我不知道你已经给了我们足够的上下文这个问题，但在这里不用。

I'm not sure you've given us enough context for this question, but here goes.

缓存必须存储不仅实际的缓存数据，也 - 对每一块的数据 - 的指数，它指的是。所以，当你查找记录N，缓存不仅要持有创纪录的N值，而且N - 这样你实际上可以查找数据。这就是看它一个pretty简单的方式。高速缓存可能有其他的元数据来表示有效性，并最后访问时间等。

Caches have to store not only the actual cached data, but also - for every piece of data - the "index" that it refers to. So when you lookup record N, the cache has to hold not only the value of record N, but also N - so that you can actually look up the data. And that's a pretty simplistic way of looking at it. Caches may have other metadata to indicate validity and last access time, etc.

示例1：字节的32位地址空间的缓存

每个高速缓存条目具有存储的数据值（8比特）加上地址（32位）= 40位，

Each cache entry has to store the data value (8 bits) plus the address (32bits) = 40 bits,

示例2：32位字的32位地址空间的缓存

每个高速缓存条目具有存储的数据值（32位）加的地址（32位）= 64位，

Each cache entry has to store the data value (32 bits) plus the address (32bits) = 64 bits,

您可以看到，例如＃1具有显著更高的开销。

You can see that example #1 has a significantly higher overhead.

与往常一样，***可能会有帮助。 http://en.wikipedia.org/wiki/Cache_(computing ）

As always, Wikipedia may help. http://en.wikipedia.org/wiki/Cache_(computing)