[SUGGESTION] TEXP leaderboard replacement

Discussion in 'Suggestion Box Archives' started by Burki, Apr 3, 2021.

  1. We all hate cactus farms right?
    Well, no.

    But I know there is plenty of people who think that there needs to be some kind of fix to stop the abundance of farms being created purely for the 'prestige' of being at the top of the TEXP leaderboards.

    These farms have supposedly caused lots of lags on utopia mainly but other servers too, whilst turning the TEXP leaderboards into a 'who can build the biggest cactus farm' competition, switching a previously competitive leaderboard into a bunch of crazy numbers with a huge difference between professional cactus farmers™ and the regular players.



    The main suggested fix for this has been to nerf the cactus XP received from furnaces.

    Personally, I don't think this is a good idea since it will just cause players to build even bigger farms to compensate for the XP loss. It also makes the smaller cactus farms, that people like me use for quick and easy repairs, much less effective. It would also be slightly annoying to all the people who have put time into their farms if they were suddenly much less worthwhile.

    My suggestion is to replace the TEXP leaderboard instead. I know the TEXP leaderboard has been around on EMC for a very long time and it seems crazy to get rid of it however it is fairly redundant nowadays. Before large scale XP farming was a thing, you could get a sense of who is active and farming XP with the leaderboards and typically anyone could compete and get high up just by using an gold farm or killing mobs and bosses in the wild however that it no longer the case.

    Replacement of the TEXP leaderboards will also mean that if these TEXP farmers really are doing it for the XP, they can still earn the same amount without any issues. It will just discourage anyone who is doing it purely for the leaderboards.

    One stat that I think would be much more interesting to have a leaderboard for is a 'blocks mined' leaderboard.
    • It isn't anywhere near as easily exploitable, since afk mining with macros has restrictions.
    • It will be a very competitive leaderboard as there's no real way to mine more blocks than others apart from simply putting a lot of time into it.
    • It also will help to boost the supply of many ores, stone, dirt or basically any block on the game since there will be more people going out to collect resources for the leaderboards than before. This is much more useful to the EMC economy than producing green dye to chuck into lava.
    I'm sure they could definitely be some other leaderboards for other stuff too, feel free to suggest any in the replies and let me know your thoughts on this controversial suggestion.
  2. Anything to help reduce the lag on Utopia I am down for. Tired of building with sub 15 TPS (usually around 13-9 now days)
    sonicol, Egeau and stormboy231 like this.
  3. Crazy how the texp has been around for YEARS and one player seemed to be on told of it for a few years with an xp farm, but now that more people want to compete for that spot, everyone wants to gone...Anyhow as I do believe if a farm (ANY FARM) is built wrong (without the care of server lag) it will hurt the tps. The claim that cactus farms/xp farms are the main reason for utopia lag is simply false information. The other day chick did a server test when tps dropped to 12 and showed us that current days tps factor. Farms such as cactus (with drop items) only accounted for 8% of all server lag, while villagers did a whooping 67% of all server lag. So please, don't use tps lag as one of the excuses. All those top 10 names only have 2 farms. Ours and random's on utopia. Ive talked to multiple SR and personally ask elfin multiple times to help us know if we are by any case causing any server lag because of the size of our farm. After multiple tests we are contributing for less then 3% of all server lag on utopia. Ive also been told that Random's farm isn't a lag contributor either. Now that we have the lag issue out of the way lets talk about the overall board.
    When I came back from break I found out the group xp gives you 5x the amount. This is something that can be changed OR make another section of the leaderboard for people who collect under the group benefit as it wouldn't be fair to someone without a group. This would both free up a lot of those leaderboard slots as well as let those who want to compete for grand xp, still have that chance.
    EfficiencyV likes this.
  4. I have had multiple conversations with SS and they have explained to me that it is not one farm in particular but it is when multiple farms are being loaded and that how is the tps dies. Just to be clear I have my own cactus xp farm but do not use it because the low tps and amount of lag frustrates me and I don’t want to be the one causing it. I agree that it should be removed and in fact it is going to be removed. I don’t believe you should build the xp farms just for competition on the leaderboard and that is not why I built mine. I have been to some of the farms and noticed things that could help reduce lag. I agree with Burki and like the blocks mined leaderboard.
    stormboy231 likes this.
  5. Couldn't agree more. Simple design changes such as not using kelp farms are fuel source will extremely help prevent lag factors. The use of less hoppers will extremely help prevent lag. And also not allowing items to float for every long through the whole farm will significantly help the overall tps. I also agree that it isn't just 1 farm killing tps, but if other farms also switch to methods to help prevent lag, all of them in use wouldn't factor a server very greatly. Ive personally been to/seen plots with a massive amount of villagers kill the tps rather then any xp farm...Dont believe me? Message me in game and ill be glad to show you myself, as I dont wish to put anyone on blast publicly.
    EfficiencyV likes this.
  6. Regardless of whether or not a single farm cases it, I feel the TEXP leaderboard itself is at least a factor in the rise of the amount of farms and trading posts that cause significant TPS issues. I do recall Aikar (correct me if I'm wrong) that they were thinking of removing the TEXP leaderboards to help with monetization and remove the "competative" feeling that they give.
  7. I’ve heard about that. Also I don’t believe trading posts are in direct correlation to the texp board and if people do build them for the texp, then I’ll just state that it is a very ineffective way of getting up the board. I understand if people build them for enchanted books and Xp for mending though
  8. Indeed. Now that a lot of people do something that does require significant server resources but doesn’t help anyone but themselves people want it to go.
    TEXP farmers have become a separate entity of the community, one which only negatively impacts the rest of the player base: bulk TEXP farming doesn’t give anything significant, it only takes.
    Happy to clear that up for you :)

