|3DNOW! Developed by AMD for their K6-2 processor, this was the first Floating point SIMD Instruction set extension for x86
processors. 3DNOW! was designed to improve performance on FPU-intensive applications, especially 3D games. With the introduction of the Athlon Processor, new
instructions were added to 3DNOW! and the name was changed to Enhanced 3DNOW! to reflect these new additions. Finally, starting with the Athlon XP processor,
SSE-support was added and Enhanced 3DNOW! became 3DNOW! Professional.
3GIO 3rd-Generation Input/Output Former codename for PCI-Express used during development of the new technology.
ACPI Advanced Configuration and Power Interface An industry standard for motherboard/device configuration and power management.
Basically, ACPI gives control over most power-management features to the operating system and defines a variety of low-power and sleep/stand-by states for all
sorts of devices.
AMD Advanced Micro Devices. The guys that hopefully made your processor.
AMD64 AMD's 64-bit extension to the x86 instruction set. Formerly known as x86-64. AMD64 is fully backward compatible with existing 32-bit and even 16-bit
APIC Advanced Programmable Interrupt Controller - A motherboard 'feature' that will allow a APIC-aware operating system (Win2K/XP, recent Linux distros) to
assign more than 16 interrupts to your devices to prevent conflicts.
Applebred Codename for the AMD Duron Processor, Model 8. The name 'Applebred' (sometimes also known as 'Appalbred' is a combination of the former codename for
this part which was 'Appaloosa' and the codename for the 130nm Athlon XP Model 8 core which is 'Thoroughbred'. In essence, this Duron core is a 'Thoroughbred'
Athlon XP core with part of its L2- cache disabled, leaving only 64KB of functional L2-cache.
Athlon Brand name for AMD's line of Desktop processors, also name for the very first AMD 7th-Generation processors (K7).
ATX Form Factor used for most of today's Desktop Motherboards. A full-size ATX Motherboard is 12" wide by 9.6" deep (305mm x 244mm). Sometimes used to refer to
the ATX family of boards. The ATX family also includes ATX, Extended ATX, and Flex ATX. See http://www.formfactors.org/, for complete listing of form factors
in the ATX family.
Bandwidth The amount of data (usually in MB/s or GB/s) which can be moved over an interconnect between devices within a certain time, usually one second.
Bandwidth can be calculated using the following simple formula: Physical clock speed (in Mhz) x Width of interconnect (in Bytes) x number of packets of data
that can be transferred per clock-cycle
Example for PC3200/DDR400 memory:
Physical clock speed: 200MHz
Interconnect width: 64-bits = 8 Bytes (8 bits = 1 Byte)
# of data-packets per clock-cycle: 2 (DDR = Double Data Rate)
So, we get: 200MHz x 8 x 2 = 3200MB/s
Barton Codename for the AMD Athlon XP or AMD Sempron Processor, Model 10. It's a 130nm 7th-Generation core with 512KB of L2-cache used on Athlon XP/MP, Mobile
Athlon XP-M and some Socket A Sempron processors. Basically, it's a Thoroughbred-core with an additional 256KB of L2-cache, for a total L2-cache size of 512KB .
Base-Clock Also known as 'Reference Clock' The 200MHz reference clock-signal generated on all AMD 8th-Generation platforms from which most other clock speeds,
such as the processor's core clock or the Hypertransport-Link's clock, are derived.
Beta Pre-release version of a Software. Most features planned for the final version are complete and functional, but being in an unfinished state, the Software
usually still suffers from a couple of bugs.
BGA Ball Grid Array A package-type for ICs. BGA-chips are soldered to a PCB with a number of small solder 'balls' on the bottom of the chip.
BIOS Basic Input/Output System - The BIOS will recognize, setup and initialize all important hardware-components when you turn on your machine. It also allows
the user to manually set a variety of parameters.
