In today’s ‘Technological world’ Cache Memory is one of the most used parts of your daily life behind your eyes. Every time you switched on your PC or open your smartphones you mainly started using Cache Memory. It plays a crucial role in deciding the performance and speed of a multi-core system. But what is Cache Memory? Among people, how many have some idea about it! In this article, Ins and outs of Cache Memory will be focused in an easy way.
In computer science, memory refers to a hardware which can store data permanently or temporarily. Stored data can be used immediately any time. It can be an electronic semiconductor integrated circuit or a mechanical combination of some magnetic disks or an optical drive as well.
Computer memory can be different types, like RAM (Random Access Memory), ROM (Read Only Memory), Cache Memory, Hard disk, CD disk, SD Card etc. Their working role, speed, size, performance or capacity everything is totally different from each other and equally important to run a PC smoothly.
Cache Memory is a very fast and small storage which is placed between the processor and the Main Memory in the computer motherboard. It is used to store computer data and instructions temporarily and making them more fast and efficient. In general, today’s processor speed is very fast compared to the speed of the main memory of PC. Cache memory can provide information to the processor at a very fast rate by acting as an interface between the faster processor unit and the slower storage unit.
Cache Memory helps the CPU by reducing the required time to transfer information and data between the storage and the processor. It increases the efficiency and performance of the CPU and reduces the cost of time and electrical energy. Data or instructions that are used frequently by the processor, are stored in the cache memory of the CPU for further uses. Thus, cache memory affects the execution time and makes your PC fast.
Speed differences between processor and memory are always a topic of concern for computer engineers. Processors are always dependent on memory for its work. But as the speed of the memory is slower than processor’s, the processor can’t provide the actual speed they can. Because memory takes a long time to move the information to and from the CPU.
In 1980, Average accessing time of a processor was around 130 ns where memories were 124 ns. In 1990 it became processor ~40 ns and Memory ~80 ns. But by 2000 processor’s accessing time become less than 1 ns but at the same time, average accessing time of a memory is 70 ~ 75 ns. This difference is huge.
Roll in the PC
Cache Memory affects the performance of your PC differently depending on the CPU. It is super fast so the more of it, the better as your computer doesn’t have to wait for data and instruction. As when a cache miss occurs, the CPU must then look for information in the slow RAM or storage. And when this happens, MHz are wasted doing nothing as the CPU has no data to work with.
In early PC’s, a common matter with computer’s hardware was that they all were really slow. Processor speed was 8 MHz or less. But it wasn’t a big problem because memories were also slow. But nowadays you have processors of 2.4 GHz to 5 GHz or more (Latest Intel™ Core i9 processor’s speed is almost 5.2 GHz). But PC’s main memory’s (RAM) speed is 1200 ~ 2400 Hz. However, a modern top of the line processor has performed over 1,000 times than that of the original IBM PC. So the importance of cache memory is very high.
Why Cache Memory is so fast but small in size
CPU Cache position in the motherboard is on the same die as the processor. It is so close to the processor than other memories. It is a solid-state memory and not still on the rotating platters as Hard Disk. This advantage reduces the round trip of this memory. Moreover, they need not have to be refreshed while working. It is a waste of cycles and reduced the efficiency of the memory. Another point is the requested data or information has already been retrieved from the storage to the cache memory. So that they don’t have to be bussed over to the processor again.
Another thing is disk drives are mechanical systems that are slower and less efficient than pure electronics device. It makes Cache memory faster than any kind of memory. Cache Memory plays a vital role to increase the performance of the PC. But why it is so small in size? The answer is cache size is at the same time proportional to latency (which makes your PC slow), to manufacturing costs of it and also to energy consumption.
So if cache size is large then it will be slower rather than fast. A huge number of transistors are needed to build a CPU cache. In the motherboard, it is one of the most energy absorbing element among others. Again the larger cache size means a larger area of memory that needs to “search” to find the requested information or data, thus the access time of the process will be much longer. Large cache memory will also increase the die area of the processor
Nowadays the manufacturer prefers smaller cache in the development of newer and developed CPU’s. Because if they use higher cache in the CPU then it will be just a waste, as an example of Intel™ core2 has 12 MB of cache memory, however more developed core i3 uses only 3 MB cache memory. But core i3 has 100 MHz more clock speed than core2. Most modern CPUs have an independent different cache memory of their own.
There will arrive another question isn’t the price of a cache memory too high according to its size? You can get a 4 GB DDR3 RAM at ~ $40 but you have to pay around $500 for a 64MB cache memory. Yes, a cache memory is expensive. But the fact is your RAM speed is around 2400 Hz but cache has a high speed as the processor has. It increases the price of the cache memory.
Cache memory needs lots of transistors, registers and they are built in a different way that makes it faster. But at the same time, it increases the manufacturing cost. The best solution to keep away the CPU from having to wait is to make every element that it needs as fast as it is. Then why RAM, ROM or Hard disk is not made as fast as a processor is.
Because that will be too costly to buy. Even it will cost more than 10 thousand USD to make a 2GB RAM which will have an access time of 0.5 ~ 2.5 ns, as fast as the processor. While the price of a 64 MB cache memory is not more than $500. So cache memory is the best solution which is as fast as the processor and small in size so that the price is not so high.
RAM vs Cache Memory
The memory of a computer is organized in a pattern that is called ‘hierarchy’ and they are placed serially around the processor. Their position is selected considering the role of them, taken the time to access them, cost and capacity. RAM and cache memory both are two important members of this memory hierarchy of the CPU. RAM is considered as the main memory of the computer where the cache is a special memory type used by the CPU.
