RAM is one of the most important components of our computer equipment. It is essential in any type of modern computer system, although it can vary in terms of capacity and frequencies. There are many parameters to take into account regarding RAM and here we will explain everything you need to know.
With the arrival of AMD Ryzen processors, DDR4 RAM has taken more prominence. This is because the performance of the company's processors directly depend on the RAM frequency. On the other hand Intel does not depend much on frequency, it depends more on the amount of RAM.
RAM is the acronym for Random Access Memory or random access memory. It is called this way because it can read and write in any of its positions without respecting any order. Virtually eliminates waiting intervals for access to information.
Table of Contents
What is RAM memory?
A computer is a system that is responsible for receiving and processing data. Not all data can be processed instantly or must be stored to be used in other processes. RAM memory is used to store all this information.
RAM memory allows reading and writing to any memory location anytime. The one who uses RAM the most is the processor, which saves information on running processes, queued or data you might need later. It is used in all current computers and devices.
We can therefore say that RAM's primary function is to store all the instructions that run on a processor. Instructions that come from different areas such as the operating system, the input and output devices, the hard disk or any element of our system.
Inside the RAM is save data and instructions from the different software that runs on our system. The data is sent from our hard drive before carrying out any kind of execution. It allows all programs to have the data of the programs that we execute available without waiting.
In the absence of RAM, we should load the information from a hard drive. Something that would be a problem due to transfer speeds and that random access to memory will slow down. This is why it is such an important element.
JEDEC and RAM standards
The asociación Joint Electron Device Engineering Council (JEDEC) sets the standards for semiconductor memory circuits and similar storage devices. It is a semiconductor trade and engineering standardization organization.
JEDEC Standards and Publications They are designed to serve the public interest by eliminating misunderstandings between manufacturers and buyers. Product interchangeability and improvement is facilitated. It also helps buyers to select and obtain the best result. They are standards that should be used by all manufacturers.
JEDEC Standard 100B.01
Are the "terms, definitions, and letter symbols for microcomputers, microprocessors, and memory ICs”. Said standard has the purpose of “promote the uniform use of symbols, abbreviations, terms and definitions throughout the semiconductor industry"
Define two common units of information:
- Bit (b) is the smallest unit of information in the binary numbering system and is represented by the digits 0 and 1.
- Byte (B) is a binary character string that normally functions as a unit. It is usually shorter than a computer word.
Unit prefixes for semiconductor storage capacity:
- Kilo(K): a multiplier equal to 1.024 (210)
- Mega (M): a multiplier equal to 1.048.576 (220 or K2; where K = 1.024)
- Gig (G): a multiplier equal to 1.073.741.824 (230 or K3; where K = 1.024)
JESD21-C
It establishes the "configurations for solid state memories"And they are maintained by the JC41 committee of JEDEC. This committee is based on manufacturers of microprocessors, memory integrated circuits, memory modules and other components, as well as component integrators, manufacturers of video cards and personal computers. Said standard 21 is published in a folder of sheets to facilitate its constant updating.
Documentation of modern memory modules, as well as the standards of memory integrated circuits, as well as the reference design of the modules needs more than one hundred pages. They contain the physical and electrical characteristics of the modules. Additionally include data for computer simulations of modules RAM memory operating in a system.
RAM memory characteristics
There are different parameters that we must take into account in the RAM memory modules. Voltage parameters (voltage), latencies and frequency are closely linked. Above all, latencies and frequencies are linked, since modifying one of these parameters modifies the others.
When we increase the frequency, unequivocally latencies increase and if the latencies are reduced, the frequencies are reduced. Changing the frequency or latencies leads to an unavoidable voltage adjustment of operation. Increase tension supposes increase heat generated and therefore many modules they implement passive heatsinks.
Highlight that modules stickers They usually offer us very useful and important information. The first thing we will indicate is the standard, the most current being DDR4. The next parameter that we can read in the label is the set capacity. It will indicate the capacity of the module, the number of modules in the kit and therefore the total capacity of the kit.
We can also see the parameter of the memory speed expressed in Mega-Hertz (MHz). We are also indicated the RAM memory latencies, indicating the four latency parameters. Finally the RAM modules voltage, the base voltage being 1.2V for DDR4.
RAM memory operation
RAM memory performs the functions of data storage that may need to be accessed quickly. It has the characteristic of storing data temporarily, until they are cleared by the system or when the system is shut down. Information, when stress is removed, is automatically erased.
This component offers great functionality because the information it stores is volatile. It apports better access speeds, avoids device saturation and is more processor friendly. It is basically a memory space where the processor stores the instructions and other system data.
