A 3477 radius planet takes some 50GB of RAM

not necessarily. let's just let the devs answer this one.

If the only reason for high ram usage would be "bad programming" then programmers in general seem to be getting worse and worse.

... or maybe they're just using the growing resources of computers to solve bigger and bigger problems.

Surface Area =4π r^2

1300 =0.21e8
3477 = 1.52e8

thats 7 times more surface area.

8gig x 7 =56

I don't see how this is surprising.

I don't think the surprise is that it scales quadratically, but rather that you used to be able to create 4500 radius planets with just 14 GB of RAM. See here.

@tatsujb is probably right in the assumption that the planets have become a lot more complex.

The maps were bigger in Total Annihilation and that game came out 18 years ago and was made by the same people.

A 3D spherical map with complex terrain is a much bigger problem than a 2D height map, regardless of the size.

If they made the game run better but then theres all that other broken things yet to be fixed, so we're not going to see anything of this till someone emulates 64 bit games in the future.

Save it for the judge.

so by your logic drawing a picture and forming a sculpture requires the same kind of work?

The terrain in TA was sculpted, it was 3 dimensional.

I'm just saying, that with all of the advancement in gaming over the last 18 years, we should probably be having games with much larger numbers of players by now. Who wants to play a 50 player FFA? Me. Now lets make it happen.

It's a design choice, not a technical limitation.

PA probably could easily handle 50 player FFA if the graphics were terrible and a whole 3D solar system of moving objects did not have to be tracked and rendered at very far and near zoom.

Considering there is pretty much no game that allows for 50 player ffa large scale RTS I guess the technical reality isn't as pretty as you (and me) want it to be.

It's a simple fact. You cannot represent an arbitrary PA map as simply as you can a TA map. Not because TA was better, or because PA has gone backwards or isn't programmed as well; it's a basic consequence of the requirements of true 3D vs height-mapped 2D terrain.

No matter how "complex" the map was it TA, or whatever tool the creators use to form the terrain, it could always be represented by a single 2D grid of height values, and a single layer of pathing. The same complexity could easily be achieved and exceeded (especially with multi-level pathing) in a far smaller map in PA; even a single terrain feature could be as "complex" computationally as an entire TA map, and that's before you even consider the performance difference of a 2D bitmap background vs individually textured 3D objects (often with multiple texture layers). The fixed camera angle also allowed a lot of cheats that simply can't be done with a rotatable camera.

Whether all that performance overhead is worth the jump from fixed camera-angle 2D height maps to full 3D worlds is of course, subjective. The former allows for far smaller system requirements, allowing far bigger battles. A TA sequel or Sup Com/PA type game set on a map system like TA could be quite interesting indeed. Sometimes less is more, in that sense.

I don't imagine there too much (relative) performance difference between 50 players with 100 units each and 5 players with 1000. Rather than performance, I imagine low player caps are largely caused by two factors -

• Network performance
• Gameplay worth / market interest

50 players sounds cool and all, but how many people get frustrated just trying to form a balanced, enjoyable small player count game? Even with a healthy population, you can expect increasing frustration with increased player count due to dropouts etc. There's also the network side of things; P2P will likely never achieve large players counts, but client/server has the burden of needing an enourmous and fast pipe to work. There have been large player games of PA in the past - it would be interesting to hear one of the players' takes on whether the experience was worth it.

To be honest, using position and orientation in 3 dimensional space was one of the best things Uber did. I mean, every time I look at those old RTSes with their ugly checkerboard bases I feel frustrated why we couldn't have this back in the time of C&C 1 and Warcraft 2.
But this all comes with a cost. Doing pathfinding within a continuous space, even if the space is flat (without bridges or tunnels, just a plain 2.5D height map) is several times harder then doing the same thing on a checkerboard, where a building occupies a cell either completely or not at all.
Doing pathfinding on a 2-manifold (that is what you get if you take a flat 2-dimensional space and add bridges or tunnels to it) is even harder. That is the sole reason why PA cheats here and units get stuck sometimes.
By the way, if someone really wants to find out what this gibberish I'm talking about means, there is a good book for planning algorithms out there and - the best thing - it is free. Conveniently its called "Planning Algorithms". (For manifolds check chapter 4. Chapter 5 and 6 will give you a glimpse of the problems you encounter when doing motion planning within a continuous state space.)

There is another reason, why large games are not desirable from game design:
The overall win-rate decreases with 1/n for n players per game. And we all know that people like to win, basically they play for that one reason.
If you only allow 1v1 in your game, it means that for every lost game there is one game that is won. Which is great, that's 50%. If you only have games with 20 players FFAs, 19 people will lose, just for one person that wins. Statistically you will suffer 19 loses for one win. I'm not sure if people want something like this all the time.

Oh come on. Do you really think people don't like a challenge? I find 10 player FFA boring because they are too easy to win! I want something harder, something that takes more than a bit of luck to win. It's a sad state of affairs when people are arguing that a game such as this, which revolves around being epic, should limit the number of players as a choice. I would gladly sacrifice the ability to have multiple planets and the ability to have 100s of units in favour of having more players.

Give us a shout when you find the other 49 willing to play with you

If the game were better it wouldn't be a problem

I haven't seen anyone argue for artificially limiting the number of players. I've only pointed out the difficulties involved and why they may not be a focus for developers due to the general interest in such games not being as high as lower player counts, for a variety of reasons. That's doesn't mean I'm pro-limitation.

You kind of make a case-in-point about player counts when talking about sacrificing other gameplay elements though. Increased player count support (from a technical perspective) often comes at the cost of something else. All developers are going to ask the question "is this a worthwhile tradeoff for my target audience?". The simple fact is, even in games that support them, large-player games represent a tiny fraction of the games played, and thus are not typically a worthwhile target. Is it better for a developer to focus on improving the game for 90% of players to the detriment of 10%, or vice versa?