Cache Coherence:

• Cache does improves the performance of a system but it creates a problem know as cache coherence.
• Before the cache is copied into main memory ,if the two copies one in cache and other in main memory are different.
• This inconsistency in value of cache and main memory is called cache coherence problem.
• In multiprocessor system, multiples copies of same data can exist in different cache simultaneously, if processors are allowerd to freely update their own copies an inconsistency may arise.
• During a Write Operation if an item is placed in cache memory then it is required to update its copy in main memory also.

Strategies:

• Software Solutions
• Hardware Solutions
  • Snoopy Protocols.
  • Directory Protocols.

Following are the protocols to solve the cache coherence.

1) Snoopy Protocols:

• When a number of processors are connected to a common bus and each processors has its own local cache then cache inconsistency can be maintained using,

  • Write –Invalidate Policy.
  • Write- Update Policy.
• In Write- Invalidate Policy, whenever changes are made in local cache of processor, all remote copies with other processors are invalidate.
• Thus, only one valid copy will remain after write invalidate.
• The Write-Update will broadcast the new data block to all the caches containing a copy of the block.
• Thus, all copies of data will become identified after updating in local cache.

2) MESI Protocol:

• Write Invalidate approach is widely used in multiprocessor system.
• Write-Invalidate protocol is also known as MESI Protocol.