Definition
Cache Write-Miss
A cache write-miss occurs when a write access targets data that is not currently present in the cache.
So the processor must decide, according to the cache write policy, whether the block should first be loaded into the cache or whether the write should bypass the cache.
A cache write-miss is the write analogue of a cache miss.
Policies
Write-Around
Write-around
On a write-miss, the processor ignores the cache and writes the new value directly to main memory.
So the missed block is not loaded into the cache.
This approach is used mostly in combination with write-through.
Fetch-on-Write
Fetch-on-write
On a write-miss, the processor first loads the relevant block into the cache, replacing the current content of the target location and updating the tag.
If the block size is greater than one word, the remaining words of the block are fetched from main memory as well.
So the write-miss is handled by first performing a read of the block from memory and then continuing as a cache write-hit, according to the chosen write-hit strategy.
This is the most commonly used method for write-misses.