Chainly | NET object a fluent interface regardless | Validation library

Everything is different.

An unordered_map has the concept of buckets -

A bucket is a slot in the container's internal hash table to which elements are assigned based on the hash value of their key. Buckets are numbered from 0 to (bucket_count-1).

An unordered_map calculates hash value of the key which points to a bucket. The desired value is in that bucket. Now note that multiple keys can point to a single bucket. In your case it may even happen that um[0],um[5] and um[9987] all lie in the same bucket! Search within bucket is linear in time.

In this situation, does std::unordered_map produce a better solution for the problem?

In case you have sparse data, use an unordered_map but with an appropriate reserve (or no reserve at all and use the default allocation policy). There's no point if you do a myMap.reserve(MAX_ELEMENTS) since that will again just lead to memory wastage.

Else, use a vector. You get a guaranteed O(1) lookup. Since its linear its super cache-friendly. Whereas on an unordered_map you may get the worst case lookup of O(N)