Ever wondered what the processor is like on the inside? Ever wondered how some of the internal structures of the processor translate into some of the arcane specs you hear about in tech circles? This is the article to read. In this article, I will cover some of the basics on how a processor works and define some of the terms which you no doubt will see in your research into the area. Like most subjects in computers, this is a VERY technical subject of which I am only going to scratch the surface here. So, if you would like to learn more, I encourage you to do your own research.
The Basics
A processor is a complete calculation engine that is completely self-contained, with all necessary units compiled onto a single chip. If you ever look at the internal wiring diagrams of a modern processor, you will see that a processor is actually a compilation of many smaller units, all inside of the same casing. Each component has its own special purpose that contributes to the overall functionality of the whole. For example, the Arithmetic/Logic Unit (ALU) of a processor performs all addition, subtraction, multiplication and division operations, while a floating point unit handles all very large calculations on floating point numbers. And you have lines in between each component to facilitate communication between the devices.
Processors also have the ability to perform decision-making and instruction prediction. These overall controller units in a processor control the operatins of all of the other components, including the FPU, ALU and the various memory caches and buffers throughout the circuitry. The entire contraption communicates with the outside world through the use of a bus unit. Different processors have different speeds at which the bus unit can operate, meaning the processor can chug around more data at a time the faster and wider (in bits) the bus is.
Here is a brief rundown of some of the internal components defined (the ones I haven’t addressed):
- Registers – temporary storage areas for holding data that is in the middle of beign worked on by the processor. It is the fastest memory in the entire PC. The width (in bits) of the registers in the processor determine its overall rating in bits. So, when you say “32-bit processor”, you are really referring to the size of the registers, and as a result how much data the chip can process per unit of time. The more registers, the more flexible the processor.
- Bus Unit – This is the component of the processor that deals with moving data in and out of the chip and interfacing with the rest of the PC.
- Control Unit – this is the primary component inside of the processor, the “brain within the brain”. It controls and coordinates all of the information flow through the processor and between the other components.
- Floating Point Unit – This is a dedicated calculation engine for computing instructions involding non-integer numbers. In all modern processors, this is integrated into the chip itself. In the older says, it was not and a separate math coprocessor had to be installed to do this.
In order to give you a little imagery to illustrate some of this, I will supply some. To the right you will see an image of a 0.13 micron Pentium 4 die. It is a close-up image, of course, of what the internal core looks like. For an internal operations diagram, you can view this diagram of the core operations of a Pentium processor. It is simplified, but gives you an idea on the various components of a processor and the data flow throughout.
Clock Speeds
Clock speed is basically the maximum calculation speed of the processor. The actual speed depends on various internal factors of the specific chip as well as variations in the manufacturing process. Every time it ticks (you don’t hear the tick) an instruction gets performed. The amount of instructions that can be executed per second is called, creatively, “millions of instructions per second”, or MIPS. MIPS is not the cut-and-dry measurement of processor speed it once was due to the fact that processors today can do an awful lot more than simply crunch instructions, but it still gives a good idea of speed. The clock speed is determined in MegaHertz(MHz), which is a million cycles per second.
As you might have known, hertz is a measurement of frequency of a wave, equal to one cycle per second. A standard sine wave starts at a point, raises, comes back to origin, goes down, and comes back to origin again. This is one cycle of a wave, and is 1 hertz. The internal system clock of a processor works on this basis. All things that occur in a processor are dictated, timing wise, by the internal clock. It used to be that one clock cycle, 1 hertz, would be the equivalent of performing one processing instruction. This is no longer the case with advancements in processor technology. First, some more complex operations require multiple clock cycles. Secondly, a technology called pipelining allows instructions to overlap, meaning that a single clock cycle could involve multiple instructions, or one clock cycle can handle one chunk of several simultaneous instructions. A pipeline in a modern processor has a width, measured in bits. While one clock cycle is being performed by a single instruction decoder and doing one thing, you could have other instruction decoders also performing their own things on their own pipelines. This large, multi-bit pipeline can peform many instructions at the same time. It is similar to comparing a single-lane road with an 8-lane highway. This ability to perform more than one instruction per clock cycle is sometimes called superscalar architecture, and is something that all modern processors newer than the Pentium do. The term superpipelining refers to the multi-tiered pipelines. With each single pipeline doing less work, but many more pipelines lumped together, you can get more work done and increase the speed of the processor.
A note for spec translation: chips are not always called directly by their clock speed rating. For example, the Pentium-200 is a Pentium chip that operates at 200 MHz. Intel’s competitors used a P-Rating (PR) System. The PR system basically compares the chip with the equivalent Intel chip. For example, the AMD K5 was released in a PR166 version. This means that it runs about the same speed as an Intel Pentium 166, although the K5 itself actually runs at 116.7 MHz. Cyrix continued to use this speed rating system until they folded, and AMD decided to abandon it upon the release of the K6. AMD, though, pulled another one on us with the release of the Athlon XP, where they are using ratings like 1800+ and 2000+ to rate their processors. While the uninformed user might think a Athlon XP 2000+ is a 2 GHz processor, it actually runs closer to 1.7 GHz, and is named the 2000+ for marketing reasons because it compares performance-wise with the Pentium IV 2 GHz. So, keep these things in mind when looking at specs.
Like what you read?
If so, please join over 28,000 people who receive our exclusive weekly newsletter and computer tips, and get FREE COPIES of 5 eBooks we created, as our gift to you for subscribing. Just enter your name and email below:
