Why is a RAM called a memory

RAM / main memory

The main memory is an important part of a computer and an essential factor for the performance of a computer system when it comes to processing large amounts of data. The reading and writing speed as well as the storage capacity play a major role here.
Since the internal memory of a processor is usually too small and access to the hard disk is too slow, the main memory serves as a storage space for data and program code that has to be processed and temporarily stored. The main memory serves as a buffer on the way between the processor and a hard disk or another input and output unit.

The working memory is also referred to as main memory. Sometimes you also use RAM, which is short for Random Access Memory. Virtually every computer, past, present, and future uses some type of RAM as memory. RAM is characterized by random access, both reading and writing. That is what a working memory essentially has to be able to do. And as quickly as possible.

Virtual memory

In principle, the physical main memory is limited to the size of the built-in main memory. But computer systems with hard disk or mass storage can virtually increase their main memory. In addition, the applications do not see the physical memory, but only their own virtual address space. This is made available by the operating system. It happens that all running applications together have more virtual memory available than the working memory physically available. If this is the case, the operating system transfers the data to the hard drive. But that also means that the overall system becomes slower. This is because access to the virtual main memory on the hard disk takes more time than access to the actually physically available main memory.

  • Paging file
  • SWAP file
  • SWAP partition

Memory modules / memory bars

In order to flexibly equip computers with main memory, the main memory is not permanently built into the computer, but is outsourced to memory modules. These are small plug-in circuit boards on which the memory chips are soldered. This design is called a storage module. These memory modules are plugged into the slots provided on the motherboard. Since not all memory modules are identical, but are optimized for different applications and manufactured by different manufacturers, the memory controller has to cope with a different number of components. So that this succeeds, the memory modules and the memory organization are standardized by JEDEC.

Storage controller

The memory controller integrates the memory modules of the main memory into a computer system. The memory modules do not have their own logic. There are only circuit boards with memory chips on them. The memory controller therefore coordinates access to the memory. The memory controller is the link between processor and main memory or between chipset and main memory. The exact connection depends on the system architecture.

Storage controllers connect the storage modules in one or more storage channels. The memory channels are equipped with one or more memory modules. At the beginning of the boot process, the memory controller reads the permitted operating parameters from the memory module and then puts them into operation. The system then has RAM.

In the early days, the memory controller was integrated in the chipset. Data and program code therefore always had to make the detour via the chipset. With increasing clock frequency and integration density, processors became faster and faster in data processing. Therefore, the speed of the working memory had to be increased again and again over time.
If the processor cannot read the data from the main memory fast enough, the effective computing power of the processor drops. This is why processor manufacturers are the ones most interested in accelerating memory. After all, they want their processors to be able to maximize their capabilities.

Processor manufacturers are not only interested in accelerating the working memory, but also in accelerating the connection between the processor and working memory. This is why processor manufacturers build the memory controller into the processor at the same time in order to connect the main memory to the processor as quickly and as quickly as possible.
The memory controller built into the processor has positive effects on the overall system. On the one hand, the direct connection of the memory to the processor shortens the latency time for the accesses. This also means that the processor can manage with smaller caches overall. In addition, the data traffic between CPU, memory and peripherals is decoupled. The data transfer of the periphery can now no longer get in the way of the data traffic to the memory.

The number of processor cores grows with each new generation of processors. And improved processor architectures increase the demands on the data transfer performance of the main memory. In addition to more cores, the processor also has ever more and ever wider memory channels with which it connects the memory modules.

32-bit limit / 4-GByte limit

Many people ask themselves: Why can't I address 4 GB of RAM in my computer? The operating system only shows me something over 3.5 GB.

Incidentally, the limit of a 32-bit Windows is not 4, but 2 GB. Most 32-bit applications cannot use more than 2 GB of virtual memory under Windows. In addition to the physical main memory, the virtual memory also includes the swap file. Windows creates a swap file in order to swap out little-used parts of the main memory to a file on the hard disk in order to keep the physical main memory as large as possible.
While normal programs can only see 2 GB, it can be more than 2 GB with special programs. To do this, the program must be compiled accordingly and Windows must be run with certain boot options. The process is called Physical Address Extension (PAE). It works with superimposed memory windows and is reminiscent of EMS from DOS times. PAE is just as cumbersome and time consuming.
The PAE mode was developed to be able to address memory and interface addresses up to 36 bits (64 GByte). But the CPU, chipset and operating system have to play their part. Because some drivers and applications crash during PAE access to high addresses, the address space in Windows XP and Vista is limited to 32 bits.

Although the 32 bits with PAE are sufficient for 4 GB of RAM, not only the RAM but also the other hardware components need to be addressed. The memory size of the main memory is mainly dependent on the memory size of the graphics card. Additional I / O addresses of expansion cards and interfaces occupy additional addresses below the 4 GByte limit. This is why you can usually only access a little more than 3 GB of RAM under 32-bit Windows.
So it is hardly worthwhile to install more than 3 GB of RAM in a computer with 32-bit Windows. Expanding the main memory to more than 2 GB is only useful in a few cases. Namely when several memory-intensive programs are running at the same time. 2 GB of RAM is a reasonable and practical limit for a 32-bit Windows.

If you need more than 32 bits, you should use a 64-bit operating system right away. But you won't get very far with a 64-bit operating system alone. Because a 64-bit operating system can only recognize 7.2 of 8 GB of RAM. What can be the reason? It's the same problem as with a 32-bit operating system.
The I / O address areas of expansion cards and interfaces overlap the memory addresses below the 32-bit limit. And then Windows cannot use it for the main memory. All components in a PC are addressed via the I / O address range, which must be below the 32-bit limit. The more components a PC contains, the larger the I / O address range and the more addresses are missing for the main memory. Depending on the equipment and BIOS of the PC, an address range that corresponds to 0.5 to 1.2 GB can be blocked.
So that the full working memory can be recognized and used, the BIOS must use memory remapping (memory hoisting or reclaim) to move the address area of ​​the physical RAM to the highest address of the built-in RAM. Usually over 32 bits (4 GB). But if 8 GB of RAM are installed, then the relocated addresses must be greater than 33. But not every BIOS and not every memory controller supports this. The same problem occurs as with the 32-bit limit.

The prerequisite for being able to use the entire RAM over 2 GB is not just a 64-bit operating system. A suitable motherboard is also required that supports addresses beyond 32 bits and also includes remapping functions.

Overview: semiconductor memory

Other related topics:

Everything you need to know about computer technology.

Computer technology primer

The computer technology primer is a book about the basics of computer technology, processor technology, semiconductor memories, interfaces, data storage devices, drives and important hardware components.

I want that!

Everything you need to know about computer technology.

Computer technology primer

The computer technology primer is a book about the basics of computer technology, processor technology, semiconductor memories, interfaces, data storage devices, drives and important hardware components.

I want that!