In talking about hardware components, we’ve covered computer cases, tools of the trade, monitors, keyboards, mice, hard drives, and DVD drives.  These components can be pretty much purchased independently of the rest of the system components.

We are about to get into the heart of the computer, and talk about power supplies, graphic cards, cpu chips, and finally motherboards.  All of these components inter-relate with one another.  For example, if you purchase a computer chip it mates to a particular type of socket on the mother board, so you need to be careful when you purchase your motherboard that it comes with that socket.

Do not purchase one of these components until after you read about all of them.  Each component has to mate with the other components in your system.  And if your careful, you’ll find that everything will come together neatly, if your not, you’ll be exchanging and returning parts. A royal pain.

At the end of the hardware components, I’ll summarize everything you’ll need to buy with cautions about what you need to make sure you have with the part you purchase.

Nuff, said.

Computer chips are made in “A Clean, Well-Lighted Place.”  Kudos to a 1926 short story by Ernest Hemingway.  I don’t think Hemingway anticipated today’s computer clean rooms and complex nanometer manufacturing technology. After all in 1926 computer chips didn’t even exist.  We’ve come a long way since then.

Two companies dominate the computer chip market, the market leader, Intel, and the “Avis” of computer chips, AMD.  Not that there aren’t other manufacturers, it’s just that your x86 desktop architecture is dominated by these two companies. Intel created the x86 architecture back in 1978 and it is the architecture, by making sure we can always run legacy code, that persists to this day.  You would think that this would make it easy to make a choice. But…if you go to the Intel web site and look at all their processors, you’ll be at a loss to pick the right computer chip for the system your building.  AMD, which also supports the X86 architecture, has less choices, but there’s enough processors to still make it difficult to know which one to choose.

We could go through a bunch of specs, comparing this against that, but you would still be overwhelmed by the choices. We are going to approach this slightly differently.  Since the computer chip is mounted on a motherboard, computer chips and motherboards are intimately connected.  It does you no good to pick out a super fast chip, and not find any motherboards that can run the chip.

Computer chips are mounted in sockets on the motherboard.  In order to increase your  overall system options, we want to pick the cpu socket that has the most motherboards available.  For Intel the current socket of choice is called “LGA 775.” LGA stands for “Land Grid Array” and the socket has no holes, instead it has posts that mate with pins on the bottom of the chip.  The current popular chip that uses that socket is the Intel Core 2 Duo.  Yes, there are faster Intel chips, like the i7 with 45nm techology that takes a “LGA 1366″ socket, and other Intel chip types, like Quad core, but at this point in time in software development, the Core 2 Duo is more than adequate.  You can tell one Core 2 Duo chip from another by their “E” number.  The higher the number, the faster the chip, and the higher the price. E numbers E6850 and below are 65 nanometer technology and E7200 and above are 45nm technology.  The 45nm are denser and faster.  That’s about all you need to know.  I admit this is simplistic, but you can’t go too wrong with this approach.  The chip you pick will be the highest E number Core 2 Duo that fits in with your overall budget.

Intel Core 2 Duo

LGA775 Socket

Let’s take a peak at the AMD chip set.  There are actually two sockets we could choose with AMD.  The AM2+/AM2, or AM3.  Theoretically the AM3 will work in an AM2+ socket, while an AM2+ chipset will not work in an AM3 socket. To be safe, if the chipsets calls for an AM3 socket use an AM3 socket motherboard only.  This corresponds to the Athlon X2 chipset for the AM2+, and Athlon II and Phenom II processors for the AM3 socket.  The chief difference is the Athlon X2 chip set uses 65 nanometer technology and the Athlon II and Phenom II uses 45 nanometer technology. Just like with Intel the higher the number of the chip the faster the performance. Both the AMD chips are Intel chips are fairly equivalent.  See my previous blog entitled, “Quad Core Battleground coming in 2009,”  for some more details.

AMD Athlon II

AM2+ socket

I favor the Intel chips sets at present, but I have to admit I don’t have experience running an AMD chip set. I don’t see any problems in using AMD, as long as the motherboard has the correct AMD socket.

Now which chip to purchase?   This is dependent on how much money you want to spend on your overall system.  You could buy an expensive chip, bear in mind, that you will probably then want to buy an expensive motherboard, and an expensive graphic card. The next thing you know, you may be way outside your budget unless you are planning on building one of those dream computer gaming rigs.  If your outside your budget you can move to the next lower E number chipset or lower number AMD chipset. I think you’ll find unless your building the hottest system you can imagine, a processor running above 2.6 Ghz is probably adequate at this point in time for a home computer.  If your going to run a Microsoft Windows7 you’ll probably want a fairly fast chip as all those 3d graphics on the desktop come at a performance price.

