INFO & FEEDBACK [Alpha 11] CPU Points and Tiers - How does it work?

I went past cpu limit on my SV and still flies just fine. No hard cap there what so ever.

 

Sigh. No, you went past the starting penalty threshold. The effectiveness decay is linear. Originally a very soft +100% = 90% effectiveness, currently a harsher +100% = 50% effectiveness, and with a proposed +100% = 20% effectiveness.

All of these have hard caps. Linear decay is like that. It's just one needs to apply primary school maths to figure out where those caps are since they aren't being disclosed directly. The current hard cap is at +200% CPU points or 3 times the listed threshold (i.e. if the HV tier-1 cap is still 5,000 then affected devices will not operate on an tier-1 HV worth 15,000 points or more). The proposed cap is at +125% CPU points or at 225% of the CPU point threshold so no device on the same HV would operate at 11,250 or more points.

 

A soft cap is the point where you receive diminishing returns. One example is weight (& more pointedly, weight from thrusters). As you add more thrusters, the propulsion provided by each is reduced. In this way, 100% CPU usage is a soft cap. Anything beyond that gives a penalty.

A hard cap presents an absolute limit- a point that's impossible to work past. Equipment limits, M/V storage limits, & optional size class limitations imposed by the sever are all hard caps. 225% CPU usage is a hard cap. Vessels stop operating beyond that point.

Since the efficiency penalty progresses at a linear rate, it has a hard cap by definition- a point where efficiency reaches 0%. The question was whether CPU represents a hard cap. As it's both a soft cap & a hard cap (at 100% & 225% CPU respectively), the answer is yes, it represents a hard cap.

 

All things in life has a hard cap. When your own use by date roles around you have reached your hard cap.
Here is something that might help you to understand.
You place bread in a toaster. you cook it to 100% perfection. Now if you keep toasting it you get diminished returns. It starts to burn so you quickly pull it out and scrape off the little bit of burnt bits because your hungry. But if you kept toasting it it even gets to the point until there is no way you could even scrape the burnt bits off. Keep toasting it and it gets even worse until it's just ash. Now you have hit the hard cap.

I think right now this this topic is past it's 100% and starting on it's diminished returns. Soon it might hit it's hard cap and sound like trolling

 

Although I understand the sentiment, it's not true. Many math functions have no hard cap. They are asymptotic to a limit, but never actually reach it. Much like some games' soft caps, they are soft capped in that there is a point where further extrapolation offers less benefit than cost to the viewer though you could if you really wanted to.

 

To be honest I think after the last CPU limit increase it's not a real issue anymore but I guess we could call it a hard cap because if you continue to increase the amount of objects your CPU has to handle then sooner or later your ship turns into a stationary base with thrusters.
Am I correct in assuming you want the soft cap to be a little "softer" ? Maybe add a little complexity to the CPU efficiency ? For example ingame maybe you're not always hit with the efficiency penalty right away. Maybe after a couple of game hours of flying your CV the efficiency penalty kicks in or maybe the efficiency penalty kicks in after you hit your CV's top speed. That is actually how something like this usually happens in the real world. It's more than size and workload that can cause lack in efficiency in a machine, it's how long it's been running without maintenance or how long the machine was pushed past it's limits. Even in the PC world people use to overclock their CPUs all the time and sometimes their machines would run just fine for a pretty good while before seeing any drop in efficiency.

 

They were warned in advance of the temporary 'mercy' in Penalty Strength fair & square, they shoulda heeded... or alternatively they don't come to these Forums at all & they've thus chosen the Path of Ignorance... suit themselves.

Really is quite Toxic...

 

I just want to understand. Since there is a hard cap confirmed, I understand better than reading Hummel's original description. One consequence is you should never bother with a lower extender type as backup. The construction will be over the hard limit of the next lower extender in almost all cases at the more draconian proposed limit. Even the +100%=50% efficiency level makes most of the immediately lower extenders worthless. If you drop a tier, affected devices will simply be 0% effective.

I think knowing how the penalties accrue is important to properly assess the system and vital to designing constructions especially since in my basic playstyle, I doubt I'll ever manage to acquire 4 tier-4 extenders or even 2 tier-3 extenders. I'm not good enough to raid much and I don't think I've ever hit over 1,000 credits in cash; I can't imagine the game holding my attention as I try to farm 5,000,000 or more credits for every extended construction I want to make. So I've been idly seeing what I could put together in tiers 1 and 2 since that's all I'll have access to should I activate the feature.

