Inside the Box: MotherBoard CPU and Memory

 Let's look at the computer's motherboard. On modern personal computer motherboards we have first full power connectors, to get electrical power from the power supply. We have sockets to install the cpu. In some cases cpu is directly soldered into the computer. We can see some slots to install the system's main memory, a chipset, which interfaces the cpu with the main memory and the buses , we see non-volatile memory chips containing the system's firmware that it's needed to load the operating system form hard disk( is actually known as the BIOS which stands for basic input output system), we have some cmos memory chips and their batteries, a clock generator which produces the system clock signal to synchronize the various components on the motherboard; we have slots for expansion cards that give access to to the buses managed by the chipset;we have an integrated controller for permanent storage devices which is typically a SATA bus driver and the connectors we need; we see here an integrated controller for keyboard and the mouse,and in legacy computers actually we will find a serial and parallel ports for these devices but all of them have been substituted by USB bus nowadays; and one or several integrated usb bus controllers to connect external devices and we can see here two and as we've seen the current USB standard nowadays is 3.1 . We also see on the motherboard heat sinks and mounting points for fans to take care of the dissipation and excess heat and, in modern motherboards, a lot of this functions that were initially provided with expansion cards are now integrated, like audio and video, we can find graphics controller, sound-card and gigabyte network controller already on the motherboard which used to be with expansion cards. So talking about expansion cards:  expansion cards that  you have to put into expansion slots, is actually the most visible point from a motherboard, you can actually use it to extend the functionalities of the motherboard. So expansion slots are connectors, you can see one here on the slide, where you can put an expansion cards, and the most used cards we have are sound-cards, network cards, graphic cards for example. But, as we indicated, the more and more functions have been incorporated into the motherboard so the importance of expansion slots has decreased because you already have all the functionality you need on the motherboard, you don't need to expand it. So the first expansion slot technology in pc cards was called ISA and then it turned into PCI which was introduced and later it evolved into PCIexpress which is also called PCIe, and in all the motherboard, you usually find PCIe slots and some PCI ones for legacy expansion cards, that's the old technology. You will also find a higher bandwidth PCIe slot to connect an external graphics card in case you need more graphics performance for computer than the one you get which the graphic card it's integrated into the motherboard. So for main memory personal computers use a type of RAM called dynamic ram or dram, and this is packed into memory modules that are called DIMM which stands for 'dual inline memory modules' and the motherboard usually has several slots to add more of these memory modules. So memory capacity of DIMM modules currently installed in personal computers goes from one to sixteen gigabytes. And these modules these dimm modules use technology called Dual Data Rate which is abbreviated by DDR and this is able to exchange data twice per dram clock-cycle,what we've seen in previous videos, and the most recent version of ddr is ddr 4. A lot of motherboards have dual-channel enabled memory controllers and these use separate channel to communicate memory with the cpu, and this theoretically multiplies the data rate exchange by two, and this means you will can do it faster. And in these motherboards memory layout tipically have color-coded dimm sockets so you know which one is which. So chipsets, which is also in a motherboard, is one of the several integrated circuits that manage the interface between the cpu, the main memory and the external devices. So it used to be made out of two chips which are known as the northbridge and the southbridge. The northbridge what it does is link the CPU to very high speed devices, especially dram and graphics controller and the southbridge connects it to the low-speed external buses as hard disks and USB buses. So the north bridge connects directly to cpu using an interface that is traditionally known as the front side bus FSB, and the southbridge connects to the north bridge. In many modern chipsets the northbridge has been integrated into the processor chip and southbridge contains some of the integrated external devices such as ethernet, USB and audio devices. So we have seen a little bit about the components that are present on the motherboard of a personal computer.

