Technical terms explained
- CPU stands for "Central Processing Unit" and is the "brain" of a computer. This is the unit
that runs all the programs in a computer. A CPU's speed is measured in Hz (MHz, GHz), (see below).
The more MHz the faster is the CPU, however "MHz" only tell about half the story. The CPU's architecture tells the other part.
Since 2003 or so, the CPU frequencies have not been rising as fast previous years. Just ramping up the frequency without making the
architecture more efficient is not a good idea as several problems come into play. For example, the power consumption (and heat dissipation)
rises to alarming levels. So rather than blindly focusing on GHz, both AMD and Intel have decided to increase the parallel processing capabilities of
their CPUs by including several cores into the same CPU (both companies released dual-core chips in 2005). This effectively gives you two CPUs in one.
Future versions will add even more cores into the same "package" (see below).
- Dual-core means that a processor (like a CPU or GPU) has two identical cores on the same die (In this case "die" is the
name of the small piece of silicon on which there are millions of transistors that make up the functionality of [for example] a CPU.)
The purpose of this is to (in theory) double the performance of the CPU without increasing frequency.
Dual processor work stations have existed for many years, but dual-core means that there is only one CPU in your computer, but within that
package there are actually 2 CPUs running in parallel. Recent developments in the miniaturization of the manufacturing process have allowed
the CPU makers to actually fit two cores on roughly the same physical space as previously occupied by only one core. As the manufacturing
process evolves, it will be possible to cram even more cores onto the same die. That technology is referred to as multi-core.
- MHz stands for Mega Hertz, meaning "Millions of pulses per second". A CPU running at 500MHz performs 500 million
"pulses" each second. If the CPU can execute one instruction each pulse, then a 500MHz CPU can execute 500 million
instructions per second (MIPS). GHz stands for Giga Hertz, meaning "Billions of pulses per second".
- Hard Drive (HD) is a unit that stores information magnetically. Once something is stored on the Hard Drive, it stays
there until it is deleted. It does not need power to maintain the stored information. Typically the Hard Drive is where you
have all your programs including the operating system, installed. When you start a program it is read from the Hard Drive
and temporarily stored in RAM. (see below)
- RAM stands for Random Access Memory. This is the "place" where programs that are currently run
are held. When the CPU is executing a program it fetches information from RAM and it needs information fast or it's full
performance will not be utilized. The faster the CPU, the faster memory is required to "keep up" with the CPU. The functional
difference between RAM and a HardDrive is that RAM is MUCH faster, but unlike a Hard Drive, "things" in RAM are only stored
temporarily. If you switch off the power, the information stored in RAM will be lost.
- Graphics card is the piece of hardware that handles the graphics in a computer system. In the old days the GFX cards
(GFX is short for "Graphics") could only draw 2D (2 dimensional) graphics. These days, all available (new) GFX cards are also
capable of drawing 3D (3 dimensional) graphics. A CPU can basically do all that stuff also, but *much*, *much* slower, which is
also why GFX cards are often referred to as "graphics accelerators".
(The 3D graphics are still projected as 2 dimensional graphics on your monitor, though. To experience "true" 3D you will need
special glasses or a good old-fashioned VR helmet and some software to make use of it.)
- MB stands for Mega Byte or "a million bytes". In the old days it was written "Mb" but there was a
confusion between MegaBit and MegaByte so it was decided that MB should be short for MegaByte. One byte is 8 bits. If you
do not know what a byte is, you should read this lesson about the binary
notation.
- 3D Graphics or 3D GFX refers to 3 dimensional graphics. Instead of drawing graphics on a 2 dimensional plane
3D accelerators can calculate the mathematics required to draw 3 dimensional graphics that look correct on a 2 dimensional
screen.
Modern GPUs (Graphics Processing Units) have programmable (pixel and vertex) units (shaders) that can be used to increase the visual realism.
Simply put: Pixel shaders can be programmed to render realistic surfaces while the vertex shaders calculate the transformations you apply to the
geometry that describes the scene you see.
- On-die really has to be put into a complete sentence to make sense. "On-die cache" for example means that the cache
(see below) is "mounted" on the same die as the CPU. Next question: what is "die"? In this case "die" is the name for the small
piece of silicon on which there are millions of transistors that make up the functionality of [for example] a CPU.
For performance & price reasons, the CPU makers are including L2 cache on the same die as the CPU itself as opposed to off-die
L2 cache where the L2 cache would be on a chip separated from the CPU itself. On-die L2 cache offers better performance than
off-die, but makes the fabrication process a little bit more difficult.
- L2 Cache L2 stands for "Level 2" and "cache" is a special kind of very fast but expensive memory that acts as a
buffer between the CPU and the ordinary RAM. The only reason to have "cache" memory is performance. Ordinary memory (RAM) is
very fast compared to Hard Drive but not fast enough to keep up with a modern CPU. While a common CPU is running at 2GHz the
memory might only be running at 400MHz. In a modern CPU the on-die L2 cache is running at the same speed as the CPU itself,
in our example that would mean 2GHz. L2 is as i said, Level 2 which suggests that there is a level 1 cache. Ever since CPU's
started to use cache memory, the L1 cache has always been on-die. Typically the L1 cache is divided in two equally sized part:
Data cache & Instruction cache. (I will not go further into detail explaining what these are.)
