What is ram memory and how does it work?

When our computer is slow, one of the first things we look at is if we have enough RAM memory. Also, one of the requirements that all programs, games and operating systems usually have is a minimum of RAM. What is RAM really and what is it for? We will see all this and more today in this article.

Index of contents

What is RAM

RAM (Random Access Memory) is a physical component of our computer, usually installed on the same motherboard. The RAM is removable and can be expanded by modules of different capacities.

The function of RAM memory is to load all the instructions that are executed in the processor. These instructions come from the operating system, input and output devices, hard drives and everything that is installed on the computer.

In the RAM memory all the data and instructions of the programs that are running are stored, these are sent from the storage units before their execution. In this way we can have available all the programs we run, if you hardly wait.

If RAM does not exist, the instructions should be taken directly from the hard drives and these are much slower than this random access memory, making it a critical component in the performance of a computer.

It is called random access memory because it can be read and written to any of its memory locations without having to respect a sequential order for its access. This allows practically no waiting intervals for access to information.

Physical components of RAM

As for the physical components of a RAM memory module, we can distinguish the following parts:

Component plate

It is the structure that supports the other components and the electrical tracks that communicate each of the parts of these.

Each of these boards form a RAM memory module. Each of these modules will have a certain memory capacity according to those existing in the market.

Memory banks

They are the physical components in charge of storing the records. These memory banks are formed by integrated circuit chips that are made up of transistors and capacitors that form storage cells. These elements allow bits of information to be stored within them.

For the information to remain inside the transistors, a periodic electrical supply will be necessary in them. This is why when we turn off our computer this memory is completely empty.

This is the big difference between, for example, RAM and SSD storage units.

To know more about SSD drives you can visit our article where the best models and their characteristics are explained in detail:

Each RAM module has several of these memory banks physically separated by chips. In this way it is possible to access the information of one of them while another is being loaded or unloaded.

Watch

Synchronous RAM memories have a clock that is in charge of synchronizing the read and write operations of these elements. Asynchronous memories do not have this type of integrated element.

SPD chip

The SPD (Serial Presence Detect) chip is in charge of storing data related to the RAM memory module. These data are memory size, access time, speed, and memory type. In this way the computer will know what RAM memory is installed inside by checking this during power up.

Connection bus

This bus, made up of electrical contacts, is in charge of allowing communication between the memory module and the motherboard. Thanks to this element we will have memory modules separate from the motherboard, thus being able to expand the memory capacity by means of new modules.

Types of RAM memory modules

Once we have seen the different physical components of the RAM memories, we will also have to know the type of encapsulation or modules they mount. These modules are basically made up of the component board and the connection bus together with their contact pins. Among others, these are the most used modules before and now:

• RIMM: These modules mounted RDRAM or Rambus DRAM memories. Then we will see them. These modules have 184 connection pins and a 16-bit bus. SIMM: This format was used by older computers. We will have 30 and 60 contact modules and 16 and 32 bit data bus. DIMM: this is the format currently used for DDR memories in versions 1, 2, 3 and 4. The data bus is 64 bits and can have: 168 pins for SDR RAM, 184 for DDR, 240 for DDR2 and DDR3 and 288 for DDR4. SO-DIMM: it will be the specific DIMM format for portable computers. FB-DIMM: DIMM format for servers.

Types of RAM technologies

In general, two types of RAM exist or have existed. The asynchronous type, which do not have a clock to synchronize with the processor. And those of the Synchronous type that are able to maintain synchronization with the processor to gain efficiency and effectiveness in accessing and storing information in them. Let's see which exist of each type.

Asynchronous memories or DRAM

The first DRAM (Dinamic RAM) or dynamic RAM memories were of asynchronous type. It is called DRAM because of its characteristic of storing information in a random and dynamic way. Its structure of transistor and capacitor means that for a data to be stored inside a memory cell, it will be necessary to power the capacitor periodically.

