CPU - implementation changes would help this feature shine

Discussion in 'FAQ & Feedback' started by Average, Nov 14, 2019.

Tags:
  1. Average

    Average Commander

    Joined:
    Jun 22, 2019
    Messages:
    144
    Likes Received:
    121
    I applaud the devs for adding CPU - creating a kind of limited 'currency' like this you get to 'spend' on the parts you most want on your ship, rather than setting boring arbitrary limits, is a great way to go. It means we can move towards ships that are specialised missile ships, or lightly armed transports, or scouts etc. (edit - in current form, CPU is doing this really badly, I hope its changed eventually)

    While I think a lot of the backlash that seems to be going on around it is from people who don't really understand the design principle behind CPU (although I get just not feeling its needed when the game is so much fun), there is some ways it could be improved to fix a couple of serious flaws and make it shine.

    ISSUE - TIERS AND LIMITS
    The biggest problem IMHO are the tiers and limits on CPU blocks. This still sets arbitrary level of CPU that any good ship will always maximise (ie. most ships flying around will have the exact same CPU). While that's not the end of the world in itself, it does mean that CPU-linked items do not scale smoothly with ship size. So you still have two fairly boring optimal designs (the theoretical 'best' ship) : the minimum ship size needed to fit the full set of CPU and weapons, and just an arbitrarily large sath-brick.

    Ship size optimums are REALLY important in a free-building game like this. You want the game mechanics to create many 'best' ship designs of different sizes, so you have a wonderful ecosystem of ships both flying around and also in fleet battles (either PvE or PvP).

    It's also an obvious problem that the "minimum sized ship that gets full CPU" and a ship 10 or 100 times the size get the same number of guns and other stuff.

    A nice ship 'ecosystem' to aim has very diverse ship sizes, each useful, but specialised for specific tasks. This is more fun than just a 'good' 'better' 'best' progression, and also feels much more authentically sci-fi.

    [​IMG]

    FIX STEP 1 - CREATE AN INTUITIVE AND ELEGANT SYSTEM BY REMOVING CPU TIERS AND HARD CODED LIMITS
    Consider removing the tiers and hard coded limit on CPU blocks, and instead linking the CPU to the ship size. This means CPU is always appropriately balanced to your ship - bigger ships get more stuff. In order to put some smooth sensible limits on this, make the amount of CPU subject to diminishing returns - doubling ship size gets only 50% more CPU, for example.

    My preferred option to realistically link CPU to size is to require large amounts of power for CPU, so that huge numbers of generators ("ships engines") are required. Large enough to be a large part of the ships interior. If you want more guns (or other stuff), you need more space (ie bigger ship) for bigger engines. Similar to most science fiction universes.

    If you didn't want to do the "engines" thing, you could alternatively link CPU more directly, like for example your CPU module outputs an amount based on the number of blocks in your ship. This is less immersive, although it's very simple.

    FIX STEP 2 - LIMITING BIG SHIPS ADVANTAGES ELEGANTLY
    Of course, now people will make the biggest ship possible to maximise CPU. So, aside from ensuring diminishing returns on how much CPU the ship receives for greater size, you can also provide a natural advantage for small ships, so that "small ship" is a great design choice in some situations. The obvious trait to link to size is maximum speed. Smaller ships get to move around faster, big ships are slower like they should be. Small ships get a chance to run and evade combat when attacked by a bigger ship they have no chance to beat. Interesting design choices emerge - should my cargo transport be large enough to have weapons in case its attacked, or be unarmed but fast enough to run away?

    Multiplayer PvE/PvP tactics also emerge where smaller ships can be use to chase and tie up an enemy while a bigger war ship arrives. If the biggest most powerful turrets had slower projectiles, you could even make it possible for a small ship to evade fire from typical big ship turrets. This is the way combat variety is created in EVE Online and a number of other games, where good pilots in small ships surprise and beat bad ones flying super big ships. Small ship, medium ship, big ship are essentially like *rock*, *paper*, *scissors*, so there's never a single 'best' ship to fly and you've got people flying all sorts of crazy stuff even in late-game.

    SERVER MANAGER COMMUNITY
    I'd also suggest that the server manager community could provide some free crowd-sourced experimentation with the way CPU works if the tiers and restrictions were removed.

