FOXBORO FBMSVH calculates the additional delay of the hash function

People have proposed other schemes, such as slant caching,[13] where route 0 is indexed directly, as described above, but route 1 is indexed by the hash function. A good hash function has the property that when mapping with a hash function, addresses that conflict with direct mappings tend not to conflict, so the program is less likely to suffer unexpectedly large conflict misses due to ill-conditioned access patterns.

The disadvantage is the extra delay in evaluating the hash function. [14] Furthermore, when loading new rows and evicting old ones, it may be difficult to determine which existing rows have been used the least recently because the new rows conflict with data at different indexes in every respect; LRU tracing of non-skew caches is usually done on a per-collection basis. However, skew association caching has many advantages over traditional group association caching. [15]

Pseudo-associative cache
The true group association cache tests all possible methods at the same time, the [①③ ③ ③ ⑥] (9) ③ ③ ③ ⑦] use similar associative memory. Pseudo-associative caching tests one possible way at a time. Hash re-hash and column association caches are examples of pseudo-association caches.

In the common case where hits are found in the first test, the pseudo-association cache is just as fast as the direct-mapped cache, but it has a much lower collision miss hit ratio than the direct-mapped cache and is closer to the miss hit ratio of the full-association cache. [14]