Memory consistency, also known as memory model entails what the programmer should know about a system in terms of the behavior of shared memory read and write which emanates from different processes. In this, the expectation of the programmer is entrenched in the memory consistency model. This is because programmers do not want any surprises while operating on a single processor or in most cases, multiple processors for any program.
At this, the memory consistency model deals with the order of instructions across accesses for the processor. The memory consistency model is also concerned with different instructions for different programs.
Memory consistency is one of the problems in the common ordering of figures. It deals with the challenge of defining how parallel processors can observe a shared-memory program. The model entails multiple processors and it is channeled towards the output of processors while working. The memory consistency models have different models that have been introduced for the easy use and accessibility of programmers. The easiest is the sequential consistency memory model although it has its acclaimed function mostly in theory.
This is an important memory consistency model that involves the operation of processors. A programmer expects that the accesses he had written is represented in the exact order and manner he had written it. By this, the accesses on a particular program remain the same and in order.
When the sequential arrangement of the accesses remains unchanged, a feature called program order is at work and it indeed maintains the order. The order maintained in this context is the order at which memory access has been generated and should be maintained by execution on the processor.
Another feature of the sequential consistency model is arbitrary. To cite an example, a casino has different decks of cards. Before the table can be opened for a game, the deck of cards must be split and rearranged in no particular order. It’s like a shuffle from each midpoint of the pile of cards. This shuffle is called the arbitrary rearrangement which also applies to the sequential consistency memory model.
Arbitrary is thus the interleaving of memory access between processors and at this stage, there is no control over the order as the accesses will be satisfied by the memory because it depends on the execution of both processors and their reach on each memory read and write. The interleaving is thus applied to the accesses of multiple processors.
It should be noteworthy that the sequential consistency memory model was proposed by Lesley Lamport in 1977/78 and it has remained integral and easy to access by programmers. Further, the sequential consistency is the result of execution maintained in the same order as presented into the program throughout the operations of each processor.
The sequential consistency requires the maintenance of program order on the individual processors and the assurance of write atomicity.
This model is applied in a high-level processor as it relaxes order constraint in pairs of reads, writes, and read-write. This means that reads are allowed to bypass writes under this model while writes are allowed to mutually bypass each other.
In this model, certain orderings as created by the sequential consistency are violated. However, memory utilization has the potential to be greatly improved. With this, there are different models under this model that permits a different form of violation which could then lead to the pitfall of particular programming.
One of the models under the Relaxed Model is the Total Score Order memory, consistency model. It provides that the writes are not reordered with other writes. However, in the stead of waiting for a write 1 (for example) to become visible, it could be placed into what is known as a store buffer. This will permit the processing of write 2. The store buffers allow hiding to write latency and also make the write visible to other threads or processors.
Another important feature is that the reads are allowed to bypass writes and the model can hide the write latency. That is, writes are not ordered according to reads. However, reads can be reordered with respect to writes to different memory addresses if, for instance, writes take longer time to process and the reads are from a different memory address.
This is another model under the Relaxed Model. The processor consistency model is more peculiar with the Total Store Order. However, the separate feature they encompass is that the TSO doesn't allow any processor to see the changes of a write except in cases where the write is obvious in all the processors.
On the other hand, the Processor Consistency permits any processor to see and access the change before the write is seen by all the processors. This makes processor consistency a more relaxed model than the total store order model.
The Partial Consistency or Partial Store Ordering (PSO) presents a more relaxed memory consistency model in comparison to both total store order. This model is fundamentally related to the TSO but it has an advantage to the consistency.
The model offers that writes in the same location will be in a sequential order while writes to a different memory location may not be ordered in any way. Also
, the processor may shuffle writes and this influences their original order in the memory system.
Our expert writers will write your essay for as low as
from $10,99 $13.60
Place your order now
