|
Introduction to Processor Cache What is
Processor Cache? Motherboard
designers often make the secondary cache upgradable from as
little as 32KB, all the way to 1MB and beyond. The technique used
to access this cache determines the number of chips used to
upgrade the cache, whether the upgrade is a module or individual
chips. As with RAM, cache memory is addressed in 8 bit
increments, which means 4 or 8 chip configurations are most
common. Although cache rarely uses parity, additional chips are
often used. These additional chips are called "TAG RAM"
or "Dirty Bit" RAM. They help the cache controller keep
track of where data is located within the cache, and for delaying
the transfer of data from cache back to main RAM. There are two
key styles of cache: Write-Back (W-B) and Write-Through (W-T).
Both styles serve the cache's primary purpose of allowing the CPU
to retrieve data from cache when it's available there. This
reduces the need to access the much slower main RAM. Sending data
to memory is where W-B differs from W-T. Write-Through means data
is sent, or written, to cache and main memory simultaneously.
Write-Back cache allows the CPU to send data only to the cache,
further reducing the need for the CPU to access slower main
memory. The cache controller performs the maintenance of main
memory, freeing the CPU from this task, thus allowing the CPU to
return to more processor intensive operations. Write-Back cache
is usually more expensive and more complex to design and build,
but offers performance advantages over Write-Through. Why isn't
cache much larger? |