Different game, different problems, but I've spent a fair amount of time looking at the source for Civ IV's core game DLL (released by Firaxis for modding).
It was fairly simple to get significant performance gains (without changing AI behavior or results) just by applying straight forward changes. Eliminating macro's hiding chained method calls. Saving the result of nested calls, e.g. save a pointer to the game map instead of making the same call repeatedly in a tight loop.
There's a huge difference between "premature optimization" and just sloppy coding. Too many take the idea too far that the (C++) optimizer is better than you'll ever be and that you should let it do it's job. Yes, the optimizer is/can be very good and can make surprising and unintuitive changes to make code faster. But, it's not a miracle worker. Optimizers work best on clear, concise, code with minimal data dependencies.
FWIW, back on my Core I2 Duo, a single late-game turn on a huge map could take upwards of 30 minutes to complete, most of that spent in AI's turns. I was able to get it down to about 5 minutes without too much effort. Along the way, RAM usage also plummeted. If I had full access to the source, I could have made even greater gains that would have broken the DLL ABI...
Sadly, I never released my changes anywhere (and I only had local SVN at the time), and they haven't survived my PC upgrades since.
No they weren't using bubble sort. When parsing JSON, they were calling sscanf (which internally was calling strlen). In addition, those parsed objects were stored in an array which was iterated through on each insert to check for duplicates.
There was an article [0] and subsequent discussion on HN [1] about reverse engineering and patching the binaries.
No? If you check your own sources, you will find that this is precisely the type of case where a developer is using an O(n^2) algorithm out of ignorance when they should have been using a faster algorithm, and that the high load time was caused by this mistake. The time complexity for bubble sort is also O(n^2), compared to O(nlogn) for reasonable sort algorithms. I don't really understand how you take all this information and then conclude "no".
The way accidental O(N²) comes about isn't "I'm intentionally using an algorithm that's O(N²)" (e.g., bubble sort), but "this method that I think is O(1) is actually O(N)" (whether because it's badly designed or just badly documented).
It's really not in the same vein as using bubblesort instead of a mergesort: failures aren't in algorithm design but in the actual mechanics of the implementation that is ancillary to the algorithm itself.
> failures aren't in algorithm design but in the actual mechanics of the implementation that is ancillary to the algorithm itself.
Strongly disagree. The GTA developer basically designed an algorithm which loops items in an array, and for each iteration in the loop, they parse the entire JSON that contains all items, and they do lookups for duplicates from an array instead of from a set. This is like a quintessential example of an algorithm design failure, so it seems weird to me that you're not attributing it to a failure in algorithm design, but instead you pass the buck to creators of the library that the developer was using. Sure, the library methods could have been documented better, or implemented differently, or whatever, but that seems like an entirely unrelated discussion.
> "this method that I think is O(1) is actually O(N)"
I'm fairly certain the developer in question does not understand / care what big-O notation means. Therefore, this almost certainly is not the thought process that lead to the slow load times in GTA. More likely the developer was just mashing things together until "it works".
Bubble sort provides O(n^2) time complexity for a task that can be easily solved with O(nlogn) time complexity. The 5-minute load time for GTA was caused by a developer using an algorithm with O(n^2) time complexity when they could have used an algorithm with O(nlogn) time complexity. So, yes, I would say it actually was the reason (unless you are specifically asking if the bad algorithm choice was bubble sort, in which case the answer is no).
It was fairly simple to get significant performance gains (without changing AI behavior or results) just by applying straight forward changes. Eliminating macro's hiding chained method calls. Saving the result of nested calls, e.g. save a pointer to the game map instead of making the same call repeatedly in a tight loop.
There's a huge difference between "premature optimization" and just sloppy coding. Too many take the idea too far that the (C++) optimizer is better than you'll ever be and that you should let it do it's job. Yes, the optimizer is/can be very good and can make surprising and unintuitive changes to make code faster. But, it's not a miracle worker. Optimizers work best on clear, concise, code with minimal data dependencies.
FWIW, back on my Core I2 Duo, a single late-game turn on a huge map could take upwards of 30 minutes to complete, most of that spent in AI's turns. I was able to get it down to about 5 minutes without too much effort. Along the way, RAM usage also plummeted. If I had full access to the source, I could have made even greater gains that would have broken the DLL ABI...
Sadly, I never released my changes anywhere (and I only had local SVN at the time), and they haven't survived my PC upgrades since.