Computer chips use a lot of power in the relatively small wafer area of the computer chip. The computer chip, CPU, offers resistance to that power as it does its work. The result of that amount of power with that much resistance is heat. The same thing happens with a light bulb. You push a lot of power through the filament to produce light, and the result of this is a lot of heat or a very hot light bulb.

The faster the computer chip, the hotter the computer chip. The heat if it gets too hot can start causing problems in the operation of the chip and the motherboard. The way around this is to use a heat sink with a fan placed directly in contact with the computer chip to dissipate the heat. Thus we have the CPU cooling fan. All current computer chips require a CPU cooling fan.

There’s a couple of ways to go. When you order a cpu, it comes with a fan. You’re all set. However, most computer enthusiast don’t think those fans are good enough, and want to order a better fan for their computer for a number of reasons: they anticipate overclocking the cpu, producing more than normal heat; they want a super quiet fan, as the packaged fan usually is noisy; you want your rig to look good, you can get lights with your fan.

If you would like to order a CPU cooling fan. There are some things to consider: The first is noise. CPU fans can be noisy. You can replace the CPU fan with a less noisy fan. Look on the CPU Cooling Fan specification for the Noise Level. The lower the dba the better.

Cooper heat sinks exchange heat better than aluminum.

CPU fans/heat sinks are made for certain computer chips and sockets. If you order a Intel Core 2 Duo, which has an LGA 775 socket, you need to order a fan made for that socket and chip.

CPU fans attach in two ways. There are four holes through the mother board around the chip socket. The heat sink of tha fan sits on top of the computer chip. The fan body either has four push pins that push into those four holes in the motherboard and are held firmly in the holes by expanding the pin after it is through the hole. Or you screw the fan into a plate on the underside of the board. I’ve found the push pin types are flimsy and the pins break easily. I prefer the type where you screw into a plate on the back of the motherboard. “Zalman” is a brand name that uses this technique.

One other thing you’ll need is thermal grease. This is a grease mixture that you place between the CPU chip and the heat sink to enhance the heat transfer.  I recommend a brand called “Arctic Silver.”  It comes in a tube with an applicator for about $7.

Graphic cards have been around since around the 1990′s.  The idea then, as it is today, was to off load the drawing of the display on the monitor from the central processing unit to speed up the computer.  This has not changed in all these years, and yes, graphic cards are very much needed today.  With the continued improvement in computer gaming software, the continued increase in both the resolution and size of monitor displays, and the movement toward hi-definition video, graphic cards have become a critical part of the computer system.

From the start, graphic cards have plugged into slots on the motherboard, and have their monitor output plugs on the card come out the back of the computer.  In the beginning graphic cards used a shared PCI bus to plug into the motherboard. Soon technology began to push the limits of the shared PCI interface, and a new interface came along called, AGP, accelerated graphics port. This provided a dedicated path to the CPU. There were several versions of AGP with different voltages required to run the card, and you had to be careful which version your motherboard could use.  Because of the confusion, and the need for even higer tranfer rates, starting in 2004 a new interface was introduced that all modern graphic cards use called PCI Express, or PCIe, or PCI-E, depending on the specs. There all the same.  To give you an idea of technology improvements. PCI had a data rate of 250Mb/s, AGP had a rate of 500Mb/s, and PCIe has a rate of 1Gb/second.

Graphic cards work by building a display in the graphic cards memory of what you will see on the screen and then transferring that display image to your monitor.  As games got faster, more graphic card memory was needed to build displays in the background, so as one image was being displayed on the monitor, the next image was being built in the graphic card memory.  High definition video and realistic 3d computer games have progressed to the point that two physical graphic cards are used to draw alternate images and to make your display operate like your television.

How do you pick a graphic card?  Well, you have three choices.  Intel has been pushing its graphic chips to be included with the motherboard, and thus manufacturers, like Dell and HP, would not need to provide a graphic card with their computers.  This resulted in a great savings to the manufacturers, and this technology is pushed today.  Buyer beware.  I view this as step back, not forward. Why?  Well these chips do not have their own graphic memory.  They use the computers main memory to draw their images. So your main memory is being used for all your computer operations, which can slow down your overall system performance, in addition, it requires you to load up on memory if you want to have decent performance.  My advice, your building your own computer, don’t get a motherboard with built-in graphic chips that run off of main memory, and hamper your overall performance.  Buy a separate graphic card.

That was choice one, back to the other two choices.  There are two companies that design and make the graphic chips, ATI and NVIDIA. They do not, as their main business, sell graphic cards, but manufacture the chips. Other vendors sell the cards with their chips on it. Most people that I know either are ATI or NVIDIA users, they don’t switch back and forth. I personally have used both, but currently favor NVIDIA.  Both companies use to be independent, but in 2006 AMD the competitor of Intel purchased ATI.  What this means is now AMD can compete with the built-in graphic chip set of Intel. Microsoft just announced they will use the AMD and ATI for their next generation Xbox so the merger is paying off.

