Buyer's guide: A basic guide to CPUs

Going to buy a computer and confused about some of the terms in use? Well look no further; for I’m going to help clear some of those doubts. I’ll be presenting a set of buying guides; one for each sub-section of the computer system.

Section A: Introduction
Today, we’ll talk about the CPU. CPU stands for Central Processing Unit. It is sometimes referred to by the term “processor”. It is responsible for a great deal of the calculations that your computer does. This is what it looks like:


Often however, if you open your computer, you will not see a part like this. It will be covered by a heat sink and/or a cooler whose job is to keep the CPU cool. This looks something like this:



Section B: What am I going to use it for?
This is the single most important question to answer. Correctly answering it will not only ensure that you are able to do what you want to with your computer; it will also ensure that you do not over-spend in buying a CPU that is meant for higher workloads. Some of the terms that I used here may not be familiar to you. In that case, you can first read the section “Terms you will encounter”.

Almost any CPU, even the cheapest models like the Celerons, Semprons, Pentiums, Atoms or any single core processor will do the following well
—Browsing web pages
—Downloading stuff from the Internet
—Facebook, Twitter, Email, Chat
—Playing Youtube videos in standard definition
—Data entry tasks
—Office tasks like document creation, spreadsheets, presentations
—Playing DVDs
—Ripping music (i.e. copying music from CDs to your computer)
—Playing music
—Sharing photos

Do you do many of these things simultaneously? In that case it would be better if you choose a low-end dual core processor. Examples include the Intel i3 series or the AMD A4 series. Such a processor will also help you if
—You want to watch HD (high definition) videos
—You are a casual gamer
—You want to do basic editing of photos

You need to move on to more heavy duty processors like the Intel i5 series or the AMD A6/A8 series if
—You are an avid gamer
—You write software for other people
—You create music
—You edit photos

You need to move to the top of the line processors like the Intel i7 or the AMD FX-8xxx series if
—You perform a lot of advanced tasks simultaneously
—You do video editing
—You run CAD/CAM or other computationally intensive software
—You are into heavy duty photo processing

If you are selecting a processor to power your HTPC (Home Theater PC), then you need to opt for a low power processor that can play HD videos. The “low-power” here refers to the heat the processor generates. The lesser heat generated means that you do not need fans running at a high speed; hence your PC will be quieter which is important in an HTPC

Section C: So which manufacturer should I choose?
You hardly have a choice nowadays. It’s either Intel or AMD. Both of them are good provided you choose a processor suitable for your needs. Currently, Intel tends to be costlier than AMD; however in return you get better performance.

If you are shopping for older second-hand systems you may come across Cyrix or VIA processors. These are suitable for low end tasks.

Section D: Is a faster processor better?
No. In the old days, this was true. Processors with a higher MHz/GHz rating were faster than those with a lower rating. Nowadays, this is no longer true. It all depends on the architecture of the processor. Newer processors with a lower speed rating may be much faster than older generation processors rated at a higher speed.

Section E: Terms you will encounter
When choosing a processor you will commonly come across a lot of the terms below.

Cores, threads, multithreading, hyperthreading
Think of the processor as a cubicle that contains one or more workers. A processor with just a single “work unit” will be called a single core processor. One with two “work units” will be called a dual core processor. One with four will be called a quad core processor and so on…

Given the same amount of time, a cubicle with 4 workers will be able to perform more work than one with 2 workers or a single worker. Therefore in the same amount of time quad-core processors can do more work than dual core processors, which in turn can do more work than single core processors

Now usually, a worker in a cubicle, at any instant of time, performs one task only. Sometimes however, you get these super-efficient workers who can do multiple things simultaneously. They might be eating their lunch with one hand while they write down things with the other hand. CPUs too are like that. Some CPUs are single threaded: this means that the core in the CPU works on one task at a time. Other CPUs are multithreaded: this means that the core in the CPU can handle multiple tasks (usually 2) at a time. Thus you can have a quad core processor with eight threads. This means that each of the 4 cores can do 2 things simultaneously: giving rise to 4×2=8 threads.

Multithreading is the generic term for this ability of a core to do multiple things simultaneously. Intel has coined the term “hyperthreading” to refer to this feature in their processors.

Clock speed, Turbo Boost, overclocking
Clock speed or operating frequency refers to how fast the processor operates. It is measured in Hz (Hertz). Nowadays, the clock speed is commonly mentioned in GHz (Gigahertz i.e. 1000000 Hz). In older computers you will see it mentioned in MHz (Megahertz i.e. 1000 Hz)

Turbo Boost is a technology from Intel, which allows the CPU to operate at a higher frequency than its base clock speed. The maximum value this can take depends on the temperature, power, current and the number of active cores. For example a processor like the i7-920XM has a base clock speed of 2 GHz. However, one single core of this quad core processor can even be made to operate at 3.2 GHz via the Turbo Boost technology. This is a form of “overclocking” i.e. making the CPU operate faster than its base frequency. This type of overclocking is automatically handled by the operating system.