- SDRAM Simply stands for "Synchronous DRAM" Which in turn is short for
"Dynamic Random Access Memory".
It is enough to say that is a enhancement of the old DRAM and the most common memory in use today.
SDRAM comes in different "flavours" however. For example "DDR" or "DDR2" SDRAM. See below (DDR) for more info.
- DDR stands for "Double Data Rate". This means that the the memory transfers data not only once
per "pulse" (clock cycle) but twice. (It transfers data both on the rising & falling edge of a clock cycle instead of only on
the rising edge as ordinary SDRAM does) In essence, a 200MHz SDRAM DIMM using DDR results in the the same performance as
a 400MHz (SDR=Single Data Rate) SDRAM. So if you see the label "400MHz" DDR SDRAM it's basically a 200MHz SDRAM using DDR
technology. Simply put, DDR technology doubles the speed/performance of SDRAM.
DDR2 is the successor to DDR2, but the differences are not as huge as one might expect. As I write this (late 2005) the
performance difference between equally clocked DDR and DDR2 memory is negligible. However, in the years to come, DDR will slowly be phased
out, replaced by DDR2 which is designed for higher clock frequencies (=more bandwidth).
DDR3 is another variation of the DDR SDRAM technology. It is capable of even higher clock speeds than DDR2 and is favored by graphics card makers
as local graphics memory. (The GPU has it's own graphics memory much in the same way as a CPU has its main RAM.)
- DIMM is short for "Dual In line Memory Module" it is a 64 bit version of the 32-bit
SIMM (Single In line Memory Module)
- RDRAM means "Rambus DRAM". Rambus is actually the name of the
company.
RDRAM is possibly the most controversial (computer related) product to come out in the last couple of years. Creating
headlines on the web on almost daily basis during 2000, 2001 it now (2005) seems RDRAM is only going to be used in gaming consoles.
All major CPU manufacturers now only support DDR/DDR2 SDRAM and it doesn't not seem likely that any of them will offer RDRAM support
in the foreseeable future.
- RIMM This package is unique to RAMBUS memory. For more info click
here.
- AGP is short for Accelerated Graphics Port. This is a bus, that was designed specifically for
graphics cards. Unlike the old PCI bus, AGP can access all RAM and offer better bandwidth. The main usage for that is to
"stream" textures from RAM via the AGP port to the Graphics Card. This is because GFX cards usually have a limited amount of
memory. Note however that AGP has been replace by the new PCI-Express architecture (see further below).
- PCI, short for Peripheral Component Interconnect. This is the common expansion bus in which
most expansion cards fit. New computers use PCI as the primary bus for external peripherals such as sound-cards, network-cards,
SCSI (SCSI is a different kind of controller used for Hard drives etc.etc..) controllers and so on.
The USB (Explained below) bus is complementing the PCI bus and is used for peripherals that do not require much bandwidth. The AGP
bus is used for graphics cards but older GFX cards were made for the PCI bus.
(For more info about the PCI bus, click here.)
- PCI Express, is the new bus that is replacing the aging AGP interface. The first incarnations of the PCI Express bus
double the bandwidth of AGP 8x and higher revisions will probably double the bandwidth in the same fashion as newer AGP revisions doubled
the bandwidth over the older version. PCI-Express also allows the signal to be split. This means that a motherboard can have more than one PCI-Express slot.
Several existing technologies already use this feature and the result for the end user is potentially much better graphics performance.
- ISA, Industry Standard Architecture is an OLD 16-bit bus architecture. It is since long
replaced by the PCI bus, but old computers might still have at least 1 ISA bus on the motherboards. It has been several years since
any motherboard used this interface, so you need not remember this one ;-).
(For an technical overview of ISA click here.)
- USB (Universal Serial Bus) 2.0 Is the newest expansion bus on todays computers. It is designed to
replace the old serial & parallel ports as well as it's predecessor, USB 1.0. (And to some extent the ISA & PCI bus.) All new computers
support the USB 2.0 bus and it is used for peripherals such as keyboards, mice, removable memory, external Hrd Drives, CD storage devices and so on..
(USB 2.0 has a maximum transfer rate of 480 Mbit/s.)
(For more extensive information about USB, click here.)
- Serial ATA was designed to replace the old parallel IDE interface.
Configuration of Serial ATA devices is much simpler, with many of today's requirements on jumpers and settings no longer
needed (no master/slave settings).
It uses physically thinner cables than IDE, and it is supposed to be more reliable than current Parallel IDE standards.
- BIOS stands for Basic Input Output System.
(Click here for more information about BIOS.)
- EPROM You might know that ROM stands for "Read Only Memory". The "EP" stands for
"Erasable Programmable". An EPROM chip is normally only readable, but can be re-written if necessary.
If you have to upgrade your motherboard you can often download a newer version of the BIOS from the manufacturers web-site.
In those cases the existing ROM has to be erased and re-programmed. (hence, Erasable Programmable Read Only Memory).
(This is in most cases done by booting from a floppy disk with the required software and following on-screen instructions.)
- IDE "Integrated Drive Electronics" is one possible combination, but there are several other
candidates. (It seems that no one can really tell what IDE originally stood for) IDE is the same as ATA. It is the old interface used to connect
HardDrives, DVDs and CD drives to the computer.
For more information about IDE/ATA click here.
|