    [​IMG]
     
    #1
    Last edited: Dec 23, 2019
  2. geostar1024

    geostar1024 Rear Admiral

    Joined:
    Jan 24, 2016
    Messages:
    1,483
    Likes Received:
    2,459
    I'm going to stop you right there on this point. Messing with max speed is both completely unnecessary and produces unexpected results. In the past, the devs decided to make max speed a function of mass, which, as one could have easily predicted, resulted in CVs that were unable to warp because they couldn't reach the threshold velocity for the warp drive to activate. As it happens, something similar was introduced (whether intentionally or not) during the introduction of better aerodynamics, and it is the main culprit behind the poor performance of low-thrust HVs and SVs (and needs to be removed).

    If you want to limit ship size, all that's really needed is the square-cube law and exposed thrusters. You can simply set the thrust/area for the largest thruster such that ships larger than a certain size (presumably packed uniformly with fuel tanks, generators, CPU extenders, etc) simply cannot develop enough thrust to match that of smaller ships. If you really want to put the hurt on large ships, a heat exhaust mechanic would be the way to go (a large enough ship couldn't exhaust heat fast enough to keep from cooking itself).

    In any case, I agree with your overall assessment that it's the architecture of the CPU system that is the problem. Simply making it actually modular would fix most of what's wrong with it.
     
    #2
  3. Average

    Average Commander

    Joined:
    Jun 22, 2019
    Messages:
    144
    Likes Received:
    121
    Thanks for the comment geo. You mentioned that messing with speed is not needed and produces unnecessary results. On the first part of that, I think there's a good case that it may be necessary. Partly to address what I mentioned in my first post, but also because it doesn't make a lot of sense that all CVs fly at exactly the same speed. As well as being aesthetically odd, this has undesirable gameplay consequences - more than 50% of space PvP encounters end in one player being just outside the range of the stronger opponent, and staying exactly there for a uninteresting chase that sometimes goes on for ages if they are determined, because they fly at exactly the same speed and they either lost their warp or can't risk changing direction long enough to warp. Even varying it a little would be an improvement in my opinion - at least someone could either cleanly get away, or they'd get chased down.

    Your point on unexpected results is well taken. I agree with caution, but I think the examples you list are, as you mention, easy enough to predict and potentially avoid. Speed interacts with a fairly low number of other things - warp and weapon fire are the biggest ones that come to mind. If we were to make the range of speed difference 40-120m/s, and set the warp requirement proportional to the max speed, would you consider indulging in the possibility that this might be worth testing, perhaps as an optional feature for willing servers to experiment with?
     
    #3
  4. geostar1024

    geostar1024 Rear Admiral

    Joined:
    Jan 24, 2016
    Messages:
    1,483
    Likes Received:
    2,459
    Acceleration is much more important than max speed. In the case that all ships can travel at the technical max speed, a ship with higher acceleration can always escape from one with lower acceleration, by spiraling away. If the higher-acceleration ship is chasing the lower-acceleration ship, then the higher-acceleration ship should be able to reach the technical max speed sooner than the lower-acceleration ship and have a good chance at getting into weapons range by the time the lower-acceleration ship reaches the technical max speed. If not, well, the attacker might just have to give up the chase.

    The thing is, it's not the max speed that's the problem, per se, it's the fact that it's trivial to equip any CV, regardless of mass, with multiple g's of thrust in every direction, allowing any ship the ability to get to the technical max speed nearly instantly. If accelerations were lowered across the board, then differences in acceleration between ships would become important (the difference between a 50 m/s^2-capable ship and a 100 m/s^2-capable ship is fairly small given how quickly each of them reach the technical max speed, while the difference between a 5 m/s^2-capable ship and a 10 m/s^2 ship is considerably larger). One byproduct of this would be limitations on what kind of ships would be capable of making planetfall safely (thus creating an incentive for dedicated landers). Another byproduct would be the likely significant reduction in lateral acceleration, making it generally harder for a ship to juke its way out of the way of incoming fire.
     
    #4
    Kieve and Arrclyde like this.
  5. Average

    Average Commander

    Joined:
    Jun 22, 2019
    Messages:
    144
    Likes Received:
    121
    I don't agree that it's much more important than speed, when it comes to chases. It seems obvious to me speed is going to be the primary factor in a chase, unless warp is involved. That's not to say acceleration isn't a factor, but I don't see how we can discount speed, its obviously heavily involved unless acceleration is so low that you can't even change your ship direction significantly during the period you are in weapon range.

    That said, independently I totally agree about greatly reducing CV acceleration. Given the unintended consequences thing you pointed out earlier would apply at least equally in that too, would you agree they could both be implemented for positive effect on the game if done with appropriate caution?
     