BTX Balance Technology Extended - Family of motherboard form factors. Includes Pico BTX (pBTX), Nano BTX (nBTX), Micro BTX ( BTX) and full BTX. BTX was designed
to replace the ATX family by providing better thermal and acoustic properties. See http://www.formfactors.org/for complete listing of form factors in the BTX
Cache Small and very fast memory used to greatly speed up access to crucial data. On modern processors, caches are integrated directly into the CPU-core (die)
and consequently run at the processor's operating frequency. Hard drives or Optical drives also have (significantly slower but bigger) caches for buffering
small amounts of data to optimize data-transfers to the host.
Clawhammer Codename for the AMD Athlon 64 processor, Model 4 (754-pin) and Model 7 (939-pin). It's a 130nm SOI 8th-Generation core with 1MB of L2-cache half
of this 1MB cache may be disabled on certain models, leaving 512KB of functional L2-cache.
CMOS Complimentary Metal Oxide Semiconductor This term usually refers to the small battery-powered memory on every motherboard which holds the user-configured
Cool'n'Quiet A power-saving technology for Desktop processors that was introduced with the AMD Athlon 64 processor. Cool'n'Quiet (C'n'Q) will dynamically adjust
the processor's clock speed and voltage depending on CPU-usage, so the system will consume less power when idle or at low CPU-utilization (e.g. Word-processing,
surfing the web, etc).
CPGA Ceramic Pin Grid Array A package type for ICs. CPGA packages were used for the first Generation of Socket A processors, i.e. Models 3, 4 and 7 as well as
some mobile Model 6 parts. The vast majority of older AMD processors (Am386/Am486/Am5x86/K5/K6) also had CPGA packages.
C PGA Ceramic Micro Pin Grid Array A package type for ICs. C PGA (aka CmPGA or CPGA/mCPGA) packages are used for the first Generation of AMD Opteron
processors (up to revision CG).
CPU Central Processing Unit The Processor, i.e. the IC that does all the calculating, etc. in a computer. The term 'CPU' sometimes used to refer to the entire
case with everything inside in the past, but it is only very rarely used in that sense today.
CPUID An instruction upon which a processor will return a string of data which allows the user (or software) to determine its Brand/Type, core model, core
revision and capabilities. CPUID-utilities can be downloaded for free and provide a means to exactly identify a processor that's installed in a running system.
CrossFire Developed by ATI, CrossFire is a technology that, like Nvidia's SLI, enables the user to link two PCI-Express video cards together for improved
performance. One of the two video cards must be a special CrossFire-card and the motherboard must be Crossfire-Ready as well. So far (Aug, 2005), ATI has only
announced boards using their Radeon Xpress 200 CrossFire chipsets to be CrossFire-ready, but more (non-ATI) platforms may become CrossFire-certified in the
DDR Double Data Rate A technology that allows for two packets of data to be transferred per-clock cycle, instead of just one. Using DDR-technology effectively
doubles the maximum theoretical bandwidth of a device.
DDR-SDRAM Double Data Rate Synchronous Dynamic Random Access Memory A more advanced form of SDRAM which uses a DDR-technology and is therefore capable of
transferring twice the amount of data per clock-cycle. Consequently, the effective clock is doubled, even though the physical clock speed remains the same, e.g.
DDR memory parts running at a 133MHz physical clock are referred to as DDR266 or sometimes 266MHz DDR memory. Common DDR-SDRAM speed grades are PC2100, PC2700
and PC3200 the numbers after the 'PC' reflect the maximum theoretical bandwidth (in MB/s) provided by that DDR memory of the respective speed grade.
DDR2-SDRAM 2nd-Generation DDR-SDRAM This is the 2nd, more advanced generation of DDR-SDRAM. It's internal architecture is slightly different from normal (1st
Gen) DDR-SDRAM, enabling higher theoretical bandwidths. In practice, however, DDR2-SDRAM still has to become a bit more mature to actually show significant
benefits over the previous generation of DDR memory. Comparably high prices, as well as clock speed/throughput and latency issues still prevent DDR2 memory from
replacing normal DDR memory in the mainstream segment at present (Aug, 2005).
DIMM Dual Inline Memory Module DIMMs are small, easily installable PCBs with DRAM-chips soldered on them. A DIMM is often referred to as a 'stick' of memory.