    Misinformation is indeed another large source of lag. People usually don’t know what they’re doing, and then create lag machines. I am seemingly the primary educational source for people when it comes to lag proofing redstone, and I have seen absolute nightmares of storage systems, some of which might have indeed temporarily caused more lag than any individual TEXP farm.
    However, most people who make these large laggy storage systems say something like this:
    “I know it is more laggy than it needs to be, but my playing style cannot afford any more time spent on Redstone – I am simply too bad at it. What about the cactus farms? Those people do know their redstone jet create just as much lag as I do.”
    You see: what-about-isms aren’t arguments. The fact that there is one source of lag doesn’t mean that another isn’t something that is a valid thing to eliminate.
    People design systems poorly and we should help them not do that. I have spent countless hours helping people out with their redstone to lag proof it, sometimes going as far as to completely remake someone’s smelter because their old one caused too much lag and they were too lazy to fix it.
    However, the fact that these issues exist does not mean that other people can just go on causing lag because they want to be on the top of a leader board, and not do anything else useful.

    I would like to see those numbers myself:
    What’s the mean lag of the farm, and what is the standard deviation of the lag produced per five minutes?
    How sure are you that those numbers do indeed capsulate all of the farm’s (and further system's) produced lag, and not only the amount that can be clearly distinguished upfront? From my own testing with bamboo farms and smeltries, the lag as measured with diagnostic tools and the lag as measured by simply WorldEditing it out of existence and testing the MSPT again where almost 200%.
    If these measurements do indeed scale up properly and do indeed work across different farm designs, that would mean your farm uses roughly 10% of all server resources for Utopia. That is quite significant, and definitely not:
    I will also add that your eagerness to immediately move away from lag issues, even though it seemingly is the purpose of this entire thread is… suggesting of something.
    This isn’t a thing where agreement is necessary. We have the means to generate empirical data, and some of it already. Empirical data transcend the need for personal opinions.
    I have personally played around with the idea of setting up a farm like this, and did some of the necessary creative testing. The statistics I got out of my lag own measurements were not in your favour. Yes, these were with the vanilla game as opposed to EMC, so they are not perfect, but they do suggest that that 10% I got earlier is a low estimate, and that we are actually looking at roughly 20% of all server performance for a farm and smeltry that roughly equates yours. The total lag production of the existence of TEXP would then be around 50% of server performance.
    This number ignores farms that are only made for enchanting: You don’t need to store 400mil TEXP in furnaces, to plop out in one go. To reflect this, this number only takes into account the amount of EXP that is generated and then used wastefully.
    That 50% (The actual number is a 95% certainty between 38% and 59% - statistics are messy) assumes that EMC is not meaningfully different from the vanilla game, which I know is false.
    Additionally, my data on what is 100% of all server performance is messy: it is based on data that I gathered whilst being (completely) alone on the server in 1.12, which make it hard to actually base conclusions on.
    However, if it is true, it means that 50% of available Utopia server resources are used solely for a few players to get higher on the TEXP leader board, and not for anything else. Yes, it has both empirical and statistical uncertainty, but it is the best number I have been able to calculate.