These dynamic memories were of the asynchronous type, so there was no element capable of synchronizing the frequency of the processor with the frequency of the memory itself. This caused that there was less efficiency in the communication between these two elements. Some asynchronous memories are as follows:

• FPM-RAM (Fast Page Mode RAM): These memories were used for the first Intel Pentium. Its design consisted of being able to send a single address and in exchange receive several of these consecutive ones. This allows for better response and efficiency as you do not need to be continuously sending and receiving individual addresses. EDO-RAM (Extended Data Output RAM): This design is the improvement of the previous one. In addition to being able to receive contiguous addresses simultaneously, the previous column of addresses is being read, so there is no need to wait for addresses when one is sent. BEDO-RAM (Burst Extended Data RAM): improvement of the EDO-RAM, this memory was able to access various memory locations to send data bursts (Burt) in each clock cycle to the processor. This memory was never commercialized.

Synchronous or SDRAM type memories

Unlike the previous ones, this dynamic RAM has an internal clock capable of synchronizing it with the processor. In this way, access times and communication efficiency between the two elements are significantly improved. Currently all our computers have this type of memory operating on them. Let's look at the different types of synchronous memories.

Rambus DRAM (RDRAM)

These memories are the complete overhaul of asynchronous DRAMs. It improved this both in bandwidth and transmission frequency. They were used for the Nintendo 64 console. These memories were mounted in a module called RIMM and reached frequencies of 1200 MHz and a 64-bit word width. Are currently deprecated

SDR SDRAM

They were just the predecessors to the current DDR SDRAM. These were presented in DIMM-type modules. These have the possibility of connecting to the slots of the motherboard and consist of 168 contacts. This type of memory supported a maximum size of 515 MB. They were used in AMD Athlon processors and Pentium 2 and 3

DDR SDRAM (Double Data Rate SDRAM)

These are the RAM memories currently used in our computers, with different updates. DDR memories allow the transfer of information through two different channels simultaneously in the same clock cycle (Double Data).

The encapsulation consisted of a 184-pin DIMM and a maximum capacity of 1 GB. DDR memories were used by AMD Athlon and later by Pentium 4. Its maximum clock frequency was 500 MHz

DDR2 SDRAM

Through this evolution of DDR RAM, the bits transferred in each clock cycle were doubled to 4 (four transfers), two forward and two for return.

Encapsulation is a 240-pin DIMM type. Its maximum clock frequency is 1200 MHz. The latency (information access and response time) for DDR2-type chips increases compared to DDR, so in this respect it reduces their performance. DDR2 memories are not compatible in installation with DDRs, because they work at a different voltage.

DDR3 SDRAM

Yet another evolution of the DDR standard. In this case, energy efficiency is improved by working at a lower voltage. The encapsulation is still a 240-pin DIMM type and the clock frequency goes up to 2666 MHz. The capacity per memory module is up to 16 GB.

As in the technology leap, these DDR3 are memories with a higher latency than previous ones, and are not compatible in installation with previous versions.

DDR4 SDRAM

As in the previous cases, it has a substantial improvement in terms of clock frequency, being possible to reach up to 4266 MHz. As in the technology leap, these DDR4 are memories with a higher latency than the previous ones and incompatible with expansion slots for older technologies.

DDR4 memories mount 288-pin modules.

Nomenclature used

We have to pay special attention to the nomenclature used to name the current DDR-type RAMs. In this way we can identify what memory we are buying and how often it has.

We will first have the available memory capacity followed by "DDR (x) - (frequency) PC (x) - (data transfer rate). For example:

2 GB DDR2-1066 PC2-8500: we are dealing with a 2 GB DDR2 type RAM module that works at a frequency of 1066 MHz and with a transfer rate of 8500 MB / s

RAM memory operation

To know how a RAM memory works, the first thing we will have to see is how it physically communicates with the processor. If we take into account the hierarchical order of RAM memory, this is located exactly at the next level to the processor cache.

There are three types of signals that the RAM controller must handle, data signals, addressing signals, and control signals. These signals mainly circulate on data and address buses and other control lines. Let's see each one of them.

