Multi-threading has emerged as a promising and effective way to tolerate memory latency. The implementation process is mainly divided into two types which are used to share memory units between processors. The first approach is the hardware approach using shared memory machines; another method is the software approach that provides an impression of shared virtual memory through a middleware. Shared memory is measured as a simple but efficient parallel programming model; Shared memory is a widely accepted model for building parallel applications. The main advantage is to provide the programmer with an adequate communication paradigm: however, with research carried out over the years, it has been shown that it is difficult to provide the illusion of shared memory to large-scale systems. Although the hardware approach through cache coherence has proven its efficiency, cost-wise, it is not scalable. On the other hand, shared virtual memory is a cost-effective method of providing shared memory abstraction over a computer network with minimal processing overhead. Say no to plagiarism. Get a tailor-made essay on "Why Violent Video Games Shouldn't Be Banned"? Get an original essay In most cases, distributed shared memory (DSM) systems and memory coherence protocols are used to support multiprocess computing. These processes have no common virtual address spaces and are assigned to different computers. Several new problems appear when extending this model to multi-threaded cases. First, the default state of multi-threaded programs is virtually shared address space. With physically separate machines, address space and code segments will be duplicated. However, these global variables must be shared both locally and remotely across data segments. Because virtual memory management (VMM) manages pages in operating systems as an alternative to data items, variable access patterns and locations can acquire a high frequency and volume of communication. Second, multi-threaded programs use mutex locks to access critical sections. In distributed systems, locks are no longer shared by these threads. As a result, the traditional locking mechanism will not work. Finally, most current coherence protocols in existing shared virtual memory systems such as Overlapped Homebased LRC, Tread Marks System, and Home-based LRC require a clear relationship between blocks and data. in advance. However, it can be difficult to obtain this information from compilers, especially if the access data is pointed to a pointer. Typically, programmers have to take care of these processes manually. The following contributions are made in this paper: Locality-based data distribution: The memory block for global variables is restructured while data segments are replicated and subsequently sent to other different locations based on the pattern accessed between threads and data. Hosting specific thread bundles reduces the communication frequency and volume of data segments. The DSM system allows processes to adopt virtual shared memory globally. DSM software offers abstraction for globally shared memory, which allows a processor to access any data item, and programmers also don't have to worry about how and where to get the data. For programs with sophisticated parallelization strategies and composite data structures, this is.