Top Products on Amazon

What does RAM Memory do

Last Updated

You may be wondering what does RAM memory do. It is one of the most important components of the computer.

It is the one that you always look at when buying a new computer, or that you think about upgrading it to get better performance, etc.

But… do you really know what its function is? Why was this memory devised? All those doubts will be dispelled after a brief tour of history and all the knowledge you should have about this component.

History of RAM Memory

The first computers began using punched cards as a means to load programs or store data, later came relays, drum memories, and vacuum tubes, which were finally replaced by transistors.

See Also: What is RAM Memory

The first Random Access Memory (RAM) came in 1947 with the Williams tube. It was a kind of CRT like the old screens, since the cathode ray tube could read and write to charged points in any order.

Although this memory could only store a few thousand bits, it was more compact and energy efficient than vacuum tubes.

That same year magnetic core memory would be invented, and it would also end up becoming random access memory.

It was based on a series of magnetized rings, being able to store 1 bit for each ring, using magnetism in one way or another.

With the advent of the first integrated circuits, and solid state devices in the early 1970s, magnetic core memory was displaced and began to be integrated into chips.

One of the first types of solid-state memory to be developed was the ROM type, followed by SRAM, then DRAM, etc.

Companies like Intel, Toshiba, IBM, Fairchild Semiconductor, etc., were some of the pioneer companies in the field of memory in these decades. Currently, there are only a handful of RAM memory chip manufacturers supplying many brands.

Some of the examples of manufacturers are Samsung, Micron, SK Hynix, and more recently others have also joined, such as Philips.

See Also:

Types of RAM Memory

Before looking at what RAM does, let’s see how many types of RAM there are in this overview, to get a clearer idea. As you can see, you can mainly differentiate them between two large families:

  1. SRAM (Static RAM)

One of the types of RAM memory that uses only 6 transistors and works very fast. It is often used as a register or for CPU cache. The biggest drawback is its cost. Robert H. Norman of Fairchild Semiconductor was its inventor in 1963.

  1. DRAM (Dynamic RAM)

Unlike static, this memory is dynamic and uses capacitors to store information. It is much cheaper than the previous one, so higher capacities can be obtained while keeping prices at acceptable levels. It was invented in 1966 by Robert Dennard of IBM.

01 – FPM: stands for Fast Page Mode, a subtype of DRAM. It was inspired by the Burst Mode of Intel 486 processors, receiving an address from the memory controller and reading consecutive memory locations without the need for more addresses.

It was used in the 486 and early Pentiums, and had 70 or 60 ns accesses.

02 – EDO: Extended Data Output RAM is another type of memory released in 1994, with accesses of 40 to 30 ns, which was a significant improvement over FPM.

03 – BEDO: Burst EDO is an evolution of the EDO, presented in 1997, although it would not be marketed due to lack of support.

04 – SDRAM: Synchronous DRAM was developed by Samsung, although Intel had already dabbled with it beforehand. In this case, access is improved to 25 or 10 ns.

See Also: How to Repair RAM Memory

It is based on capacitors and if the electricity supply is cut all its content will be erased. It needs to be constantly fed. From this, in turn, variants have emerged:

  • RDRAM: Rambus became very popular, and was adopted by Intel for its Pentium 4s. However, it was soon displaced by DDR. It should also be noted that the Rambus company would launch other variants such as the XDR (Extreme Data Rate) or the XDR2 (Extreme Data Rate 2).
  • SDR SDRAM (Single Data Rate): single data rate.
  • DDR (Double Data Rate): A type of SDRAM with double data rate to improve performance.
  • DDR2: Doubles the previous one.
  • DDR3: Twice the data rate than the previous one.
  • DDR4: Doubles again at the previous rate.
  • DDR5: Twice the data rate than the previous one.
  • DDR6: Doubles the previous one.

Of course, there are other types of memory, but these are the most interesting given the topic of the article.

See Also: How to change RAM memory on PC

What does RAM Memory do?

In the introduction to the previous story you can see how efforts have been made to improve the speed of memory.

The problem is that this evolution has been very dissonant compared to the evolution of processing units, which are much faster. That is why some buffers were introduced, thus reducing latency and bottleneck.

That said, I’ll use this schematic of a Von Neumann-type computer to explain what RAM does. The first thing is to present the elements:

  • On any computer or PC, we have the operating system and any other software stored on the hard drive or secondary memory. It can be SSD, HDD, or hybrid. This memory, although a giant step has been taken with the introduction of SSDs, is still much slower than the CPU.
  • On the other hand, we have the virtual memory that appears in this scheme, and that in Windows corresponds to the pagefile.sys file, while in UNIX environments it would be the SWAP or exchange memory. Data and processes will be stored in this memory when they cannot enter the RAM memory.
  • RAM, or main memory, is an intermediate memory between the secondary and the CPU. It is considerably faster than the secondary, although slower than the cache.
  • Within the CPU there are much faster memories, such as the cache at its different levels, also acting as a kind of fast buffer between RAM and the functional units of the CPU. And then there are the register banks, where data or instructions are stored and which are very fast, but limited in

Now, once these elements have been presented, in case you did not know them, now we will describe how a computer works, explained in a simple way, to help you understand what RAM memory does:

See Also: What is STT RAM Memory

  • When any program (code made up of instructions and data) is executed, the operating system treats it as a process, loading said program into main memory or RAM. Depending on the kernel scheduling algorithm and priority, it can move processes up or down from RAM to/from virtual memory.
  • By doing this, everything needed to run that program is in faster memory. At this moment, the CPU will access the RAM to bring the instructions of said program and the appropriate data.
  • The CPU will then interpret the instruction and perform some operation on the data. This is how the result is obtained. When all the instructions in the process are processed, it has finished executing.

The answer to the question what does RAM memory do is simple: it acts as a buffer between the secondary and the CPU, improving performance.

If it were to go all the way to secondary memory, the delay would have a significant negative impact on performance. Note that:

  • Registers are almost instantaneously accessible, while L1 can take anywhere from 1 to 5 ns. The L2 cache between 5-10 ns, and the L3 between 10 and 20 ns.
  • RAM memory has access times of between 50-100 ns, depending on the type.
  • While the secondary is around 0.1 ms for some SSDs, or between 5-8 ms for a HDD.
  • That is, we are talking about a jump between the CPU and the secondary memory is of million times slower in some cases. While the jumps from CPU to RAM are lower, only a few tens or hundreds of times slower.

Conclusion

Without RAM between the CPU and secondary memory, a huge number of clock cycles would be wasted accessing data, resulting in very poor performance.

See Also: Different Types of Computer Memory

In addition to accelerating the transfer rate, creating RAM memories with increasingly lower latencies, and increasing the cache.

It is also pointing to a “full of light” future, since it is experimenting with fiber optic transfers to interconnect units that now use conventional conductors and electrical impulses.

The arrival of new optical devices could represent another great leap that would once again lower latency and increase the bandwidth between memory and the CPU. Will this technology arrive soon?