Data bus

This line is responsible for carrying the information from the memory controller to the processor and the other chips that require it.

This data is grouped into 32 or 64 bit elements. Depending on the bit width of the processor, if the processor is 64, the data will be grouped into 64-bit blocks.

Address bus

This line is responsible for transporting the memory addresses that contain the data. This bus is independent from the system address bus. The bus width of this line will be the width of the RAM and the processor, currently 64 bits. The address bus is physically connected to the processor and RAM.

Control bus

Control signals such as Vdd power signals, Read (RD) or Write (RW) signals, Clock signal (Clock) and Reset signal (Reset) will travel on this bus.

Dual channel operation

The dual channel technology allows an increase in the performance of the equipment thanks to the fact that simultaneous access to two different memory modules will be possible. When the dual channel configuration is active it will be possible to access blocks of a 128 bit extension instead of the typical 64. This is especially noticeable when we use graphics cards integrated in the motherboard since, in this case, part of the RAM is shared for use with this graphics card.

To implement this technology, an additional memory controller located in the chipset of the north bridge of the motherboard will be necessary. For a dual channel to be effective, the memory modules must be of the same type, have the same capacity and speed. And it must be installed in the slots indicated on the motherboard (usually pairs 1-3 and 2-4). Although do not worry because even if they are different memories they will also be able to work on Dual Channel

Currently we can also find this technology using triple channel or even quadruple channel with the new DDR4 memories.

RAM memory instruction cycle

The operating scheme is represented with two dual channel memories. For this we will have a 128-bit data bus, 64 bits for each data contained in each of the two modules. In addition, we will have a CPU with two memory controllers CM1 and CM2

One 64-bit data bus will be connected to CM1 and another to CM2. In order for the 64-bit CPU to work with two blocks of data, it will spread them over two clock cycles.

The address bus will contain the memory address of the data that the processor needs at any given time. This address will be from both the module 1 and module 2 cells.

The CPU wants to read a data from memory location 2

The CPU wants to read the data from memory location 2. This address corresponds to two cells located in two dual channel RAM memory modules.

Since what we want is to read the data from the memory, the control bus will activate the read cable (RD) so that the memory knows that the CPU wants to read that data.

Simultaneously the memory bus will send that memory address to RAM, all synchronized by the clock (CLK)

The memory has already received the request from the processor, now a few cycles later it will prepare the data from both modules to send it over the data bus. We say a few cycles later because the latency of RAM makes the process not immediate.

The 128 bits of data from the RAM will be sent over the data bus, a 64 bit block for one part of the bus and a 64 bit block for the other part.

Each of these blocks will now reach memory controllers CM1 and CM2, and in two clock cycles the CPU will process them.

The reading cycle will be over. To do the write action it will be exactly the same, but activating the RW cable of the control bus

How to tell if a RAM is good

To know if a RAM has a good or bad performance we will have to look at certain aspects of it.

• Manufacturing technology: the main thing will be to know which technology implements the RAM memory. In addition, this must be the same that supports the motherboard. For example, if it is DDR4 or DDR3, etc. Size: Another main aspect is the storage capacity. The more the better, especially if we are going to use our equipment for gaming or very heavy programs, we will need large capacity RAM, 8, 16, 32 GB etc. Board capacity for which channel: Another aspect to consider is if the board allows dual channel. If so, and for example we want to install 16 GB of RAM, the best thing to do is to buy two modules of 8 GB each and install them in dual channel, before installing only one of 16 GB. Latency: Latency is the time it takes for memory to do the data search and write process. The lower this time, the better, although it will also have to be weighed with other aspects such as transfer capacity and frequency. DDR 4 memories for example have high latency, but are counteracted by high frequency and data transfer. Frequency: is the speed at which the memory works. The more the better.

You may also like:

This ends our article on what a RAM is and how it works, we hope you liked it. If you have any questions or want to clarify something, just leave it in the comments.