    The lag data generated this way would be anecdotal, but, besides, that is just another shameless what-about-ism, not worthy of a response.

    My personal calculations do indeed support this argument as well. The 10MSPT that one farm uses is significant, but not drastic. Multiple farms would indeed cause the TPS to drop.

    If developers say it, and further properly gathered data supports it as well, I think it is reasonable to conclude that it is true.

    ---

    Now for the further argument: Assuming that my calculations are correct, and TEXP uses 50% of server performance on Utopia, or roughly 25MSPT, I do agree that the TEXP leader boards should be removed. TEXP leader boards to not help anyone but the people trying to get up the board.
    I have previously largely defended the construction of large farms, as it is an integral part of the way people play Minecraft: I definitely think that, if there is a way to automate something, it should be automated.
    However, there is a limit to this, and I think that
    -the uselessness of the TEXP leader board;
    -the general inefficiency of the way the EXP and items that are generated are being used; and
    -the relatively large amount of server resources that these farm use, in combination with them always being loaded,
    make this go across the line.

    The TEXP leader board has become a server minigame that has become to expensive too continue supporting, and should therefore be removed.

    (Sorry for my messy writing in this one… I’m somewhat tiered)

    EDIT:
    Because of Chicken’s post, as quoted below, I changed some phrasings to make it clear that all my lag testing did already include smelters and further processing of the items: The testing was done on a complete system of fuelling, cactus farming and smelting.

  9. I am just at my computer for a minute. But wanted to correct some misinformation stated here. The cactus growing do not cause lag. But the resulting items and other tile entities involved are contributors to the lag. (not the sole cause of utopia lag). Reminder to everyone to be kind to each other. Pointing fingers at each other is not beneficial.
  10. ahahah every time we collect someone goes and off and makes a post on the forums.
    Building massive xp farms across multiple servers is a lot of money, time and resource's, especially for the clout of taking that #1 spot. It gives us all something to strive for, with a goal in mind and a reason to log on.

    Once we do achieve our goal, who knows what's next?. An expansion of our gigamassive gold farm?, boss farming for a dc of vv's? all sound great but stupidly large cactus farms peak my interests, for now atleast.

    Yeah the leaderboard is useless, but removing it wont make us remove the farm, it will still get used and will be opened up to the public for tool repair, enchanting, etc, which has always been the plan as we know the board will eventually be removed anyway.

    People make it out utopia is full of xp farms when there's only 2 large active ones. (ours and Randoms).

    There are a few ways to build these farms,
    Bamboo or kelp
    you probably used flying machines to harvest these.
    We use bamboo which is harvested in 9 block sections using a observer once the bamboo reaches 4 blocks in height. Reason for this is because we first used 2 large kelp pools, which when harvested would tank the servers tps to sub 15.

    Also doesn't emc apply optimisations to improve server performance?. All these answers do need clarification from the developers.
    Also I would like to see actual testing, if you have it and not some number jotted down on the forum, which as far as i could know could be out of nowhere.

    as for a replacement for the leaderboard, tokens.
    EfficiencyV and stormboy231 like this.
  11. You might have noticed that it took me a while until I respond. That is because of this part of Envine’s post here:
    You couldn’t: all that was left of my testing were the results written down in a notebook and a world with different possible designs in it.
    So, I redid my testing. Here is everything again:
    I spent over four hours on this post, so I hope you appreciate all 3000 words.

