Ok so I do want this to be in the description but I need to quickly send this to a community that doesn’t use discord so
- the reason why the spike gets larger when you repeatedly reload the spawner is not because of some java memory cache loading speed issue. it is not just related to, but quite literally caused by treasure bug: when a spawner is unloaded, it doesn’t get properly removed from the list of tile entities while it gets unlinked to the physical spawner block, so when it is reloaded, another copy gets added to the list, so there exists two copies of the same spawner in that list. when the spawner block is mined, only the one copy that’s linked gets removed, leaving a ghost copy that still produces particles and causes lag that can be detected by the root.tick.level.entities.blockEntities pie. the same thing happens here, you can quickly verify that by seeing the amount of particles emitted by a repeatedly loaded spawner grows linearly as the amount of times you reload it increases. note that this exact process happens to other block entities like chests. also note that this is just a client side issue, so you can’t just repeatedly reload a spawner for 10 times and get a 10x speed spawner.
- the way that ranked fixes the treasure bug i believe is it removes all copies of that same spawner when you mine the spawner(but not any other block entities), but not fix the duplicates caused by the incorrect spawner unload, so this repeated reload can still cause linearly more lag that shows on the pie. this means that chests for example, also gets duplicated in this process.
- as a result, if there are enough chests in the same render distance ring of chunks as the spawner, when you load that ring, the chests can negate the effect of the added spawner on the root.tick.level.entities.blockEntities pie and the spike can sometimes be even smaller(normally, as the chests and magma cube spawner is close to the player, they don’t get repeatedly reloaded and so don’t get duplicated in this process so they have less and less of an impact as you reload more times). to solve this, you can use the second value shown in the same row as […]minecraft:mob_spawner, which shows the overall percentage the lag caused by it, so it has a much bigger denominator and therefore less prone to the effect of other block entities.
- that value is normally 0.00%, so we will have to repeatedly reload the spawner enough times that it at least shows 0.01%. to somewhat decrease the denominator to make the number bigger, one thing that i found helped quite a bit is turning off VSync and FPS set to unlimited. other things i tested such as looking into the sky, flashing F1, turning down all video quality settings, seems to help very little. the amount of times that i have to reload the spawner in the bastion practice map increases as the amount of chests increase(they seem to do more than just increasing the denominator) from ~10 for no chests to ~40 for max amount of chests i encountered to get a stable spike that doesn’t flash between 0.00% and 0.01% at 18 render distance. also, for my hardware it seems to be working even if i immediately start unloading chunks after reloading and also the other way around(you can verify by checking the particle count when close to the spawner in the debug menu below entity count, if you did it correctly you should have pretty much the exact multiple of times reloaded compared to what you get from a normal spawner), so i was able to fit ~15 fully overlapped reloads into the treasure route.
- that is not exactly a lot so unfortunately you probably have to deal with flashy spikes with this method. also note that the final spike number(or in this case, more like ‘flashiness’) you get from loading the spawner with basic pie ray(negative pie ray seem to be worse for dealing with flashy spikes) is not exactly the same spike you get from 18 render distance, because you can unload all the outside chests. its not necessarily even the same spike you get from the render distace you raised to get the spike, because you might have never loaded the other parts of that render distance ring. also, the spawner you are pie-raying is not necessarily even the closest spawner to you, because it might take two spawners to stack up the lag to break through that 0.01% barrier and when the first spawner loaded it never spiked. to deal with all of this, i recommend refreshing pie 3 times after each action if you get a flashy spike and comparing the results to that of what the previous result is. this is indeed much slower but it saves you from potentially having to reload 30 more times and is still faster than breaking the spawner.
added information:
- the normal glowdar spike seems to be heavily surpressed when the spawner is culled, where as little as 1 chest in the same ring can negate the spike completely; it also happens just slightly more in real ranked games because ruined portals do not spawn in the bastion practice map.
- always using the second value is slow so maybe when seeing the second value not spike to consistent 0.01% at 18 rd you can use the normal value first and when that doesn’t work you can resort to reloading a bunch more times and use the second value.