During my study, I come to know,

pages = groups of data
page frames = physical addresses in main memory

In the main memory (RAM), how the data & address are stored? my current assumption is, in RAM memory address are not stored only data are stored,

is it right?

You'll have to give us some more context for this question. A program (or the operating system) will store data at an address in RAM---and it will also have a mechanism for storing what the address is (so it can find the data). How all this happens depends on the program and what the data is.

All programs are using virtual memory, to Read or Write the data in to RAM, the Kernel uses the Page Table concept and retrive the Physical Address and request CPU to Read or Write the data in to that Physical Address.

Now CPU access the data from the RAM by using the Physical Address.

My question is, using Physical Address how CPU retrives the data from the RAM ?

Hi -

At the lowest level, address translation is done in hardware. The kernel "manages" the process, the CPU and associated hardware actually "does" it.

These links might help:

http://en.wikipedia.org/wiki/Virtual_memory
<= Fairly high-level

http://duartes.org/gustavo/blog/post...es-your-memory
<= Much lower-level detail about kernel implementation...
... but *don't* forget about the importance of the MMU hardware!

That appears to be part of the documentation of a specific strange substitute for assembly language. Very little there is relevant to the topic of this thread.

That's a big shift from what you seemed to be asking earlier, so I wonder what you're really trying to ask. But taking that literally:

The physical address is passed from the CPU to the memory controller. The memory controller translates it again in order to select the right bank of memory and generate row and column numbers and other stuff that I don't 100% understand myself.

In current AMD x86_64 designs, that memory controller is inside the same CPU package with the CPU itself. In Intel designs, the memory controller is in a separate part on the motherboard.

It simply discusses how memory can be accessed in 80x86 without delving into MMU internals (it's part of assembly teaching book "Art of Assembly", which is available online).

Here's how it works...

All of the addresses issued by a program are virtual addresses. This means that they are subject to "address translation" by the hardware... although the program in question cannot be aware of what is actually going on. This technique allows the operating system kernel to use any physical address to satisfy the storage-requirements of any virtual-address that might be issued by any program.

The actual implementation of this hardware feature divides the virtual address references into equal-sized chunks, called "pages," and therefore also the physical-address space into "page frames" of the same size. The mapping that takes place, in other words, does not occur on a byte-by-byte level, but rather in terms of chunks that are (say...) 4K in size.

This feature also allows the operating-system to say that a particular virtual address, at the moment, maps to "nowhere at all" in physical memory-space. When this happens, the running program is interrupted with a "page fault" and the operating-system must figure out how to resolve the problem. Running programs are not specifically aware that they have been interrupted in this way.