There is another form of overclocking that is possible. This type of overclocking is made possible by modifying various system parameters such as increasing the voltage that is supplied to the CPU. This makes the CPU work faster. However, the CPU will also heat up faster and therefore needs to be appropriately cooled. Note: This second type of overclocking will immediately void the warranty and if used inappropriately can damage your CPU or reduce its lifespan.

Amongst the newer processors, the K series of processors from Intel allow overclocking while the non K ones are limited in their overclocking ability.

Architecture: 32 bit v/s 64 bit
The CPU operates on data and instructions which are fetched from the RAM (i.e. main memory) of the computer. This RAM is made up of a number of small cells. There are special components called address registers inside the CPU which are loaded with the address of the memory cell. Data is then fetched or sent to that memory cell. How wide this “address register” is, determines whether the CPU is 32 bit or 64 bit. A 64 bit address register is wider than a 32 bit one. Hence the 64 bit CPU can refer to more memory cells. Therefore a computer with a 64 bit CPU can have more memory as compared to one with a 32 bit CPU.

Older CPUs were typically 32 bit; most modern ones are 64 bit

Cache
As mentioned previously, CPUs fetch data from RAM. This takes time. Therefore CPUs have a special high speed memory inside them called the cache. Certain parts of the CPU are responsible for pre-fetching data from the RAM and putting it inside the cache. Other parts of the CPU then operate on the data inside the cache instead of having to fetch the data from the RAM. This speeds up operation.

Caches are typically multilevel. There will be a small amount of L1 (level one) cache followed by a larger amount of L2 cache inside the CPU.



Integrated graphics, APUs
In older computers, interaction with your monitor was controlled by a specialized chip or card called the graphics chip or graphics card. Today’s CPUs have all the graphics control circuitry inside the CPU itself. Thus we have the term, integrated graphics i.e. the graphics control circuitry is built into the CPU. Examples of these are the Intel HD 2000, HD 3000, HD 2500 & HD 4000 graphics which can be found in the i3/i5/i7 family of processors. AMD uses the term Fusion CPU or APU (Accelerated Processing Unit) to refer to the package containing both the CPU and the GPU (which is responsible for the graphics control circuitry). In modern CPUs, the integrated graphics are powerful enough to let you play most modern games at low to medium level settings. If you want even better quality, you can of course buy a separate graphics card.

TDP
This stands for Thermal Design Power. It refers to the amount of heat that the CPU generates. CPUs with a lower TDP are better
—In laptops, because the battery lasts longer
—In HTPCs because the cooling fans do not have to rotate so fast to cool the system and hence the system is quieter
A lower TDP also usually means lower electricity bills.

Process
This refers to the semiconductor process step used to manufacture the CPU. It is typically measured in nm (nano meters). The latest generation processors from Intel are 22 nm processors. Older processors manufactured at 32 nm or 45 nm are also fairly common.

The smaller the manufacturing process used, the easier it is to fit more components on the CPU.

Socket Type
This refers to the socket on the motherboard, in to which the CPU fits. A CPU is always designed for one particular socket only. Some examples of sockets include:

From Intel, in increasing order of age:
LGA 2011: Used for the extreme edition Intel Sandy Bridge processors
LGA 1155: Used for the mainstream Sandy Bridge and Ivy Bridge processors
LGA 1156: Used for older generation i3/i5/i7 processors
LGA 775 & Socket M: Used for older processors like the Pentium 4, Pentium D, Core Solo, Core Duo & Core 2 Duo processors



From AMD, in increasing order of age
Socket AM3+: For the FX series
Socket FM1: For AMD Llano
Socket AM3: For Phenom II, Athlon II, Sempron
Socket AM2+: For Athlon 64, Athlon II, Phenom & Phenom II


CPU family
Depending on the manufacturer CPUs may be categorized into several different families

Intel uses the brand name Intel Core to refer to its medium to high end range of processors. Celeron & Pentium are the brand names used to refer to the entry level processors that Intel offers. Atom processors from Intel are extremely low powered ones; these are suitable for used in ultra portable computers like netbooks.

The Intel Core family includes the i3, i5 and i7. It also includes the older models like the Core 2 Solo, Core 2 Duo, Core 2 Quad & Core 2 Extreme processors.

The latest, third generation i3/i5/i7 models have the model number 3xxx. This includes both the 22 nm Ivy Bridge lineup (e.g. 3770, 3770K) and the Sandy Bridge-E lineup (e.g. 3960, 3820). The older second generation i3/i5/i7 belonged to the mainstream Sandy Bridge family and used 2xxx as model numbers (e.g. 2600K, 2500).

AMD’s current offering includes Bulldozer CPUs (FX series) and Llano (A4, A6, A8) CPUs. Older models include the Athlon 64, Opteron & Phenom which in turn were preceded by the K5, K6, Athlon, Duron & Sempron models