    The first part of my testing is going to be the most difficult: Exactly measuring what 50MSPT on my system is for the system EMC operates with.
    I measured it twice, once in 1.12, and once roughly half a year ago, which confirmed my previous measurements. The result I got is that 1MSPT on EMC roughly equates 2.5MSPT on my system.
    When I did my testing in 1.12, which was using some weird behaviour in the hopper code to generate massive amounts of lag whilst completely alone on the server (no one online from SMP1 to Stage) (When deliberately trying, you can get a single hopper 1240 times laggier than a normal empty hopper: https://youtu.be/yINcXy-ajiU?t=562)
    The 1.15 testing I did on my farming residences at 1070, and was a lot less rigorous: EMC has (or had) some really weird code with entity collisions, which I found out by singlehandedly dropping the TPS to 7 whilst working on my hostile mob switch. I used this knowledge to make the TPS drop to 18, and then tested loading and not loading my farming district, this time only being alone on SMP1, and measured the difference. This could give me an MSPT value for my farming district, which wich I could calculate the rest.

    This 150% more performance on EMC is really weird, and Tom and I still think I might have made a mistake in my methodology. You see, assuming the data Aikar has given on what system he uses to run EMC is still accurate, there shouldn’t be any difference in performance on the hardware side of things:


    Minecraft is written in Java, and therefore doesn’t really do all too much multi-threading. Chicken has recently told me that they have managed to get some performance from other cores as well, but it’s still mostly single-core performance.
    Okay, but, maybe that CPU isn’t used anymore, and the people who run the server have upgraded. To another server-grade CPU. The fastest server-grade CPU on the market for sinlge-threaded tasks seems to be the Intel Xeon W-2275, with a simmilair preformance of that of the i5-10600


    Lastly, let’s compare to the CPU that currently has the best single-threaded score on the market, ignoring server-grading or anything of that nature:


    If 50% increase is litteraly the fastest sinlge-core increase you can get, that 150% must largely come from software upgrades.

    This is less likely than you might think because this:
    isn’t entirely true. I have made quite a lot of changes to the vanilla game, and, at the Minecraft client I use for lagg testing, all of those are to simulate the EMC system.
    Because I am quite obsessed with lag (and because I have apparently become the primary source people go to for lag busting, as suggested by the fact that I have helped with the design of two of the four large TEXP farms, among other things,) I have obsessed over running similar optimisations to EMC, and this 150% just seems rather unlikely.
    To give you an example, when 1.15 was on its way, I had a few conversations with Aikar on redstone dust optimisation (In the vanilla game, redstone dust turning off is by far the most laggy thing that is regularly done.) I had tested some different optimisations, he had too, and we both came to different results (only by a few percent,) which granted a short conversation.
    This also means that I know that the code EMC runs for redstone dust isn’t available to run on my system, unless I patch it in myself, something that has never lead me to anything but frequent crashes. (I know just enough about code to break everything, but far from enough to actually fix something.) Aikar’s testing has shown that the optimisation I run is within a few percent of the optimisation EMC runs, though, if I remember that conversation correctly.

    I mostly said what I quoted above because I know I cannot be certain that my lagg improvements match those of EMC, even though I try. I didn't want to explain all this, so I just didn't.

    Even though it seems unlikely, I am still going to use this 150%. It is the only empirical data I have, and I am going to give you as much leeway as possible, so that you cannot complain about my testing.
    Also, I cannot prove that my system’s lagg improvements roughly equate those of EMC, so I am going to have to go with the empirical measurements.

    Also, to do the tests, we need a decent null measurement, which is going to be of my redstone test worlds: a blank white stained glass layer at y=0, and nothing else.
    Here is it without anything in it:


    For all further tests farms are going to be the only thing present in a world like this, which is why I am going to assume a background noise roughly 1MSPT, which is also roughly my error margin.
    To be more precise: I am going to round down to the second nearest half, as that makes the maths a bit more clean.

