Sub-Millisecond Query Optimisation
Here’s a summary of the queries needed to add 28,000 poker hands to Poker Copilot 2.
The average query time is 0.24 thousandths of a second. Which seems okay, except that 660,000 queries in total were performed. A very small number multipled by a very large number can result in too large a number.
I’ve never had to optimise queries before that were already running in less than a millisecond. Typically, a few milliseconds per query is satisfactory. This is just one of many technical challenges behind Poker Copilot. I enjoy technical challenges.
An advantage of such fast queries is it allows good concurrency.
Now I have two possibilities for future optimisation. Make the queries faster still. Or execute less queries, possibly through more aggressive caching.