Limit Theory Forums

So I was reading the thread about AI recently, and found a post from josh saying that 100s of units will be the expected maximum during battle due to CPU constraints. Is this limitation to all ships in general, or just total number of ships in combat at any time? If there'll only be a few hundred ships in the entire galaxy the entire galaxy is going to feel empty. I hope I'm misunderstanding what's been said.

Thanks for your useless and time wasting reply. That's the thread I just read, and was referring to.

JoshParnell wrote:My target has always been ~100 ships in combat for a "large" battle.

obviously you did not read closely.

the limitation is only intended for battles, not for the whole universe

Edit: when you already knew the thread which is abou ship count, why are you creating a new thread? It just spreads the information around different threads and makes it harder for others to find the informations they search.

Necro posting is encouraged in this forums

Cornflakes_91 wrote:

JoshParnell wrote:My target has always been ~100 ships in combat for a "large" battle.

obviously you did not read closely.

the limitation is only intended for battles, not for the whole universe

Yes, but what about the people in the system during the time of the battle? Aren't they going to take up space? My guess is that the game will be able to support up to 100 comfortably and then raw CPU-horsepower will make up the difference.

Maddix wrote:

Cornflakes_91 wrote:

JoshParnell wrote:My target has always been ~100 ships in combat for a "large" battle.

obviously you did not read closely.

the limitation is only intended for battles, not for the whole universe

Yes, but what about the people in the system during the time of the battle? Aren't they going to take up space? My guess is that the game will be able to support up to 100 comfortably and then raw CPU-horsepower will make up the difference.

Its completely raw CPU power, if your rig does not get to 100 ships it does not get to 100 ships.

besides that its not directly about the ship count per-se but the amount of independent physics objects (bullets, missles) that they create while fighting.

Cornflakes_91 wrote:

JoshParnell wrote:My target has always been ~100 ships in combat for a "large" battle.

obviously you did not read closely.

the limitation is only intended for battles, not for the whole universe

Edit: when you already knew the thread which is abou ship count, why are you creating a new thread? It just spreads the information around different threads and makes it harder for others to find the informations they search.

Necro posting is encouraged in this forums

While he did specify that 100 ships will likely be the number for battles due to CPU, he didn't ever say what the number will be in total, in and around sectors. And the main reason I posted this in another thread is that it felt important enough to warrant a second thread - if it was the case that only ships in the hundreds would be around, I wouldn't bother buying this game, and I'm sure others wouldn't also. I figured I'd ask in a thread where that info wouldn't get burried.

EDIT: for clarity

I'm with mogthew on this. I'd like to know how many ships we can expect per system, not just per battle. I feel this warrants its own thread if another one like it doesn't exist. The thread Cornflakes linked is specific to battles and not ship count in general. It'd be far better to use a new thread like this called "Ship count" since that will make it easier for people to find this information in the future with the search function. Welp, back to trying to get to sleep.

Having just played with the Starswarm Mantle stress tester, I can certainly say that when it comes to ship count (or performance in general), an optimised Mantle implementation is the way to go.
I'm hoping Mantle will appeal to Josh's keenly developed sense of optimisation and efficiency... one day

u3z05en wrote:Having just played with the Starswarm Mantle stress tester, I can certainly say that when it comes to ship count (or performance in general), an optimised Mantle implementation is the way to go.
I'm hoping Mantle will appeal to Josh's keenly developed sense of optimisation and efficiency... one day

Um what?

First and foremost. Mantle is a GPU system. In the video introducing the benchmark some time ago they showed you can get similar FPS with DX11 settings turned way down.

Second. There is a MASSIVE difference between the two games. LT also has to handle OOT AI for an entire game universe. And the combat AI for LT is likely going to be far better per unit than that of Star Swarm.

Third. To be frank none of us know how LT is coded and Mantle may not be the way to go. People were thinking Mantle was going to save BF4 and benchmarks show in my opinion unimpressive gains.

u3z05en wrote:Having just played with the Starswarm Mantle stress tester, I can certainly say that when it comes to ship count (or performance in general), an optimised Mantle implementation is the way to go.
I'm hoping Mantle will appeal to Josh's keenly developed sense of optimisation and efficiency... one day

Mantle very well might make it's appearance eventually, but for right now, Josh's main development machine is Linux (due to it's stricter adherence to coding standards) and I don't believe Mantle is cross platform yet. This is why OpenGL is the focus.

At the same time, recent developments might suggest Mantle isn't doing as much for GPUs as originally thought.

Despite all that, I think we'll start seeing more in regards to how much the engine can handle once we get a few more features.

Josh, can we get a comment? Be great to know what you expect in terms of numbers. Even fakes per system? Anything, just don't tell me you're planning on only having a couple of hundred ships in the galaxy at all times! :<

It's honestly too early to tell what kind of numbers we'll be seeing, but my goal has always been "1000+ active NPCs." I would be very happy if that number could be pushed to 10,000, but again, it's too early to tell, as I haven't started mass-scale AI optimizations yet (that won't start until the AI behavior is fairly well locked down). As someone already stated, 100 active ships in battle was my goal in terms of individual battle scenarios

Also: Mantle really isn't as relevant in a situation like LT, where we're going to hit a CPU bottleneck long before the GPU taps out. The difficulty will be scaling up the AI and the physics, not the graphics. Mantle cannot help with that part

EDIT : Keep in mind that 1000 "active" NPCs does not mean there are only 1000 NPCs. It just means that you're only seeing "activity" from 1000 NPCs. Regions will be paged in and out as you move about the universe, so it's still entirely possible for you to meet more than 1000 NPCs (or whatever number ends up being the limit of the NPC working set).

Another thing - this number will be very easy to expose so that people with more powerful PCs can try to push the limit

JoshParnell wrote:Also: Mantle really isn't as relevant in a situation like LT, where we're going to hit a CPU bottleneck long before the GPU taps out. The difficulty will be scaling up the AI and the physics, not the graphics. Mantle cannot help with that part

I am not a specialist at all, but from what I've read these last days about Mantle, it is especially efficient on performances in case of CPU limitations. http://www.extremetech.com/gaming/17567 ... s-not-gpus
Don't know if it is relevant for LT but I just wanted to share it.

I think the problem is that Mantle doesn't do AI, it does the graphics engine. Josh is offloading the graphics onto the GPU already, so there's no real problem where's that concerned.