    #5
  6. geostar1024

    geostar1024 Rear Admiral

    Joined:
    Jan 24, 2016
    Messages:
    1,483
    Likes Received:
    2,459
    If neither ship has reached the max speed, then the max speed is irrelevant. Even when both ships are traveling at max speed, it's still acceleration that lets a a ship with more acceleration escape from one with less (by spiraling, as I mentioned).

    It's not the same, though. Changing thruster output doesn't change anything about the physics model, while a mass-dependent max speed does. All the effects of reducing thruster output are trivial to predict, because the physics is the same. Consider what happens if block/device masses are increased: with normal physics, some ships might not be able to make it off a planet without additional thrusters; with mass-dependent max speed, all ships will see their max speeds drop.

    Deviating from normal physics is just not a good idea in general (sometimes it's unavoidable, like the technical max speed), because you start to lose the self-consistency of physics that keeps everything properly balanced.
     
    #6
    Kieve and Arrclyde like this.
  7. Average

    Average Commander

    Joined:
    Jun 22, 2019
    Messages:
    144
    Likes Received:
    121
    How likely is that? You're talking about some VERY low acceleration there. Like multiple orders of magnitude. I think you have to justify that a lot more to propose it as the alternative that is less likely to cause huge unexpected changes in gameplay?

    But how do you not feel speed is a factor at all? Even a slight speed difference would resolve a chase faster than minor acceleration advantages used to slip away slowly with spiralling. Perhaps we have a very different idea of spiralling - I assumed you meant deviate from the chase vector in certain ways that caused distance to increase when the other player didn't or couldn't correct. I've done plenty of PvP, perhaps you could elaborate, as this doesn't make any sense to me at the moment?

    As the arbitrary and unrealistic restriction is in place, and as it currently impacts the game, what's wrong with changing how that works? It doesn't increase the amount we're already deviating from physics. And keep in mind a game with accurate space physics isn't much fun, as you'd be flying past eachother at 50,000m/s (and be obliterated with one cannon round).
     
    #7
  8. geostar1024

    geostar1024 Rear Admiral

    Joined:
    Jan 24, 2016
    Messages:
    1,483
    Likes Received:
    2,459
    As I noted, the major change in gameplay would be that it would be harder for ships to get off of planets without more thrusters. Dedicated lifters would be quite necessary as soon as you started to haul around significant resources (and the square-cube law would act to limit the size of those ships anyway, due to running out of ship surface for thrusters). HVs would see a lot more use as haulers and tanks (given that hover engine stats wouldn't change). It would take longer to get up to speed in space, but that's not the problem it used to be given that we now have an adjustable cruise control. The difference in time to go 10 km with instantaneous acceleration vs 1 m/s^2 acceleration is just a factor of 4 (200 seconds vs 50 seconds; note that at 16 m/s^2 acceleration, the times are equal). So, I'd argue that there's no huge unexpected changes in gameplay that will result from reducing thrust, and that the changes that would result would be quite beneficial.

    That is indeed the same kind of spiraling I'm talking about. I'm not arguing that differences in max speeds wouldn't make it easier to escape, just that it's not necessary.

    But it does change the fundamental physics for all speeds below the max speed: you go from being able to reach any speed if you burn for long enough, to having your speed capped due to your ship configuration no matter how long you burn for. Is the technical max speed a deviation from physics? Yes, but that doesn't mean we should do further deviation when at speeds below the technical max speed.

    Another thing to consider is that mass-dependent max speeds don't combine nicely with cargo mass; you're essentially getting double-taxed for the mass that you're carrying (once in acceleration and once in max speed).
     
    #8
    Kieve likes this.
  9. Arrclyde

    Arrclyde Rear Admiral

    Joined:
    Mar 6, 2015
    Messages:
    238
    Likes Received:
    449
    Well to be fair: speed is limited to a maxed put value. Everybody can reach the limit. But the one reaching it faster is further ahead of the one reaching it later. It is actually pretty simple.

    Faster acceleratoin does not mean lower top speed. It is not like a Tesla dragracing a Lambo where the tesla is capped at 250 km/h and the Lambo can go 350 km/h. The Tesla will be the first on the first 200 meters, but because it reaches topspeed at 250 km/h the Lambo will close in onnthe tesla and surpassing it on the last few meters before the finish line.

    In empyrion there is a speed limit everybody can reach. So it is a non factor, and the only important thing is accelerration. The one reaching the limit quicker will always stay ahead as there is no way for slower accelerating ships to catch up.
     