There are different types of DIMMs, distinguishable by their pin-count as well as the position of the notches on their connectors, which are used for different
types of memory (SDRAM, DDR-SDRAM, DDR2-SDRAM).
DRAM Dynamic Random Access Memory Volatile Memory used as system memory on modern computers, although they usually use some more advanced types of DRAM, e.g.
DTR Desktop Replacement A 'class' of mobile processors designed to provide Desktop-like performance on mobile computers. DTR processors usually also have a
power-consumption similar to their desktop-cousins.
Dual-Channel This term refers to the use of two memory-controllers rather than just one, each driving their own memory-channel and providing the full 64-bit
bandwidth of modern DDR- or DDR2-SDRAM. A dual-channel setup effectively doubles a system's available memory-bandwidth. The term is also often used to refer to
the modern 939-pin and 940-pin AMD Processor's 128-bit memory-interface, although this is not technically correct.
Dual-Core This term refers to a processor with TWO full-fledged cores on a single chip, like the AMD Athlon 64 X2 series or the AMD Opteron Dual-Core series,
currently comprising the x65/x70/x75 models.
Duron Brand name for AMDs first line of budget processors, based on the AMD Athlon and Athlon XP's cores, but with only 64KB of L2-cache on all models as well
as slower bus speeds on most models.
ECC Error Checking and Correction A technology used for memory which allows for the detection and correction of single-bit errors as well as the detection
(but not correction) of multiple-bit errors. Like Parity (which was used on older SIMMs), ECC uses one additional bit for each byte (=8-bits) of memory in order
to be able to detect errors, so an ECC-DIMM is 72-bits wide instead of just 64-bits (which makes it more expensive as an additional chip is required on each
DIMM). Performance will decrease slightly when using ECC, so it only makes sense in mission-critical systems like Servers/Workstations...that's why most
desktop- motherboards don't support ECC.
EMI Electromagnetic Interference Electronic devices (such as computers) emit electromagnetic radiation that may affect other electronic devices nearby (e.g.
mess up your TV reception). Don't worry, your computer's case provides adequate shielding against EMI.
FID Frequency Identifier A 6-bit binary number that represents the processor's core-clock multiplier.
FSB Front Side Bus Connection between the Processor and Chipset (and thus the rest of the system) on AMD 7th-Generation (and earlier) processors.
GDDR3 DDR memory for Graphics, 3rd-Generation This is the 3rd generation of DDR memory developed specifically for use on video cards.
GDDR3 memory is typically capable of reaching significantly higher clock speeds and bandwidths than its desktop-cousins DDR- and DDR2-SDRAM. It is, however,
also quite a bit more expensive, especially at the very high speeds required for high-end video cards.
GPU Graphics Processing Unit Specialized processor used on video cards to deliver fast, high-quality graphics.
Hammer Codename for AMD's 8th-Generation of Processors, also see K8.
HT HyperTransport As with HTT, the acronym HT might also be used for HypertThreading.
HTT HyperTransport Technology see Hypertransport. The acronym HTT may also be used for HyperThreading Technology which has nothing to do with Hypertransport
Hyperthreading Hyperthreading enables a single CPU to act as two 'virtual' or 'logical' CPUs. It will only offer improved performance if applications are
specifically optimized for Hyperthreading, otherwise there will be little to no benefits - in some cases, performance may actually be slightly reduced when
running non-optimized applications. AMD processors currently do not support Hyperthreading.
Hypertransport A fast, scalable, serial, packet-based interconnect used by AMD 8th-Generation Processors (Athlon 64/Opteron, etc) for communication between the
Processor and Chipset as well as for inter-CPU communication on Multiprocessor systems (Opteron 200/800 series).
Hypertransport is not limited to AMD Processors, it can also be found on other devices (e.g. North- to Southbridge communication).
IC Integrated Circuit A complete electrical circuit made out of semiconductor material, comprising components (e.g. transistors, resistors, etc.) as well as
interconnects between all these components. Modern silicon chips like microprocessors, but also DRAM or flash memory are all ICs.
