::::::::: :::::::: ::::::::: :::::::::: :+: :+: :+: :+: :+: :+: :+: +:+ +:+ +:+ +:+ +:+ +:+ +#++:++#+ +#++:++#++ +#++:++#: :#::+::# +#+ +#+ +#+ +#+ +#+ +#+ #+# #+# #+# #+# #+# #+# #+# ######### ######## ### ### ### http://blacksun.box.sk ____________________ _______________________I Topic: I_____________________ \ I Bascis of am I / \ E-mail: I 3mp33j3w I Written by: / > I I < / fu@ckz.org I____________________I Ralph \ /___________________________> <_________________________\ Basics of am 3mp33j3w by Ralph (fu@ckz.org) -AWC (http://awc.rejects.net) 1. Introduction -About this thing -Who should read this? -Brief intro to Mircoprocessors 2. Basics -ALU -Accumulator -Data Register -Address Register -Program Counter -Stack Pointer -Instruction Register -Instruction Decoder 3. Architecture -Buses -Instruction Responses -Fetch -Decode -Execute -Microprocessor Instructions -Modes of Addressing -Inherent Mode -Immediate Mode -Direct Mode 4. Machine Cycles 5. Interfacing -Digital -Analog -Robotics 6. Other 1. Introduction ================ About this thing ---------------- This article will explain the basics of general MPUs. It is by no means in-depth or anything, but it should be enough. I'm taking a fairly lowlevel approach to this thing. As in, I will be talking about the circuits and such. Please note that while I tryed to keep this information as acurate as possible by reading multiple views on something, I can't guarentee that this info is actually all true since I don't know anyone who wants to and knows enough to proof-read it. Also, this thing is very general! Just by reading it you won't be able to distinguish an Apple from an IBM or even a calculator. Finally, even though this thing only explains the very basics of CPUs, it's not really recommended reading material for people who are new to computers. I assume you're familiar with concepts such as the binary numbering system, PC hardware, circuits, etc. Who should read this? --------------------- I have honestly no idea. Maybe low level programmers, or people new to this might be able to put this to some use. This is basicly just background information you should know, but will most likely never actually use it. Kinda the way most people in high school math are thinking. In addition, this information is fairly old, thus some of this shit has long since been replaced or added on to. Brief intro to Mircoprocessors ------------------------------ Inside every computer, there is a little ceramic block of purified crystaline silicon, coated with impurities that give it electronic swithing abilities, and etched with poisonous gases. This little block was developt in 1971 by Integrated Electronics (Intel). The earliest version of this thing was the 4004 chip used in Japanese calculators. One year later, the founders of Intel, Robert Noyce and Gorden Moore, designed the 8008 chip. These things sold rapidly and Motorola produced it's own chip, the 6800. I little later more companies like Signetics (the 2650) and Rockwell (the PPS8) produced their own chips. Still later many companies started releasing a new generations of chips with more functions build into a single chip. Those are for example Fairchild and Mostek's F8, Intels 8048, and my favourite, Zilog's Z80 (still used in GameBoys). The Z80 stands out because it can handle the numbers DEAD and BEEF. Ph33r! Anyway, eventually most of these companies droped out of the race, and only Intel and Motorola kept on developing new processors. Intel started to work with IBM and produced such famous chips as the 8088 and the 8086. Later a number was added with each new version: 80286, 80386, 80486. The 80586 was renamed to Pentium. Motorola on the other hand went to Apple. I don't like Apples, so I have no idea what the current chip versions are. Basics ====== A microprocessor is a chip which must perform at least all following functions: * Receive and store binary data * Perform arithmetic operations * Make Logical decisions * Deliver processed data to output circuits These statements are very general, and along the way, I will explain how each is performed. ALU --- ALU stands for Arithmetic Logic Unit, and is just that. It performs all the arithmetic and logical operations based on binary data being fed to it. It performs additions viva combinational logic based on usually a Full Adder Circuit. Subtractions is done by first negetating the numbers, than adding them. Mulitplication and Division is done by repeaded adding or subtracting respectivly. This k-rad ASCII shows you how the ALU would go about adding two numbers (well it's supposed to anyway): +----------+ | Memory | +----------+ | | Number 2 Number 1 | | +-----------+ +-------------+ |Accumalator| |Data Register| +-----------+ +-------------+ \ / Number 2 Number 1 \ / \ / +---------------+ | ALU | +---------------+ | Result | +-----------+ |Accumulator| +-----------+ NOTE: The accumalator that stores the result is the same that stored number 2, I just couldn't figure out how to express that in ASCII. Accumulator ----------- This is a multi-purpose circuit section. As you can see from my ASCII picture it will not only hold a number to be mulitplied, but also store the result. Usually, the result will than be transfered to some form of memory. Data Register ------------- Temporary storage for data comming from, and going to the bus. Address Register ---------------- Conains the address of the