    -

    The testing of the actual farms: 1: cactus.
    This one is both the hardest and the easiest: It is the easiest because there is a trivial choice of design type, which everyone I know uses, and it’s less trivial because the more aggressive stacking mechanisms is one of the things that I might have partially emulated, but haven’t copied exactly. That 150% should cover for that, though.
    This is the farm I used:

    It’s the trivial Illmango design, at exactly the size you have on 5045.
    The first glace isn’t at all in your favour:


    Here is the more precies testing data. I didn’t test it for hours, I am doing this for the second time, after all. These few minutes should give a rough idea, though:

    As you can see, it is pretty consistent at 20.5MSPT, or a 20MSPT increase over the standard.

    I tested this by having two emulated players standing on top of the farm, to then fly away myself, only loading the farm server-side, not client-side.
    I also tested this farm’s output:

    This is almost exactly 6 2/3 hopper speed.

    This concludes that the cactus farm has a total lag impact of roughly 3MSPT per hopper speed output.

    This is post 1/2
  12. Second farm: bamboo

    This one is a lot less trivial.

    I have tested a lot of bamboo farms, and, just for reference, this here is the least laggy bamboo farm I know of, which is also the design I use myself.

    This farm, at this scale, produces a little under 4 times hopper speed, once it has settled:

    These measurements suggest that this farm has a total lag impact of 0.6 MSPT per hopper speed output, which is close enough to my error margin that I would have to scale this farm up to be able to properly test it.

    There is a very good reason for not make this farm, though: it uses a 4-directional flying machine: those break every time you unload them without first stopping them, and break by starting to fly in a random direction once reloaded, meaning you have to rebuild every flying machine every time you get afk-kicked. For me, this is no issue, and I only had to rebuild it twice since making it last Christmas, but I understand it would be an issue for you.

    We’re going to leave flying machines behind and look at more traditional farms, which is also what I based my original testing on.
    The first design I want to show you is the one I personally use when I only need a bit of bamboo, scaled up. I usually only use it for 20 or so plants, or less than a SC an hour, when another farm I have sometimes produces stuff that can be smelted. (like a cooked chicken farm which, for some reason, also sometimes produces raw chicken)


    This is, as compared to the other farm designs, a pretty laggy one, at 5.6 MSPT per hopper speed output, but I still wanted to show it so you can see what an non-optimised normal farm would look like. It’s a bad idea to make this farm at a size of more than 20 plants, but it’s probably what most people use regardless.

    Here is your farm design. I saw it during construction, it is an exact copy of the farm you use (worked to a different scale, of course.)

    You already know this farm. As you can see, it works at 2MSPT per hopper speed output. It more than halves the lagg of the previous design, but it is also more than three times as laggy as my 4-way flying machine.
    Also, I would want to add that making this farm in the vanilla game is a bad idea: It uses quite a lot of redstone dust, which, as I said, is amongst the most laggy components in the vanilla game. I wouldn't be surprised if the impact of this farm is twice as large in vanilla.

    This is a new design I want to introduce to you. It’s a design by me, based on a SciCraft suggar cane farm. Instead of the observer-based farms we’ve seen previously, this farm runs on a clock, with the power slowly snaking through the farm.

    This screenshot here is probably enough to see how it works:

    I mostly like this farm because of its simplicity. Yes, it does have quite some slightly more complex redstone, but the base concept is really simple and clean.
    As for the lagg:

    This test here concludes that this farm would be at 2.5 MSPT per hopper speed output.
    In the longer tests I talked about in my previous post, this design was better for lag as compared to your design ever so slightly, but not statistically significantly. They are within error margin of each other. Aditionally, this design uses way less redstone dust, making it a lot less laggy in the pure vanilla game.

    The last design I want to share here last is the design that, in my testing, was the most lag friendly:


    This one looks a lot like your design, but it is mechanically quite different. It has similar 10-wide sections, but it runs on a clock / in a loop powering only a single 10-long slime structure every 3 full-delay repeaters, which this cross-section here should perfectly illustrate:

    Additionally, the grass sections are not on the same level:


    As for the lag, this design is slightly better than both designs I showed previously, ticking in at roughly 1.6 MSPT per hopper speed output:

    (As for why I tested this design so much longer: I was tick warping / running the game at 100TPS, and went for lunch. I knew I wanted to do the long-term testing with this design, as I already knew it is the best one)

