Why aren all caches as fast as L1?
To make all the cache memories run at the same speed as the L1 cache – 10 nanoseconds – would be extremely expensive and in fact the computer doesn’t need large amounts of cache memory due to locality of reference. Locality of reference means that no matter how large or small in size a running application is, only a small part of the respective program is being stored into the cache at one given time. This is because the vast majority of programs mostly contain fairly large amounts of code lines which are executed only once. The other lines are represented by program loops which are executed over and over again. Less than 10 percent of processor time is spent by the programs running at one time.