current memory data. Used to access specific points in memory. Program Counter --------------- As the name implies, this thing keeps track of the memory location containing the next instruction to be executed. Whenever a break in the flow of the program (a jump, call, interrupt, etc), the current number in the program counter is stored in on the stack and the new location goes into the program counter. When the break is finished and control is returned to the original routine, the number on the stack gets poped back into the program counter. On most modern CPUs this part would be the Instruction Pointe (IP) register. Stack Pointer ------------- A data structure that records the last program step prior to a break in the routine. There are two basics systems. One is to store the last program step in memory and use the stack-point counter to index the address. The other is the one used on almost all systems today, Last In First Out, or LIFO. Think of it as a stack of books. You put one on top of it, and that one will be the first one to come of next. Putting stuff on the stack is called Pushing, getting stuff from the stack is called Poping. For example, say you have 5 books called A, B, C, D, and E stack on top of each other like this: A B C D E Now you add (push) book F to the stack: F A B C D E If you pop the stack, you get book F back and the stack looks like this again: A B C D E Instruction Register -------------------- The IR holds the binary data fetched from memory during the decoding and executing of that instruction. Instruction Decoder ------------------- The ID sets up the logic which controls the entire MPU. This includes the control lines, the clock, and bus access. Everything happending in your processor is controlled from this section of the chip. Sometimes you may here that everything is controled by the Control Unit, but in fact CU is just a name that summarizes the previously discussed units. 3. Architecture ================ Architecture in terms of processors means the layout of a chip. This might not seem as important, but the architecture is what has so far kept Intel alive and dominant. Every single micrometer has to carefully layed out for best performance and efficiency. Buses ----- Buses are a conductors that transfer binary data in a system. A bus can be external, or internal. External buses are known as System Buses, but we're only interested in the internal buses, as they dictate the architecture of the CPU. The most basic bus design is the Single-Bus System. In this design one one set of conductors is used to connect all the units. 31337 ASCII warning! =======<>============================= | | | | | | | | +---+ | | | | |ACU| | | | | +---+ | | | | | | +---+ +---+ | +---+ |R1 | |R2 | | |ALU| +---+ +---+ | +---+ |===<==| <> - Two-way data flow < - One-way data flow Sorry, didn't feel like making a detailed diagram here, but you can kinda see what I mean. There is only a single major bus connectiong all registers and the ALU. There are other conducting paths branching, but those aren't important. This design is not being used anymore as it is very slow. You can see that doing a simple multiplication would already require to be done on a time-sharing basis. This of course slows the whole thing down by quite a bit. However this design does have one advantage, size. It's very small, thus the next generation of pens will most likely be using it (that was a joke by the way). To get around this speed problem, designers used the Triple-Bus System. In this design 3 seperate main buses are being used. Sorry, no ASCII for this one, just too complex for notpad Actual chips today are even more detailed. They use a decendent of the triple-bus design and the bus can be anywhere from 8-bit to 64-bit in size (that is, between 8 and 64 conducting wires). And again, no ASCII for this one, if you're a good ASCII artist, please contact me and I could send you a .bmp or something. Might not be easy to draw a detailed 8080 chip with characters though :). Instruction Responses --------------------- Every instruction invloves a three-stage sequence, Fetch, Decode and Execute. Fetch In this stage the Program Counter is switched to the address bus and the binary sequence is read by the Address Decoder. Then latched into the Instruction Register. Decode When the fetch was completed successfully, the decoder kicks in. The binary data recieved represents a specific instruction, and will be conditioned in the Programmable Logic Array (PLA) in the decoder. This part of the MPU is a complex network of combinational logic. The sequences that occur in the decoder are controled by an internal ROM called CROM, or Control Read-Only Memory. This thing sets up a control sequence corosponding to the instruction code entered (OpCode). Every MPU has its own unique CROM with a set of OpCodes (Operating Codes) called Instruction Set. Back in the days, the 8080 has 78 different opcodes, today a Pentium has about 700. Some (well most) of these have additional paramters called Operands following them. These are fetched as the opcode is executed. Execute After decoding the instruction, the execute unit directs the synchronzing signals which control the ALU, memory, and I/O circuits. This sequencing and timing is kepy in sync by the Program Counter, which keeps track of every step in the program and control breaks as mentioned earlier. Microprocessor Instructions --------------------------- Every MPU is designed to respond to a fixed set of instructions. As mentioned above, the old 8080 has 78 of them. However, opcodes actually consist of two parts. One part indicates the address mode, the other the actual code. Usually, only 2 bits are used for the address mode part. A 8-bit computer can thus theoreticly have 64 opcodes, but manufactures usually push this limit a bit and include a few variations. Some opcode examples are: Binary Hex Mnemonic Describtion 01110010b 72h JC Jump short if carry flag set 00001100b 0Ch OR AL Perform logical OR on contents of register AL Binary is the raw instruction opcode being used in the Decoding process, Hex is the same number converted to hexidecimal, mnemonic is a more human readable form of that instruction. Mnemonix are what assembly programmers code in. Modes of Addressing ------------------- As stated before, a opcode usually contains two bits telling the MPU what addressing mode to use. I will cover the three basic (and outdated) 16-bit address modes, Inherent, immediate and Direct. Inherent Mode This mode invlovles the MPU chip itself as data is being stored on one or more of the onchip registers during the execution of the instruction. The opcode itself holds all the information required to execute the instruction. Each register is identified in the opcode. For example, say we want to add registers A and B, where A = 000 and B = 111, the opcode for that might be 01000111 The 01 indicates Inherent Mode addressing, the 000 the source register, and 111 the destination register. As you can see, this opcode contains all the information the decoder needs in order to execute this instruction. Steps to execution: 1. a) Fetch instruction b) Decode instruction 2. Execute instruction Immediate Mode This mode involves the opcode PLUS the operand. The operand can contain (at the time this was writen) 8, 16, 32, or 64 bit information. This information can either be an address where data to be used is located, or the data itself. The decoder is told how many more bytes to read in by the opcode. An example of this is a jump instruction. First comes the opcode telling the MPU to perform a jump, the next few bytes contain the address to jump to. Steps to execution: 1. a) Fetch instruction b) Decode instruction 2. a) Obtain operand b) Execute instruction Direct Mode This mode is used when transfering data from the MPU to memory or vice versa. Instructions based on this mode come in three bytes, one for the opcode, two for the (16-bit) address. For example, the following statement moves the value DEADh into the AX register: 10111000 10101101 11011110 10111000 is the opcode for move AX. Since AX is a 16 bit register, the decoder will assume that the next 16 bits are are the value. Steps to execution: 1. a) Fetch instruction b) Decode instruction 2. a) Obtain address b) Decode address 3. a) Obtain operand b) Execute instruction 4. Machine Cycles ================== Timing is a very important thing in MPUs. Many circuits share the same data bus, and must be precisly timed. I briefly mentioned this earlier, and now I'm covering it a bit more in-depth. MPU instructions are executed by the control section and timed by a series of machine cycles. The cycles are subdevided into 5 parts called Clock Periods. A instruction from memory is given 5 clock periods in which it must complete the instruction. If it finishes in less than 5 clock periods, the remaining periods are idled. The following k-rad ASCII shows a instruction fetch cycle: |---Machine Cycle---| +---+---+---+---+---+ | 1 | 2 | 3 | 4 | 5 | +---+---+---+---+---+ |-----------| | | | | | Instruction | Idle Fetch | | Increment PC and decode On a modern computer one machine cycle occurs ever 1193180/65536 of a second, or about 18.2/sec. 5. Interfacing =============== Without interfacing, a MPU is about as usefull as condom made of ice. Even the simplest task requires some sort of way to input data, and some way to get the output. In fact, interfacing is so fundamental to the MPU that a interfacing apparatus is usually already part of the MPU design. Digital ------- When the MPU recieves data, and performs some function, the result is placed on a data bus. These bits can then be read by any imaginable device. All that is necessary is a decoder than can read those bits. 31337 ASCII Alert! +------------+ +------------+ +------------+ | | | | | | | MPU | | Interfacing| | Display | | |----| Chip(s) |----| | | | | | | | +------------+ +------------+ +------------+ 6. Other ======== I cannot be held responsible for anything that happens to you as a result of using this information. You may freely distribute this text as long as you don't change anything. If there's something you think should be changed, contact me first. Send feedback to fu@ckz.org And very important! If you noticed a mistake (speling, grammer, and especially technical!), please contact me asap. As I said before, most of this information is fairly outdated and useless, if you know of a specific change that has happened to anything discussed here, please let me know. Please always get the newest version of this and other tutorials at http://awc.rejects.net as they usually contained updated information, and addons. Greetings to: cozgedal, Alf (the cat eating fury dude), Sad1stick, rpc, moJoe, Lindex, and everyone I forgot to mention. This space intentionally left blank.