, because it is much more expensive and bulky than DRAM, it is used almost exclusively for cache memory.
A device used for storing data indefinitely. Examples include hard drives, floppy disks and CD-ROMs.
The process of using virtual memory to temporarily store data when physical memory is full. See Virtual Memory.
Memory that operates at the same clock speed as the system's CPU. See SDRAM. DDR, DDR2 and Rambus are usually also used synchronously creating some confusion.
Memory that does not use buffers to communicate with the chipset. See Buffered Memory for more information.
When all physical memory (i.e., installed memory modules) are being used, the system can simulate more memory by using empty sectors in the hard drive to store additional data. The process of storing extra data on the hard drive and treating it as extra memory is known as swapping. Since hard drive access times are much slower than memory access times (about 60,000 times slower), using virtual memory can severely decrease performance if it is commonly performed.
VRAM (Video RAM)
DRAM used on video or graphics cards. Sends two streams of data, one to refresh and update the image being shown, and the other to communicate with the CPU and change data about the image.
A waiting period between operations for the CPU. Wait states are caused by a difference in the bus speeds between different components. For example, if the memory is slower than the CPU, the CPU will have to wait for the memory to "catch up" before it can transfer data again.
The amount of time it takes the memory to carry out the operation of storing data in the memory once the command is given.
Noticeably improve your system's performance or your money back
100% compatible memory for your system
7 levels of rigorous testing and a lifetime warranty