Cache Address Breakdown
What you'll learn: How a memory address gets split into tag, index, and offset fields so a cache can quickly locate data.
The Core Idea
When your CPU needs to access memory, it doesn't send a single blob address to the cache. Instead, the address is cleverly divided into three distinct fields: the tag, index, and offset. Each field has a specific job in helping the cache find (or confirm) the right data block.
Think of it like finding a book in a library:
- Index = which aisle to go to
- Tag = which specific shelf in that aisle
- Offset = which page within the book
The Three Fields
Offset
The offset (rightmost bits) tells you where inside a cache line the specific byte lives. If your cache line is 64 bytes, you need 6 bits for the offset (since 2^6 = 64).
Index
The index (middle bits) selects which cache line (row) to check. In a direct-mapped cache with 256 lines, you need 8 bits for the index.
Tag
The tag (leftmost remaining bits) is the unique identifier stored alongside the cached data. After the index picks a line, the cache compares the tag in the address with the stored tag to confirm it's the right block (a hit) or a different one (a miss).
Why This Matters
Breaking addresses into fields lets the cache hardware work in parallel: it uses the index to jump straight to one location while simultaneously comparing the tag. This is far faster than searching every cache entry.
Key Takeaway: A memory address is split into tag (unique ID), index (which cache line), and offset (position within the line) to enable fast, direct cache lookups.