Some countries have integer numbers printed on the transport tickets, for example, in ex-USSR. The ticket is "happy" when the sum of the first half of digits equals to the last half one. Then you should eat it and make a wish.

In theory, hash map structure should be very fast on retrieving by key operations, close to O(1) like for an array. A sorted map (keys are sorted) should be O(log₂ N) instead. The following test checks the difference.

Common scenario

We will use one tester class per type of tested structure. The keys are string[10] and are generated randomly in prepared array to avoid additional checks whether the key exists or not.

Let's make a simple benchmark test for random and sequential access. I will compare:

• array with predefined size (allocated on stack)
• TList container class (allocated on heap)
• dynamic array (allocated on heap)
• generic TFPGList (allocated on heap)

I use an array of 10M elements having integer and varaint data type. You can easily change its type definition in "type" section if need.

I know not much feedbacks about real tests on hyper-threading in OLAP world. Here is a SQL Server hyper-threading from Linchi Shea.

So, using hyper-threading doesn't mean that you will get the advantages always. At least some performance tests are required to ensure this technical choice.