CPU

 So let's look inside the box, inside the computer and look at the cpu, the central processing unit. So the central processing unit or cpu is the component of the computer that executes all the instructions of software programs stored in the computer's memory. So the CPU is actually the brain of the computer. And here on the slide you see you see a wafer. A wafer is the way processors are being manufactured silicon wafers and each one of those can include five hundred to a thousand processors at once. And in modern CPUs we can actually find more than two thousand million transistors. This number is actually increasing because the miniaturization process achieves that we can manufacture components that are smaller and smaller every time. And this was actually are outlined already by Gordon Moore and he proposed in nineteen sixty-five that the evolution was going to be that components we're going to get smaller and smaller and he actually predicted that the number of components in an integrated circuit will doubled each eighteen months. Here you can see the theory on this slide. This is known actually as Moore's law and nowadays it's still followed because things are getting smaller and smaller, evidently won't last forever that's impossible and there are some studies that suggest that the final of this law will arrive some about twenty years. So if we look at the processors we have in our computers nowadays are developed with fourteen nanometers technology. So you can see on the slide we had twenty two before, we are now with fourteen and development and research is going towards ten and seven. And this means that with fourteen nanometers technology in a plate of eighty two millimeters we can have near like two billions of transistors, that's a lot. So just to see what is fourteen nanometers exactly, here you can see on the slide some things that we know. Here is a guy, his name is Mark, he's one sixty six meters, here's the fly that is the seven millimeters here's a mite, is getting smaller, a blood cell, here's a virus that is a hundred nanometers and here's somewhere are fourteen nanometers processors that we have seen. So this is really really small and is about a fifty times bigger than a single silicon atom. So the computer with this CPU is capable of doing very complex tasks. But the CPU itself actually only executes very simple instructions. Is the software and the way is written that decices how it does the hard work of converting very complex task into the simple constructions that CPU does. So, on the slide we can see how CPU works a little bit and what does it's able to get data from the memory, perform very simple arithmetic or logical operations like comparison, greater than or equal and it can jump to different parts of a program depending on the results of the comparison that has been done. And it can put some data into memory and nothing more. It's basically that. So, how it works? Here you can see first there is a prefetch unit that extracts the next instructions from memory and then there's decode unit, tha decodes it to obtain information of the operation. So then, it fetches the data that is needed from memory  and put it into the CPU internal memory, this is called the CPU registers. It's the internal memory that is in the CPU. So then there is the arithmetic and logic unit the ALU which is in charge of performing the operation with the data stored in registers and it obtains the results. These results are return later on back to the main memory from the internal registers. So, once this have been done the next instruction is loaded from ram memory. And sometimes before loading new instructions we have to jump to another memory location and execute instructions that are there. Where you have to jump depends on some comparison that CPU has done with the information that is present in the registers. So the control unit actually organizes all this complete process synchronised to a central clock. So the performance of the CPU depends on several factors and one of the most relevant is the speed. The speed of CPU is measured in hertz, which is cycles per second. Herts is not a measure of speed, it actually measures the frequency of the internal clock and usually more hertz means more processor speed, but this is not always like that because there's other factors that actually influence processing speed. So CPU that are used in nowadays computers have clock frequencies in the gigahertz range where one gigahertz is equivalet to one US billion hertz. So the problem is that with high frequency the temperature rises and with miniaturization heat dissipation becomes harder. So you will have noticed that several years ago there was an increment in processor clock speed of new processors, but this increment got stuck. And actually nowadays they are trying to improve the performance using other techniques, as the heat dissipation limit is near of what technology can actually achieve today. So the clock gives a signal with a given frequency and each instruction that the CPU needs to execute needs a specific number of clock cycles to be executed. So let's look at an example, here on the slide we see the clock and we see and we assume that we have an instruction that needs three point five clock cycles to be executed. Here you see three point five clock cycles one two, three and a half, three and a half clock cycles and this instruction needs that. So if the computer works at one hertz, this mean one cycle per second. This means that this instruction needs three and a half seconds to be completed but evidently if it works at two hertz, to cycles per second, this instruction would only need one point seventy five seconds to finish and it would be quicker. So the speed of modern CPUs is measured in millions of operations by second and that's why given the same processor the more gigahertz the faster it is. But let's look at other issues that determine the speed of a processor. Another important feature is word length and this is basically the number of bits that the CPU can receive when it accesses the memory. So you can imagine it as the number of lanes in a highway. When we increase the number of bits that can be transferred simultaneously the performance of the CPU evidently improves. So another characteristic that affects the CPU performance is the number of cores that the processor has. A processor with one core can only execute one instruction at the same time if we had another core and we have two then it can execute two instructions at the same time and different programs can be executed in parallel. And this is obviously results in a processor which is twice as fast. Here the slide we see it a dual core processor which is a processor with two cores. So current CPU for personal computers have four or even six cores. You can see on the slide for example a quad core processor, a processor with four cores. These cores can be combined with another technology that is called hyper threading and hyper threading allows actually to execute two different threads of the same program almost in parallel. So for example each tab of a web browser or each avatar of a video game can run in an independent thread. And as a result a single processor can execute eight instructions in parallel, so this will obviously increase the performance of the CPU. Here you see on the slide a quad core processor that can have like eight threads running at the same time. So in the consumer market there are two main processor manufacturers Intel and AMD. And the products are compatible so we can execute the same instructions in all of them. SO for desk computer Intel has the i3, i5, i7 models with some specific design for laptops and tablets. And equivalent in AMD are the athlon processors. So both brands have very powerful CPU oriented to workstations and servers segments, there are the Xeon and the Opteron  respectively for these sectors. So this is the computer and it's components and we've looked inside the box at the CPU.

