INTRODUCTION:
Shared memory is a memory that can be used by multiple programs to avoid redundant copies or to provide communication among them. In other words we can say that, it is a technique, in this exchanging of data in process of program is done more quickly than by reading and writing using the services of OS. It is the fastest form of IPC available. A number of basic issues in the design of shared memory system have to be taken into consideration. Once the memory is mapped into the address space of the processes that are sharing the memory region, no kernel involvement occurs in passing the data between the processes. It refers to a multiprocessing design where several processors access globally shared memory.
These include access control, synchronization, protection and security.
 Access control determines which process accesses are possible to which recourses.
 Synchronization constraints limit the time of access from sharing processes to shared resources.
 Protection is a system feature that prevents processes from making arbitrary access to resources belong to other processes.
 At the memory module the requests arrive through its two ports.
 The simplest shared memory system consists of one memory module that can be accessed from two processors.

 Classification of shared memory:

Depending on the interconnection of network, a shared memory leads to system can be classified as :-
• UMA: -
 Shared memory is accessible by all processor through an interconnection network in the same way a single processor accesses its memory.
 All processor have equal access time to any memory location.
 Because access to the shared memory is balanced, these systems are called SMP (symmetric multiprocessor). (UMA Architecture)
• NUMA :-
In Non uniform Memory Access each processor has part of the shared memory attached. However, the access time depends on the distance to the processor. In NUMA, memory access time depends on the memory location relative to a processor. A processor can access its own local memory faster than non local memory. (NUMA Architecture) • COMA: -
Cache Only Memory Architecture is similar to the NUMA. In this each processor has part of the shared memory in the COMA. In this case the shared memory consists of cache memory. A COMA system requires that data be migrated to the processor requesting it.

(COMA Architecture)

 Advantages:

• Programming for IPC become simple that we simply read and write to an address pointer available in our process address space.

• We need not use system calls like read and write.

 Architectural Support for Shared Memory:

BIBLIOGRAPHY
Books:
 Advance Computer Architecture
 William Stallings
 B RAM

Websites: http://siber.cankaya.edu.tr/ParallelComputing/ceng471/node18.html