I did not speak clearly enough. We can have thousands of items in the game at one time without tanking FPS, so the FPS impact of a single item is basically nil. To gain significant FPS reduction via items would need to create tens of thousands of items. So cleaning out 10 here 100 there is counting pennies, unless you are talking about rotting items (clothes, remains). Did food in stockpiles reduce FPS less than food sitting on the floor rotting? In 18 pages I didn't stop to read every word >.> It also seems like save/quit and reload is required after destroying a large batch of items to properly "reclaim" the FPS. By and large, empty spaces, repeating bridges, and caravans were the real FPS bandits (unless you go hog-wild on item creation).
If I understand your science correctly, you were creating batches of the same item. Thus, each time you tried to do something with a goblet/stone, EVERY goblet had to have a path check for it. In a proper fort, we don't have 10,000 of a single item- we don't even have 1000 of a single item typically (outside of stones). Did you ever try a test where each sweatshop produced a different (non-rotting) item, but kept the overall item count the same? Would 10k goblets create the same FPS issues as 1k goblets, 1k stone, 1k logs, 1k (stacks) of bolts, 1k shields, 1k picks, 1k bags, 1k mechanisms, 1k blocks, 1k soap. For just idling I'd doubt it would be so, but it would mean that during play limiting the total list size of any one category would be more important than total item count.
That isn't the case.
Nobody was pathing for these boulders or goblets.
They were simply sitting in place when the tests were taking place. I did not count the FPS drop-off from the need to pathfind, as that was not the purpose of the test. This was
purely the existence of the items, themselves, all of which were sitting in just 16 piles, which is
perfectly consistent with an undump or quantum stockpile, so your claim that it would have been better with a quantum stockpile is unfounded. I was not moving any of these items to some giant stockpile area and back, they were simply sitting there, not being pathed to or used in any way. (At least until mass dumping and elimination, whose main point was to test "contraction", and whether the game recovers FPS after items are destroyed, where FPS measures were primarily taken after all dumping and atom smashing was complete.)
In short, a quantum stockpile with 10,000 items on it will have a noticeable impact on FPS
even without anyone trying to pathfind to it at all. This does not have to all be of the same item. This is merely the effects of memory size bloat on tens of thousands of total items that have quality. (Qualityless items like boulders are less notable by about an order of magnitude, while rottable items like corpse parts are significantly more lag-inducing than mere qualitied items. Dead creatures are never removed from the list, and eat significant FPS forever.)
That said, there is certainly a
significant improvement in performance when a quantum stockpile is used with regards to pathfinding, which is certainly capable of being the single largest drain on FPS in many forts that are not optimized for pathfinding or which do not have significant item quantities, and if you are worried about FPS, quantum stockpiling what you need helps a great deal. However, saying that it is "A common myth is that more items=lower FPS, but this just isn't so," is just plain false.
Also, if you look over Loud Wispers's Silentthunders, he was talking about how his drink supplies had dropped from lack of trade to a "mere" 5400 from previously keeping over 12k drink, alone. (One of his screenshots posts over 50k drink...) (And he reports a consistent 3 FPS for the past few decades of fort time, so...) I have seen people with hundreds of thousands of boulders from positively huge megaproject excavations.
There certainly are people who will produce 10k cloth items "just to keep their dwarves busy" and clear out whole z-levels to stockpile it all, and hundreds of thousands or even millions of items total, and then wonder where their FPS went.
Author
Topic: FPS death (Read 22771 times)