All four construction types are concerning to me since I have a negative experience already with the tier-1 limits on HVs and that was when the penalties were quite weak. I am especially concerned with lower level CVs as bases can be balkanized and the two smaller vehicles are much less resource hungry.

 

@Hummel-o-War
I know this question has been asked a lot, but I still haven't seen a clear answer from any of the developers, just lots of guesses from forum members.

What specific gameplay issue is the CPU system attempting to address?

Forcing specialization to discourage jack of all trades ships?

Limiting ship size for MP performance reasons?

Slowing game progression so it takes longer to get good ships?

Some combination of the above? Something else entirely?

Without knowing the fundamental goal of the CPU system it's difficult to give feedback or know if it's working as intended from a gameplay perspective.

 

Why tiers rather than a modular system like we have for storage containers?

 

Then your implementation is poor. You do have a hard cap; it isn't too far ahead of the penalty start point in the case of the lower tiers and your proposal will bring it much closer. Saying otherwise is just wrong. Whether that is caused by innumeracy or simply being disingenuous is for others to determine.

 

The fact that there is a hard limit currently at 3 times the penalty threshold and 2.25 times the threshold at your proposed penalty rate? Linear decay will always impose a hard limit. Constantly saying the limit is "soft" when in reality it is a linear decay at a threshold and not disclosing the actual hard stop point is disingenuous at best.

There are a lot of functions that produce asymptotic curves that fit your disclosed requirements far better. A simple one that meets your needs would be effectiveness = 1 - Max(0, tanh( spent_cpu - threshold_points ) * scaling_factor).

If you want a smoother curve with less of an elbow there's the ever-popular sigmoid curve that is more computationally expensive (which shouldn't matter much -- you are only calculating effectiveness when block count changes, right?) that could be bashed into shape.

And there's the trivial effectiveness = 1 / max( 1, scaling_factor * (spent_points - threshold_points)) that could be used.
Or maybe 1 / max ( 1, spent_points / threshold_points) would give a better curve for your needs?

THERE. ARE. SO. MANY. CHOICES.

Claiming "math is the problem/reason for the hard limit" is just..... wrong.

 

Thanks, somehow I missed that first paragraph description (unless it was added later).

My feedback then is why introduce an entirely new CPU system to encourage specialization and limit certain parts rather than a block/mass/weight/power/damage/armor rebalance? The way CPU penalties are applied sounds suspiciously like power with the efficiency dropoff.

In terms of jack of all trades ships, IMHO the reason they're so common is that all the weapon/constuctor/storage blocks are so small that it's trivially easy to add a few of each to anything. If you really want to encourage specialization, add some truly large guns/refineries/constructors and some improved automatic mining/crafting (and remove the magic blueprint factory infinite storage) to provide a need for dedicated freighters capable of hauling massive amounts of resources.

In terms of gameplay, I do exclusively SP and what this limit will do more anything else is make me build smaller, not necessarily more specialized due to the difficulty of managing multiple ships. Especially with regards to CVs. The weights and volumes and physical block dimensions for mounting weapons already forced some specialization in HV/SV designs (miners, haulers, POI busters), and other than adding extra grind for the cpu components I don't see it really changing these vehicles as they're already size limited by hangar doors in my usage.

 

I don't particularly care for math, but it's easy enough to illustrate this point.

The point is basically "fix your math" and it won't be a 'hard cap' anymore.
(The second equation is of the form y=1/x)

 

The second method would cause a very quick downturn of performance the moment you step 1% over the limit. Which is unlike a CPU.

If this is an acceptable side effect, then second method would be good to implement from a gameplay standpoint.

 

Isn't that exactly what i wrote a couple of weeks ago? ;-)

Seriously, i would love such a system way better than the current one. Because it is simple, intuitive and has no hard cap by reaching a point where adding more CPU comes with a very high cost (energy, volume, weight). And it is up to every player to determine if its worth the cost.

Edit:
Another plus for modular system is => who would have guessed, modularity.
To make the current system work it would mean that no additional devices com into the game, otherwise they have to change limits and tiers again.
With a modular system that is dynamically, you don't need to change anything. New devises will fit in and need additional cpu modules, making the vessel grow. The only thing that you have to adjust: CPU modules in CPU give and Energy taken. Pretty easy to balance.

 

That's just a crude example, and while I don't know what the actual equation for that particular curve is, I know it's not y=1/x specifically. There's a fair amount of adjusting and number fiddling that could be done to obtain a less brutal decline and longer curve, but - I hate math.

 

I hate math too.

I know! Lets ask Mikey! er... geostar!