    Lastly, here are the four bamboo farm designs next to each other:

    The measurements of these farms are, from left to right:

    1:
    13 x 70 x 82 (74 620) – 42 718 blocks to place at this size.
    2.7 volume per item/hour
    1.5 blocks to place per item/hour
    5.6 MSPT per hopper speed output

    2:
    17 x 70 x 82 (97 580) – 56 106 blocks to place at this size.
    2.4 volume per item/hour
    1.4 blocks to place per item/hour
    2MSPT per hopper speed output

    3:
    19 x 56 x 78 (82 992) – 42 142 blocks to place at this size.
    2.5 volume per item/hour
    1.2 blocks to place per item/hour
    2.5 MSPT per hopper speed output

    4:
    17 x 70 x 82 (97 580) – 49 370 blocks to place at this size.
    2.0 volume per item/hour
    1.0 blocks to place per item/hour
    1.6 MSPT per hopper speed output

    I am going to assume a 1.6 MSPT per hopper speed output for bamboo as of right now, even though I am really quite sure that no one on EMC is actually running a farm that efficient. Your farm is 25% more laggy than this.

    Also, I will add that, in my testing, a well-designed kelp farm is slightly more lag efficient per XP gathered than this bamboo farm. It’s not quite as drastic as I previously thought, since Equable made me aware that the amount of XP per smelted item differs per item smelted, which is something I didn’t think to test, but it should still be significant.
    However, crafting the kelp is a lot of work, and I don’t think that doing this is worth the slightly better lag performance. Kudos to those who do do it, though. :)
    (I've spent enough time today re-doing my testing - I'm not going to re-do that, too)

    Lastly, the actual smelting system.

    There are a lot of different smelters out there, most of them differ in the ways they distribute their items. I am not going to be able to test every single possible system, so, instead, I am going to test this system here:


    In the top barrels, there is cactus, in the side, there is bamboo. The dropper dispenser is just the trivial hopper speed design everyone uses (I hope.) Indeed, here I am assuming that the item distribution is so optimal the lag impact is negligible. I know it is not negligible, but this is simply a way to make sure everyone is happy with the results. No one can be more lag friendly than this.
    Or, well, it *might* be more lag friendly on a very large scale to pick the items up with hopper mine carts instead of with normal hoppers. I know most large-scale furnaces designed by expert redstoners do this, as they want to dispense to shulker boxes, but doing this on a level that it is significantly less laggy requires a level of redstone expertise I don’t think I have… I am a world record holder in the piston door community… I would consider myself quite good at redstone…

    Anyway, here are the numbers on that smeltry:


    As you can see, it isn’t all that bad. Being somewhere around 0.2 MSPT / hopper speed output.
    Let’s compare that to the system I know you’re using, with, instead of a barrel in the back, a DC with two hoppers on it in the back:


    This is quite a lot more, already, at 0.5 MSPT per hopper speed output.

    Let’s compare that to a system where we just don’t care about distributing the cactus input, probably placing different small furnaces next to the cactus farm, to then only distribute the bamboo properly.
    The cactus just inputs to a hopper line, and flows over different filled hoppers until one furnace needs the item.
    I know most people do it this way, and, honestly, I would too, as it is less laggy, at 0.3MSPT per hopper speed:


    Indeed, distributing the cactus with hopper mine carts (which are in this case even left out – only the exposed hopper is tested) is more laggy than doing it with just plain hoppers. I showed previously that you can generate a lot of lag by doing some weird stuff with hoppers, well, if you know some things about hoppers, you can also lag bust like this. Even though we use more hoppers here, it is less laggy overall.
    This is why the bamboo distribution for the furnace I just use to smelt stuff (which also is the furnace I use that 4-way flying bamboo farm for) looks like this:


    This is just genuinely the most lagg-friendly way to distribute an output of 4 times hopper speed.
    As you can probably guess, my smelter works at hopper speed, I’ve got that 4-way-flying machine farm at a neatly perfect 4 times hopper speed: it is a really neat and clean system.
    If you want a more equal and fast distribution for your smelter, which you do if you are just using it normally, I would suggest making the distributer of Illmango’s perfect smelter.