JEDEC Joint Electron Devices Engineering Council An organization responsible for the standardization of semiconductor devices, most importantly for us
K5 Brand name for AMD's Family of 5th-Generation Processors.
K6 Name for AMD's entire Family of 6th-Generation Processors, comprising the original K6 as well as the K6-2, K6-III and their respective mobile counterparts.
K7 Name for AMD's entire Family of 7th-Generation Processors, comprising Athlon, Duron, Athlon XP/MP, Mobile Athlon 4, Mobile Athlon XP-M, Mobile Duron as well
as some Desktop Sempron (Socket A only! Processors. The very first (Slot A) AMD Athlon processors were also called 'K7 Athlon', but the K7 name was dropped
after the 7th-Generation Family moved to the Socket A platform.
K8 Name for AMD's entire Family of 8th-Generation Processors, i.e. Athlon 64, Athlon 64 FX, Opteron, Sempron (754-pin) as well as their respective mobile
counterparts. Unlike the K7, they were never officially referred to as K8, however.
LDT Lightning Data Transport former codename for Hypertransport Technology. Although the term is not commonly used anymore, you may still come across it when
making adjustments in BIOS (CMOS Setup).
LGA Land Grid Array - A package type for ICs. Unlike most other package types, this one is completely pinless. However, some kind of pins are still required to
connect an LGA-chip to its socket - they have just moved from the package to the socket.
Longhorn Codename for Microsoft's upcoming Operating System, officially called 'Windows Vista'.
Manchester Codename for the Athlon 64 X2 Dual-Core Processor, Model 2B. It's a 90nm SOI 8th-Generation DUAL-CORE part with 512KB L2-cache per core for a total
of 1MB L2-cache.
Memory Timings There are several important timing parameters for your memory that can be changed/adjusted on most motherboards.
Java has made a nice Guide to Memory-Timings right here. http://forums.amd.com/index.php?showtopic=12017 in our Technology Guides and Tips section. Read it to
find out what things like CAS, RAS, tRP, Command rate, etc. mean.
Morgan Codename for the AMD Duron Processor, Model 7. It's a 180nm 7th-Generation core with 64KB of L2-cache and SSE-support.
Multiplier A number (usually encoded by an FID) with which a base- or reference-clock signal is multiplied to get an output clock speed. On an Athlon 64
processor, the system's 200MHz base-clock is multiplied by the processor's core-clock multiplier to get the actual core-clock speed (e.g. Athlon 64 3500+:
200MHz x11.0 = 2200MHz).
Newcastle Codename for the AMD Athlon 64 Processor, Model C (754-pin) and Model F (939-pin). The core was also used on some Semprons (with 64-bit capability and
part of the L2-cache disabled). It's a 130nm SOI 8th-Generation core with 512KB of L2-cache part of this 512KB cache may be disabled on certain models,
leaving 256KB or 128KB of functional L2-cache.
OPGA Organic Pin Grid Array - A package type for ICs. AMD started to use OPGA packages with the introduction of the Athlon XP processor.
This package-type was used for all Athlon XP processors as well as the newer Athlon MP models and all Socket A Semprons.
O PGA Organic Micro Pin Grid Array - A package type for ICs. O PGA-packages were first used by AMD for some of their Low-power Mobile Athlon XP-M models. The
entire 8th-Generation Family (with the exception of the early 940-pin parts up to revision CG) uses O PGA packages.
OPN Ordering Part Number A Code that is printed/etched on every AMD processor. It allows for the exact identification of the Processor's Brand/Type, core,
revision and specifications (e.g. speed, voltage, maximum temperature, L2-cache size, etc.).
O/S Operating System The most important piece of Software installed on any computer. The O/S loads after the computer is turned on and has passed POST. It
allows other software (applications) to run on the machine and take advantage of the installed hardware which is managed/configured by the O/S.
Overclocking Running a device (e.g. processor, video card, etc) outside of, i.e. faster than its specified parameters. Overclocking may give you better
performance, but it will void your warranty in any case and it may also damage your hardware unless you know what you're doing.