    #9
  10. zztong

    zztong Rear Admiral

    Joined:
    Apr 12, 2016
    Messages:
    249
    Likes Received:
    269
    And at that point they'll stay the same distance from each other because they're at the same speed.

    Until, as Geo points out, the one with better acceleration turns 45 degrees. Then there's an acceleration difference again. Basically, if you accelerate faster than your opponent, turn when you reach top speed and you can continue to extend your lead.

    Notes:

    Interestingly, this can lead to power problems for ships that don't have enough power plant to afford firing thrusters in multiple directions at once.

    I would say turn 90 degrees, but then the following opponent gets an obvious hypotenuse... err, might get to cut the corner.
     
    #10
  11. Average

    Average Commander

    Joined:
    Jun 22, 2019
    Messages:
    144
    Likes Received:
    121
    This is actually the point where I must not be getting across properly. I've been in this exact situation in PvP on both sides of it quite a few times. Initially this tactic actually causes you to get closer to them, and whether you actually then go on to increase the distance depends on them either not paying attention and just sailing by, or them having acceleration that is so low that taking the inside circuit of the turn doesn't work. Basically if they respond by turning 30 degrees for every 45 you do, they catch up because they can use the 'shorter' triangle to travel at a intersecting vector. If they have MUCH lower acceleration, it works because they can't adjust fast enough. Or if they don't notice your turn and don't adjust, you can increase the gap. Which is why in PvP currently people use this but only do it a few degrees at a time, because it is really hard to spot. Then over a *long* time they get away. But if they pursuer sees what you're doing (because you try to do it too quickly), they take the shorter path and catch up.

    Hey I could be wrong, but open paint and I think it's easy to draw what I mean, a ship fleeing and the vector periodically changing by 45 degrees, and a ship pursuing and changing by 30 degrees. Each line segment the same length for same speed. As long as the responding ships acceleration is not MASSIVELY slower (in which case effectively they keep going close to straight, and after initially catching up they lose ground), they will basically follow an inside curve that intersects with the other curve at a closer point than before.

    That's why I say geo's talking about orders of magnitude difference to make spiralling effective enough.

    So again I actually agree with nerfing thrust. And I like how you've thought through what you're saying. But I think I can list a consequence which will be unexpected for you.... dodging. No more dodging fire from other people's turrets, even in an agility CV (which I use in PvP sometimes btw). Even if we slow turret projectiles down. That's a significant thing to remove from the game. I'm not even sure I disagree with you, but I don't think the unexpected effects thing is greater for thrust over speed.

    The speed thing fixes the boring chase issue if you just make a small change. The acceleration thing needs orders of magnitude change to resolve chases. And the speed thing fits better aesthetically - big stuff should fly slower than small stuff in a sci fi (as opposed to realistic) setting. Lastly, it increases diversification in a sensible way because it adds to small ships late game advantages.

    I'm actually agreeing with you in many ways, and I appreciate that you put more thought into suggestions than most folks on this forum :p Acceleration should get nerfed! But it's better together with the speed change I suggest. Let the ideas team up and it will actually enhance them both! :)
     
    #11
    zztong likes this.
  12. geostar1024

    geostar1024 Rear Admiral

    Joined:
    Jan 24, 2016
    Messages:
    1,483
    Likes Received:
    2,459
    Your analysis is correct for sudden large changes in heading, so incremental changes in heading is definitely what I had in mind.

    However, I should have clarified that both linear and angular acceleration have a role to play in spiraling. As ships grow in characteristic dimension, their moment of inertia grows as that dimension^5 (assuming constant density). Neither RCS nor thruster torque can keep up with that growth rate, so larger ships should have much lower angular acceleration (right now they don't, for various reasons that have nothing to do with the underlying physics, and probably it's hard to notice the effect of rotation rates on spiraling as a result). Plus, right now the moment of inertia isn't even computed properly (it simply assumes that the ship's mass is spread throughout the bounding box instead of computing the contribution of each block relative to the center of mass). So, a large ship chasing a small ship is going to lose it pretty quickly even if both have the same linear acceleration and max speed. And a ship with a thick shell of armor is going to have a harder time of it than a ship with a thin shell of armor or no armor (which gives lightly-armed/civilian craft a fighting chance to escape).

    I didn't mention dodging, it's true, but I guess that's largely because I don't use it in combat. My ships are typically built with relatively little lateral thrust, so I tend to make passes at my targets (turrets on SVs would help with this a lot . . .). It will change combat, to be sure, but then perhaps we can get away from the idea that ships need equal acceleration in all directions.