You will choose one of the other, either ATI, or NVIDIA, and this will affect which motherboard you purchase, as will see later.  One other wrinkle is the use of two graphic cards instead of one in your computer.  I do not recommend this unless you are a state-of-the-art computer gamer.  One card is usually enough.  ATI calls their dual graphic card set-up, “Crossfire.” and NVIDIA calls theirs, “SLI.”  You’ll see motherboards and power supplies that claim the are either “Crossfire-compliant” or “SLI ready.”  There are pluses and minuses of each.  If you decide to go the two card route, I recommend you research the differences first.

Since the graphic cards plug into the motherboard, you will have to pick a motherboard that fits the graphic card, or visa versa pick a graphic card that fits the motherboard. For example, if you want two graphic cards, your motherboard must have two PCIe slots the proper distance apart on the motherboard.  More on this when I talk about motherboards.

My recommendation is you get one of the latest graphic cards, that fit in your budget, produced by a graphic card manufacturer, like XFX, PNY, or EVGA.  You can get two card compatibility, i.e. SLI ready, and just buy one board with the idea of picking up the second board later, but do some reading first, if you want to go that route in the future.  You want a good chunk of memory with the card.  Usually the higher the number of the card, the better performance, and the higher the price.  Just like computer chips.

There is an exception to this.  NVIDIA recently got to their 9800GX2, and instead of going into GTX10000 for their next products decided to start their numbering over with the 210.  There latest is the GeForce GTX 295.

The number one thing you can do to improve the performance of your computer is to increase your memory size.  I’ve heard this for a number of years, and I beieve it is true. Why?  Let’s start with speed.  Currently DDR3 memory has a transfer rate of around 9,000 Mb per second. The latest SATA hard drives transfer data at about 300 Mb per second.  Memory is about 30 times faster.  This doesn’t take into account disk latency, or the time it takes the disk to arive at the right spot on the disk to transfer the data.

When your computer runs an application, it loads your application into memory from the disk.  If you open another application, another space is set aside in memory for that application, and so on until there is no memory space available.  If another application is loaded after that, your computer starts to do what is called paging.  The oldest application is taken out of memory, and loaded to a swap space on your disk.  As you go back and forth between your applications.  You swap your pages in and out of memory from your swap space.  Needless to say this going back and forth to the disk slows the performance of your computer.  The more memory, the less swapping.  Enough said.

The only problem is when you turn the power off and shut down your computer, you lose everything in memory, while your hard disk retains all your data and applications on the disk surface.  You need both memory to run your applications and hard disks to store your data.

Let’s look at a typical memory name:  OCZ Reaper HPC 4GB(2 x 2GB) 240-Pin DDR2 SDRAM 1066 (PC2 8500) Dual Channel.  What is all this gobly-gook or should I say gobly-geek?

Let me take it apart a step at a time.  OCZ is one of your top memory board manufacturers.  OCZ Reaper HPC is the brand name.  4Gb(2 x 2GB) means there is a total of 4GB of memory in two sticks of 2GB each. 240-Pin is the number of pins needed on the socket of the motherboard. DDR2 SDRAM stands for double-data-rate two synchronous dynamic random access memory.  What’s important here is the type of technology, DDR2.  DDR2  describes the DDR chips themselves, whereas PC2 8500 denotes theoretical bandwidth, and is used to describe assembled DIMMs. 1066 is the bandwidth, the higher the faster the memory.   Dual channel implies that it will work with a motherboard that can transfer data on a dual bus for improve thoughput.

One other way a spec might be presented is: DDR3 2000(O.C)/1600(O.C)/1333 This is usually what you’ll see as a memory spec for the motherboard.  This says the motherboard takes DDR3 memory boards running at one of the following bandwidths: 2000, 1600, or 1333.  The O.C. means the memory can be overclocked.

You always want two matched pairs of memory cards and thats the way their sold. No worries. DDR3 is the latest memory technology replacing DDR2 and they both use a 240-pin socket.  DDR3 is faster than DDR2 for the same clock rate.

Ok, here’s what you need to know.  When picking out memory, you have to match the memory board to the motherboard.  If you don’t you’ll have a heck of a mess as the motherboard may not work and you won’t know why.  If you look at the above motherboard spec, that motherboard requires DDR3 memory, not DDR2, and the bandwidth on the memory has to be one of the three speeds listed: 2000, 1600, or 1333.  You may be able to get away with a lower frequency, but I don’t recommend it. I recommend you look for DDR3 2000 memory for that motherboard.

If you look at the first example that memory is made for a motherboard that requires DDR2 memory that can handle a 1066 bandwidth.  I recommend you pick out your motherboard first, and then purchase the memory called out for by the motherboard spec, not the other way around.