Physically it is a set of chips soldered onto a PCB which is assembled into a specific DIMM socket. Under the DDR standard, it has a different notch, so a DDR4 module cannot be installed on a motherboard with DDR3 support.
That the data be erased when the working voltage is cut off, it's a positive and a negative. If the changes have not been stored in a persistent memory (hard disk or pen drive) they will be lost. It will be the processor who decides which data should be saved and which not, depending on the needs.
DIMM interface
All the components that are installed on our motherboard require a connection interface. The one for RAM is called DIMM (Dual In-line Memory Module) and varies according to the DDR standard. This is due to a notch in the connection interface that prevents the installation of the wrong modules. Due to this notch we could not install a DDR3 module on a DDR4 module.
The RAM memories for this connection interface allow the installation of memories on both sides of the printed circuit. This makes it possible to increase the capacity of each module significantly. Some DDR4 modules can reach 64GB of capacity per module.
The standard DIMM is intended for desktop computers, while for notebooks it is reduced. For the notebook computers using the SODIMM or small outline DIMM interface. While a full DIMM has 288 contacts, SODIMMs only have 144 contacts
RAM memory latencies
All RAM memory modules offer us four latency parameters that will vary by manufacturer and frequency. These parameters have a specific meaning and we are going to explain it to you so that you can identify them. For the example we are going to use latencies 17-18-19-35.
- CAS Latency (CL) - 17: Clock cycles from when a column address of a memory is sent and the beginning of the data that is stored in it. The time it takes to read the first bit of memory from a RAM module in a correct row already open.
- Row Address to Column Address Delay (TRCD) - 18: Clock cycles that are required since a memory row is opened and an inner column can be accessed. Time to read the first bit from memory without an active row is CL + TRCD
- Row Precharge Time (TRP) - 19: Clock cycles that are needed to send a preload command and to open the next row. Reading time of the first memory bit if a different row is open is CL + TRCD + TRP
- Row Active Time (TRAS) - 35: Clock cycles required between a row actuation command and the sending of the load command. Time required to internally refresh a row with TRCD overlap. For SDRAM (Syncronous Dynamic RAM) type modules the value is CL + TRCD. For other cases it is close to (CL * 2) + TRCD.
Interpretation of RAM latencies
We know what each of the parameters means, but, What use is this data to us? This allows us to know the effective latencies at each of the different frequencies and how it will affect our performance. We must bear in mind that this will affect the final performance of the system, in some cases.
The latency must take into account that it is the time between the processor makes a request and receives the response. This is the time it takes to perform the action. To calculate it we must apply the following formula:
(1000 / actual frequency) * CAS latency = Time in ns (nanoseconds)
Example for a frequency of 2666MHz with a CL15 and CL 16
(1000/2666) * 15 = 5.626ns | (1000/2666) * 16 = 6ns
Example for a frequency of 3000MHz with a CL15 and CL 16
(1000/3000) * 15 = 5ns | (1000/2666) * 16 = 5.333ns
Example for a frequency of 3400MHz with a CL15 and CL 16
(1000/3400) * 15 = 4.4ns | (1000/2666) * 16 = 4.706ns
Impact of latency on processor performance
Now that we know the different names as well as the calculation of latencies, it is time to explain their impact on system performance. The lower the system latency, the shorter the times between the system request and the data reception.
The processor implements cache which stores primary data for different operations. For high-volume data or data that is not required on a recurring basis, it is sent to RAM. So the lower the latency between the processor and RAM, the performance will vary.
We must bear in mind that latency above all impacts games more than in synthetic applications. This is because games are based on 3D image modeling and processing. For this type of applications must store a lot of data in RAM and it must be accessed a lot.
We see how in PCPer they have performed a latency performance test using the Intel vTune tool. We can see how the latency dependence of gaming memories is higher than that of productivity tools. We see that WinRAR is an exception, this is because compression tools are highly dependent on processor load and communication with RAM.
Dual Channel RAM configuration
La Dual Channel technology of the processors for the domestic market it is essential for all kinds of tasks. This type of configuration allows access to two modules interchangeably. It is achieved by doubling the bandwidth, which becomes 128 bits instead of the typical 64 bits.
This is achieved by a memory controller that is integrated within the same processor. Processors can access two modules simultaneously, improving overall performance.
We should note that the two modules must operate at the same frequency, they must have the same capacity and the same latency. It is recommended to buy the two-unit kits offered by all manufacturers on the market. We can mix modules, but keep in mind that we can have problems. The first is that they do not work due to lack of compatibility and the second is that they work according to the performance of the worst RAM module.
Better RAM memories
No products found.