Package The ceramic or plastic/organic Substrate on which an IC is mounted, so that it is protected from mechanical damage and able to communicate with the
outside world via connectors (processors usually use pins) on the package.
Palomino Codename for the AMD Athlon XP Processor, Model 6.
It's a 180nm 7th-Generation core with 256KB of L2-cache used on the first Athlon XP/MP parts as well as on Mobile Athlon 4 processors.
P-State A pre-defined performance state on AMD 8th-Generation processors determining which clock speed/voltage combinations may be used by the processor's
PCB Printed Circuit Board A fiberglass board on which the connections are printed/etched, so no wires are required as an interconnect between components
soldered to the board. Modern PCBs usually have multiple layers, i.e. there are connections running inside of the board which cannot be seen from the surface.
PCI Peripheral Component Interconnect A bus-type, parallel interconnect used to connect various devices to the chipset's Southbridge chip.
PCI expansion slots can still be found on all modern motherboards and are used by today's consumer-level add-in cards (excluding video cards PCI video cards
do exist but are not widely used anymore). Running at 33MHz with a 32-bit width, the PCI- Bus's 133MB/s of bandwidth are not suitable for modern high-bandwidth
devices such as Gb-LAN or S-ATA RAID cards. Consequently, it is about to be replaced by
the more advanced PCI-Express standard.
PCI-E PCI-Express A new, fast, serial interconnect developed to replace the aging PCI and AGP buses. With a peak bandwidth of up to 4.0GB/s, PCI-Express (x16
slots) has already replaced AGP as the main standard for video cards. The smaller x1, x2 or x4 slots are supposed to eventually replace the old PCI-slots
as a computer's main expansion slots. Being a serial interconnect with radically different signaling, PCI-E is incompatible with the old PCI bus. You can NOT
put a PCI card (or AGP card) in a PCI-E slot, it won't even mechanically fit.
PCI-SIG PCI Special Interest Group The standardization body for PCI, PCI-X and PCI-Express.
PCI-X PCI-eXtended A new, faster version of the conventional PCI-standard. Developed before the introduction of PCI-Express as a solution to the
bandwidth-constraints of the old PCI bus, this 64-bit PCI-variant clocks at up to 533MHz, providing bandwidths up to 4.2GB/s more than enough for even the
latest high-bandwidth devices. PCI-X slots are used EXCLUSIVELY on Server/Workstation type devices you will NOT find a PCI-X slot on a Desktop motherboard.
Unlike PCI_Express, PCI-X is fully backward-compatible to the old PCI bus, meaning you can put a PCI card (32-bit or 64-bit) in a PCI-X slot.
POST Power On Self Test A set of basic tests and hardware initializations performed by the system's BIOS right after power is switched on and before any O/S
can be loaded. A successful POST is usually concluded by a single short beeping sound.
PowerNOW! A power-saving technology for Mobile processors that was introduced with the Mobile AMD K6-2+ Processor. PowerNOW! will dynamically adjust the
processor's clock speed and voltage depending on CPU-usage, so the system will consume less power when idle or at low CPU-utilization (e.g. Word-processing,
surfing the web, etc). ALL Mobile AMD Processors released since the Mobile K6-2+ support PowerNOW!. Recently, AMD has added PowerNOW! support to their
Opteron-line of Server/Workstation processors.
RAID Redundant Array of Independent Disks A technology developed to improve both redundancy (i.e. securing against loss of data) and performance by connecting
multiple individual hard drives together to appear as a single drive to the system in which they're installed. There are various 'levels' of RAID, with each
level representing a different configuration of the array of hard disks. If a RAID is present on today's desktop systems, it will in most cases be a RAID 0.
RAID 0 is sometimes not even considered a true RAID-level because of the absence of the redundancy-component: In RAID0, two (or more) hard drives are just
striped together for improved performance. If one drive fails, however, all data will be lost. We have a guide about RAID right here" http://forums.amd.com/index.php?showtopic=10335
in our Tech Guides and Tips section.