    Anyway, I think that that is all testing that needed to be done. Now it’s time to fact-check myself:
    I have one measurements that give away how fast your farm is: A cactus farm of the exact same size and design of your farm outputs 60000 items per hour, as I have shown previously. This would be 151 furnaces smelting continuously.

    For that you need: 6.6*0.5+6.6*3+6.6*4*1.6 = 35 MSPT on my system. Which, with that 150% I got earlier, is 14.12 MSPT on the system of EMC, or a little under 30% of all available server performance. As you can see, I always round down when I discuss stuff without giving proof. My longer testing was at 25%, so this one is probably on the high side. (I know you shoulnd't round half-way, but the difference is insignificant, so I'll just write it down like this.)
    Also, this testing was a lot less proper. My previous testing was a lot more scientific: for example, here, I assumed the increase in lag as the farm gets larger to be linear, which I tried to prove for each farm in my previous testing. Also I tested 1, 2 and 3 modules, and then calculated the average marginal increase, which I still think is the proper way to do it.

    I think I should give some more context to that 35MSPT on my system. Here is a sorting system that I designed for the new recycling centre:

    This system sorts through 800 different items, and has a storage capacity of 2DC for each type. Using roughly 1500 hoppers.

    Here is my personal storage:

    This system uses this many hoppers:

    It stores items in more than 3DC of chests. Even though I have optimised it quite a lot, I still call it “the lagg machine.” Also, I think that the fact that I can run my storage 6 times on my system without the game lagging (that is what that 7.2 means, after all) shows quite well how good my lag optimisations are, both in-game redsone wise, and from a modding perspective.

    -

    TL;DR:
    My, this time really transparent, testing, shows that the farm Stormyboy, Envine & co. use, uses roughly 30% of all available server performance on utopia. Some context on how much that is can be found in the paragraphs right above.
  13. Alright so I want to fix a few things that you've stated above....First off, you should have just copied over our bamboo farm just like you've copied somewhat of our cactus farm. We used observer per section rather then a clock or a full observer as both cause way more lag than ours. We only cut 9 blocks worth of bamboo once that section reaches a height of 5 blocks. Also, I understand and believe your testing but I also have to lean towards the testing of actual server staff. They seem to have direct analysis based on the server's performance. What you have showed would tell me that you are somehow getting 10x whatever has be told to use from multiple testing based on the farm running 100% on the server rather then a test world
  14. I see this as an invitation to use world downloaded on your residence. Using this without permission is not allowed via EMC rules, so I didn’t do it jet: I just wrote down the proportions of the cactus farm.

    This is what I figured out from the world download:
    1: Bamboo farm #2 is identical of your farm as currently located on Utopia, and all testing of that farm is accurate to your actual design. I can definitively say that your bamboo farm is roughly 25% more laggy than the best design I have been able to come up with in a few hours.
    The only difference between your farm and my re-creation is that I, whilst copying it over, found a small resource optimisation (something that only affects the amount of materials used, and nothing else) so obvious that I assumed you would spot it whilst making the farm. I pre-emptily corrected it, as I assumed you would, too. You didn’t.




    2:
    Just so you know: you don’t have to do item pickup like this:

    This works too:

    3:
    I can also definitively say that my recreation of your smelter is accurate.

    Speaking about the furnace: Why on earth do you have 310 furnaces, all equipped with 5 hoppers? Your cactus farm works at 6.6 times hopper speed: you only need 152 furnaces for that. You are using 158 furnaces, and 790 hoppers without necessity. That is roughly 1MSPT or 2% of all server performance, just because of plain bad design.
    4:
    What the actual Flippin’ crap were you thinking when you designed this:

    Honestly, the lagg impact of a dropper dispensing system like this isn’t even all that much, but it’s just stupid. It’s completely unnecessary and you should be glad that EMC runs good redstone wiring optimisations, otherwise this alone would be 5-6 MSPT of unnecessary lag.
    When I said this:
    That “I hope” was a joke, because I thought no one in their right mind would be so bad at redstone as to not know how to do a basic dropper dispensing system, and still say they are confident with lagg proofing. You proved me wrong, kid. You proved me wrong.

