Direktmappad

Direktmappad (direct-mapped) cache, in computer architecture, refers to a cache organization where each block of main memory maps to exactly one cache line. This one-to-one mapping contrasts with set-associative or fully associative caches, which allow a block to reside in multiple lines or sets.

Operation: An address is divided into a block offset, an index, and a tag. The index selects one cache line; the block offset locates the data within the cache block; the tag is used to verify that the stored block corresponds to the requested address. On a cache hit, data is read from the line; on a miss, the cache fetches the corresponding block from the next memory level and stores it in the selected line, evicting existing contents if necessary.

Characteristics: Direct-mapped caches are simple and fast, with low hardware complexity and predictable timing. They typically require less silicon area and power than more associative caches and provide straightforward circuitry for tag comparison and data retrieval.

Limitations and performance: The fixed mapping means two memory blocks that map to the same line contend for that line, causing conflict misses if accessed in alternation. This can reduce the hit rate for certain access patterns, particularly with poor spatial locality. Overall performance depends on workload locality and cache size.

Applications and configuration: Direct-mapped caches are common in simple processors, embedded systems, and microcontrollers where cost and power efficiency are priorities. Typical designs use power-of-two sizes, so index and offset lengths are simple to implement; block sizes often range from 4 to 64 bytes, with the number of lines determined by architectural goals.