    What model would you use to set the max speed then?
     
    #12
  13. Average

    Average Commander

    Joined:
    Jun 22, 2019
    Messages:
    144
    Likes Received:
    121
    Ideally giving some freedom for server managers to tweak the values would be good. And minimal warp speed would need to be set proportionally. But as an initial guess I'd say, starting at 150m/s for ships under 500-700 blocks, knock 15-20m/s off the max speed for each time the number of blocks doubles. So if I calculate correct for 15m/s version, a starter CV or minimal corvette might go around 130m/s or 140m/s, a middling destroyer close to maybe 110m/s, a massive warship might go around 80m/s, and a ridiculous behemoth would be more like 60m/s. SVs might need to be modelled a little different.

    Agreed, this would be great. I'd love to see some servers set up with example configurations that could be tested by players, if the more realistic acceleration model described could be consider by the devs. What sort of indication have you got from other players and/or devs about interest in this model?
     
    #13
    Last edited: Nov 16, 2019
  14. geostar1024

    geostar1024 Rear Admiral

    Joined:
    Jan 24, 2016
    Messages:
    1,483
    Likes Received:
    2,459
    Tying this to number of blocks will penalize decoration, both internal and external. It'll probably also drive designs to use the largest possible devices to minimize the effect (which will end up meaning that every combat ship has ridiculous amounts of thrust as a result of using XL thrusters).

    I don't know, maybe it's the physicist in me, but when I see a big ship in sci-fi, my reaction is, "wow, that must be a pain to start and stop" rather than "wow, that must be a slow ship".

    From the devs, nothing. I don't know that I've seen any server owners explicitly express interest in testing such a configuration, either.
     
    #14
  15. Average

    Average Commander

    Joined:
    Jun 22, 2019
    Messages:
    144
    Likes Received:
    121
    True. My intuition is that they would only represent a tiny fraction of blocks overall. I haven't checked though, if I am wrong perhaps they would need to be weighted in the speed calculation to be less (10%?), as they are fairly fragile and probably not going to be effective deployed as speed-neutral shielding unless they are free in that regard. Ideally though, they are overall a very low cost block-wise.

    Have you ever played Children of a Dead Earth? I've been meaning to play that one for a while. It's apparently space combat with realistic orbital mechanics etc.

    Yeah my guess is they avoid commenting on most stuff to avoid getting too entangled. But I get the impression they sometimes will respond to good ideas, you never know! Judging on what I can gather from your description, I hope the acceleration one gets taken up.
     
    #15
  16. Average

    Average Commander

    Joined:
    Jun 22, 2019
    Messages:
    144
    Likes Received:
    121
  17. geostar1024

    geostar1024 Rear Admiral

    Joined:
    Jan 24, 2016
    Messages:
    1,483
    Likes Received:
    2,459
    Right, but remember that exterior decoration also includes hull/armor pieces. Maybe it's not a huge number of blocks in the end, but every block would count under this system.

    Possibly. I'm not sure which way that might push things.

    I have, and did enjoy it. It took a bit of trial and error to adapt to the tactics required, and the opening parts of the campaign were sometimes annoying because of deficiencies in the ships you're locked into using. However, the AI ships are easily outclassed by custom creations once you unlock the customization option (30 km/s railguns and TW-class lasers FTW). The neatest part in some ways is the module design system, as there's a lot of good physics models implemented for all the various subsystems. Also, last I played, the orbit controls were somewhat rudimentary compared to KSP.

    I'm always happy to be surprised, but I've been suggesting it for years now without any success.

    Realistically, balanced masses for all blocks are needed before any of this can proceed, followed by retuned thrust. Only then would it make sense to deploy additional balancing systems like CPU or this system.
     
    #17
  18. Average

    Average Commander

    Joined:
    Jun 22, 2019
    Messages:
    144
    Likes Received:
    121
  19. geostar1024

    geostar1024 Rear Admiral

    Joined:
    Jan 24, 2016
    Messages:
    1,483
    Likes Received:
    2,459
    I really don't know; certainly, they don't seem to intend to budge on the actual architecture of the CPU system no matter how much criticism is heaped upon it. They seem to believe that it's at the stage where all they need to do is tweak a few numbers to make it work right, whereas I don't think their formulation actually does what they want it to do. <shrugs>
     
    #19
  20. Average

    Average Commander

    Joined:
    Jun 22, 2019
    Messages:
    144
    Likes Received:
    121

Share This Page