It seems that RAM and Cache memory are kinds of similar as they both help processor by storing CPU data temporarily for further uses. But there are some fundamental differences between this two. Cache memory is closer to the CPU when compared to the RAM position. Cache memory is much faster in speed and more expensive than RAM. Cache memory management is done fully by hardware devices but RAM is managed by the operating systems of the PC.
The capacity of the RAM memory is larger. RAM can be 16 GB or more but Cache memory is usually made around 3-12 MB. RAMs are divided into two categories. Further, the cache memory is organized as L1, L2 and L3 caches that differ in the performance of the PC, speed of the processor, cost and capacity of it.
Levels of Cache Memory
There are mainly 3 levels of a cache memory on modern CPU cores. They refer to the distance from the CPU where Level 1 is the closest. They are:
- Level 1 (L1) Cache: L1 cache is called primary or internal cache. It is very small and tightly bound to the actual processing unit of the CPU. It tends to be around 4-32 KB depending on the CPU architecture. It can fulfill data request within 3 CPU clock ticks.
- Level 2 (L2) Cache: L2 cache is larger than L1 but it is a bit slower. It needs 20 CPU ticks to fulfill a request. It is generally tied to a CPU core. Its storage capacity is more, from 64 KB to 8 MB. The current CPUs used advanced transfer cache on their processor chip that is one kind of L2 cache.
- Level 3 (L3) Cache: L3 cache is much larger but slower than rest of the caches, though still a lot faster compared to main memory. It is separated than the main processor. Today’s personal computer often has up to 8 MB of L3 cache.
Cache types and specialized caches
There are different types of caches. They all work together for different purposes.
- L1 Cache.
- L2 Cache.
- Virtual Memory.
- Buffer Cache.
- Disk Cache.
- Network Buffer Cache.
- Browser Cache.
- Web Cache.
There is some special type cache memory used by different CPUs. Like:
- Trace Cache: found in Intel Pentium 4.
- Victim Cache: found in Intel’s Crystal-well variant of its Haswell.
- Multilevel Caches: in IBM Power4, AMD Phenom II, Intel Core i7
- Write Coalescing Cache: in AMD’s Bulldozer
In CPU, the processor is at the core of it. And then cache memory, then RAM and at last storage device. Cache is also a type of RAM, but it is a static RAM which is called SRAM. SRAM is faster and costlier than DRAM. Because it consists of flip-flops (6 transistors) to store data, unlike DRAM which uses 1 transistors and capacitor to store data.
When an application starts or any data is to be read/ write, then the data are shifted from the storage device (magnetic device – Hard disk, optical device – CD drive etc.) to DRAM. Whenever any data or instruction is required by the processor, they provide them at a faster rate than slower storage devices.
Although they are much faster, the processor processes at a more fast rate than DRAM. They are not able to transfer the required data or instructions from storage to processor at that rate. So the required data is transmitted to the next level of fast memory, which is known as Cache memory.it provides data to the processor at a very high rate and the processor can work with that data.
Cache memory also stored recent data which is used by the processor, temporarily. So next time if the processor needs that data again, they can provide that data at more high rate because this time they don’t need to collect it from slower DRAM or Storage device. It makes the overall process faster.
Cache memory mapping is the way in which CPU map or organize data in the cache memory. Cache is mapped every time when the data is to be used by the Processor. This is done by increasing the efficiency of storing data. There are three Cache mapping techniques.
- Direct Mapping: in Direct mapping both (RAM and Cache Memory) are used to store a data. An address of data or instruction is divided into individual two parts. Index part for RAM and tag part for Cache. But this process is not flexible.
- Associative Mapping: in this type, the associative memory is used to store content and addresses. This enables the placement of any word at any place in the cache memory. It is considered to be the fastest and easiest mapping technique among others.
- Set-associative Mapping: this mapping technique is a hybrid mixture of the associative mapping technique and direct mapping technique. It is a reasonable compromise between the easiest direct mapping and the complex hardware which is needed for associative caches.
Is it changeable?
It is not possible to change your PC’s cache memory as it is one kind of part of the processor. To change your cache memory you have to change your processor. But you can maintain cache size for different works (i.e. browser cache) by command.
Despite lots of advantages, caches are not totally problem less. They don’t know which instruction or data the processor is going to required next. It selects groups of data that happen to be in the database and close to the last program that was run or instruction that was required. A number of important factors can affect the performance and the average speed of the processor as well.
When a client computer or mobile device is accessing to an online server or a website over the internet, they might store them on their own memory or a disk of a collection of the documents and images that have already been used. Then, if those servers or websites are accessed again, they can be reloaded from the local cache rather than reloading them again which is a threat to the security.
First documented the use of a CPU cache was on the IBM System/360 Model 85. But after that, there has been much improvement in cache memory technology. In past, engineers focused on the cost of the cache and RAM and also on the average execution speed.
But now a day’s energy efficiency is considered more strictly. However, researchers are trying to use e-DRAM (embedded DRAM) and NVRAM (non-volatile RAM) for designing Cache Memory. Their main thought is in the cache cycle time, energy consumption and the area.
Hopefully, after reading this article, the answer to What is a Cache Memory is now cleared for you. In Cache Memory, each technique is associated with its own design constraints, advantages, and limitation. There are many advantages and at the same time many rooms for improving the performance of cache memory. Engineers are working hard to create a new technology which will remove the obstacles and will make your PC more fast and efficient.