Memory
 So let's look again inside the box and let's look at the memory. That is also a very fundamental component of the computer. So the main memory of the computer is called random access memory which is abbreviated to RAM, RAM memory. So according to the Von Neuman architecture the RAM stores temporarily both the instructions of the program that are being executed and the data that these instructions need to execute. So the positions of the memory are like postal boxes, each one of them is identified by number and it contains a sequence of binary digits in it which is called a word and this is of a fixed length. So the access time to any memory location is the same and the access time is independent of the address and this is actually why it's called random access memory, because it takes the same time to retrieve the information from any random address. This is different that in other storage medium like the hard disks or the CD, because actually these have to rotate to arrive to the desired position and then the access time varies depending on the position where information is on the disc. So another important type of memory computer is ROM. This is an abbreviation of read only memory and is a type of memory recorded by the manufacturer and we can only read it and the advantage of it is that the content is never erased. So ROM memory plays an important role in the boot process for computers since the programs it contains are the first ones to be executed and so they check the computer components and load the operating system to continue the starting process and since it is read only, this content is never erase. So there's another type of memory which is CMOS memory and cmos stands for complimentary metal oxide semiconductor (CMOS) and this makes reference just to the material it's made off and this is very low power memory. And so remember that ram memories are volatile, store content temporarily and all the information that is stored is lost when you power off machine. So to keep some information about basic hardware settings as the date and time and things like that a small cmos memory powered by battery can be used to remember these things for the next time we start the computer. So there's several types of computer memory and each one of them has obviously its advantages and disadvantages, like everything in life actually. There's always a trade off between prize, speed and persistence and you cannot for example have a memory that is cheap, fast and permanent at the same time. But you need this three characteristics to make computers run. So this is why memory hierarchies are established in the computer and you put fast, expensive and small storage options close to CPU and slower but larger and cheaper options further away. The first level of the memory hierarchy after CPU registers is actually the cache memory. So if you want to compute to be fast you need fast access to the program instructions that you need to execute and the data that is needed to execute them. So fast memories are expensive so cache memory was created as a mechanism to have faster access to program instructions and most used data without spending too much memory. So the data you would read from memory and put into registers would come from the cache. This is just a small amount of ram memory that CPU can access very fast and it uses an algorithm to select the most frequently used data and store copies of it. So most of nowadays computer processor integrate cache memory in the CPU chip. Having independent caches for instructions and data. Data caches usually are organized as L1, L2 and L3 and this depends a little bit on speed. So the next level of memory hierarchy is ram that we've seen, random access memory and this is known as the main memory or primary storage memory of the computer. It's fast but volatile as I said, you need another level in the hierarchy that is slower but can be permanent and there you have, for example, magnetic and solid-state hard drives that are also known as secondary storage, where you can keep data and it will not disappear. So in most personal computers we have a lot of different applications open at the same time but we're not really multitasking, we're actually switching between. So when there is not enough ram memory left what operating system doess is a clever trick which is called virtual memory. This consists of setting aside a certain portion of storage on the hard disk to act as additional ram, and moving there the content of the memory of the application that loses focus and bringing it back into ram when it gets focus again. So this way operating system can work with a virtual memory space which is much bigger than the physical ram memory it really has. And this is actually one of the reasons why sometimes your computer freezes for a short time when you switch from one application into the other, because actually it's reloading memory content from hard diks into ram memory. And to manage all this process the operating system organizes memory assigned to different applications in pages that can be move between ram and hard disk. And this is called memory pagination and files were RAM content is stored in the hard disk are called pagefiles or swapfiles. So this has been the memory where we've looked inside the box of the computer.

UPValenciaX: ISC101.2x Information Systems and Computer Applications, Part 2: Hardware