Within a CPU, cache memory provides low­-latency access to frequently used information. Because access to system main memory (usually DRAM) is relatively slow compared to CPU operation, cache memory can dramatically speed up program execution. Usually cache memory is integrated as SRAM alongside the CPU rather than being implemented as a separate discrete device.

style="margin: 0in 7.85pt 0.0001pt 5pt; line-height: 118%;" class="normal"> When any of the processes running within the CPU need access to system memory, the read or write request is first checked against the cache. The cache contains data that has been recently retrieved from memory, along with the corresponding address (which is known as the tag). If the requested address of the new read or write operation matches a tag in the cache, it is referred to as a cache hit. The CPU can then use cache memory to fulfill the request and continue execution. If the requested address is not found, it is referred to as a cache miss. Read misses are costly in terms of performance because the cache must be updated before CPU execution resumes. Write misses are less significant because the data is still written to the cache, although that data will have to be copied off to main memory at some point. The decision of when to make that transfer from cache to main memory is one aspect of cache management. Read more Read less