RAM Random Access Memory Volatile Memory used as System Memory by computers, see DRAM.
RDRAM Rambus Dynamic Random Access Memory A proprietary memory technology developed by Rambus Inc. RDRAM uses a serial architecture which is radically
different from the parallel approach used on 'normal' DRAM such as SDRAM or DDR-SDRAM. The few RDRAM implementations found on PCs suffered from high latencies
and heat. They also came at a pretty steep price due to RDRAMs proprietary nature.
Registered DIMM also known as 'Buffered' DIMM This type of memory module contains very small bits of extra memory (on small, seperate chips) called a 'buffer'
or 'register'. Control and data signals from the host (memory-controller) are first stored in the buffer or register and then distributed to the actual
DRAM-chips. Due to its buffers, registered memory's performance will be slightly below that of 'normal', unbuffered memory and registered DIMMs are also a bit
more expensive. However, using registered memory will improve stability, especially under high loading conditions, i.e. when using lots of DIMMs or very high
capacity DIMMs. That's why it is predominantly used in Server/Workstation type systems.
RIMM Rambus Inline Memory Module A memory module holding RDRAM-chips. Due to RDRAM's architecture and the heat resulting from its power-management policy,
these were the first consumer-level memory modules that actually required the chips to be covered by a metal heat spreader.
San Diego Codename for the AMD Athlon 64 processor, Model 27. It's a 90nm SOI 8th-Generation core with 1MB of L2-cache as well as SSE3-support Half of this 1MB
cache may be disabled on certain models, leaving 512KB of functional L2-cache.
SDRAM Synchronous Dynamic Random Access Memory A type of fast DRAM designed to run synchronously with the system bus.
SIMD Single Instruction, Multiple Data A means of processing multiple data streams with a single instruction. Using SIMD can and will improve performance, but
only if the software is optimized for the SIMD extensions that the processor is capable of.
SLI Scalable Link Interface Developed by Nvidia, this technology enables the user to link two PCI-Express video cards together for improved performance. Two
cards with identical GPUs must be used ideally, video cards with the same manufacturer and model should be used. At present, SLI will only work with
PCI-Express video cards based on Nvidia GeForce 6600, 6800 or 7800 series GPUs.
SMP Symmetric Multiprocessing An architecture that allows for multiple processors to work on individual processes or threads. Any processor may work on any
task and can access the system's entire memory.
SODIMM Small Outline DIMM A smaller version of the standard DIMM used for mobile devices, i.e. Notebooks/Laptops.
SOI Silicon On Insulator A technology used for silicon wafers (from which processors are made). Basically, an insulating layer is placed between a thin layer
of silicon and a thicker substrate silicon layer. I won't go into physical/electrical details, but the benefit of this technology is that transistors may become
faster while consuming less power. SOI is a key technology that helped significantly in developing high-performance processors while keeping power consumption
and heat at an acceptable level. AMD's entire 8th-Generation Family of processors (i.e. Athlon 64, Opteron,etc.) uses SOI technology.
SSE Streaming SIMD Extensions A Floating Point SIMD instruction set extension for x86 processors. SSE2 and SSE3 are further, more advanced extensions to SSE.
As with all SIMD instructions, Software must be optimized to actually benefit from SSE/SSE2/SSE3.
Sledgehammer Codename for the AMD Opteron processor, Model 5 as well as the first 940-pin Athlon 64 FX processors (also Model 5).
It's a 130nm SOI 8th-Generation core with 1MB of L2-cache and with three active Hypertransport-links (non-Opteron processors only have one active
Hypertransport-link consequently, the additional links are disabled on the 940-pin Athlon 64 FX models).
TDP Thermal Design Power (or Thermal Dissipation Power) This is the maximum theoretical amount of power (in Watts) that a processor may consume and therefore
dissipate as heat. TDP-values are crucial when it comes to designing cooling solutions as they specify the maximum amount of power (=heat) that a cooling
solution must be able to dissipate. Note that AMD's TDP-values are absolute maximum, i.e. 'worst case' ratings. During normal operation