    I know you use an observer per section, as stated above, my recreation of your design was accurate. Your design is #2, those tests are of your design.

    As empirically tested above, the second statement of your phrase is false: a farm that runs on a central clock and slowly moves the power through the farm is less laggy than your farm. Farm #2 is 25% more laggy than farm #4
    Honestly, most of the lagg optimisations that farm #4 makes over farm #2 are trivial. I spent less than two hours on the design (as compared to 6 hours on #3, which I thought, when designing it, would be less laggy) It honestly astonishes me that you do not see the lack of optimisation that your farm has. The concept is decent, but the execution is sub-par at best. You should have really asked someone who actually knows what redstone is, there are plenty of decent redstoners on EMC.
    That a farm that runs on a central clock and slowly moves the power through a farm is less laggy than an update-based farm should be basic knowledge among people who do lagg proofing. It should be obvious in itself: You are testing for a single bamboo to grow a certain height, and then cut it. You only measure one of the 10 bamboo plants in every section, meaning you just effectively put 9/10 bamboo on a clock of random delay, that is incorporated in the farm.
    Of course that is sub-optimal! By reading the game code and using a logarithmic distribution, you can calculate the optimal time to harvest a bamboo by simply minimising the cost function. The most complicated thing about that is just taking the derivative of a function.
    By definition, for those other 9/10 bamboo plants, this timing is more optimal than the timing you get with your system. This is just basic logic and some simple maths at this point! Additionally, by doing it like this, you can be sure that the distribution of the lagg is equal, as you always know when every section is going to fire.
    You seem to think that flying machines are more laggy, but that is not true, they simply have a worse spread when poorly designed. This is why the 4-way fling machine I use is less heavy on server-resources than any clock-based farm: it, just like the clocks, harvests all the time, and, with that, spreads out the strain evenly.


    Having now seen your (partially deconstructed) kelp farm, you simply used the worst kelp farm imaginable, that did not spread out the lagg at all. No wonder that your kelp farm was a decrement to the server every time it harvested!

    So:
    You furnace array can be made 50% less laggy, your bamboo farm can be made 20% less laggy: the only thing you didn’t mess up was the cactus farm, and only because that is a farm so trivial that it would be special if someone managed to do something stupid with it.
    Also, you did kind of did mess the design up:
    This

    Should be done like this:

    As the main distributer of the farm you used said.

    Tested to be false.
    Implies that flying machines are more laggy than non-flying machines. Also tested to be false, in case two well-designed farms are compared.
    Tested to be false, if two well-designed farms are compared.
    Tested to be false in specific cases. The "type" (location, average use) of a hopper matters greatly.

    I know that that dropper dispensing system might seem like a trivial problem, but it is a great way to capture my general feeling going through the world download. Your redstone is poor at best. You clearly have no clue what you’re talking about: take it from a world record holder, take it from someone who worked with the SciCraft team, your claims that your farm is as lagg-friendly as it can be is just utter crap. Your designs are sub-par and generally poorly executed at best.

    I am probably more angry than I should be, but it seems like any form of logically structured argument just goes right past you. See this anger as a clear indication that 1: I am fed up with your untruths and don’t want to bother talking to you anymore; and that 2: you should maybe try to properly read a structured argument and then reply accordingly, instead of poorly reading something to then word-vomit one large unstructured paragraph filled with what-about-isms and anecdotal evidence.

    I am not going to be responding to any further comment of yours. I have spent several hours debunking all the blatant and utter untruths you have spread, and I have enough. If you manage to get SS to do some proper empirical testing, you should contact me through them, as I am not going to respond to anything you say: too much of what you say is simply false.

    You should be ashamed of yourself.
  15. To claim the texp leaderboard being replaced and/or removed will not fix a majority of the Utopia lag, is absurd and frankly ignorant. It will most definitely help with lag, especially on Utopia.

    Also kudos to Jelle for the time and detail they have put into researching this issue given the circumstances.
    pewdiepiggy and Egeau like this.