Limitations curiosity (Full Version)

All Forums >> [New Releases from Matrix Games] >> Command: Modern Operations series



Message

DMAN -> Limitations curiosity (3/21/2019 7:46:47 AM)

Hi all,

I'm curious as to the physical calculation limits that the sim can perform.

Let's say you had a "super computer", with numerous CPUs and about 1TB of RAM (and all the other necessary bits and pieces).

Could the sim run an actual world-wide conflict seamlessly in 1:1 time with hi-fidelity mode on?

Cheers!



SeaQueen -> RE: Limitations curiosity (3/21/2019 11:07:47 AM)

Why would you want to?

There's a lot of stuff Command doesn't do. Ground combat, for example, is pretty basic. Global logistics meh... sorta. You can't do things like build GBUs out of bomb bodies and guidance kits. That limits the time scope of the scenarios. It also doesn't work well over many planning cycles without a lot of very sophisticated LUA coding.

The other problem is that it stops making sense after a certain point. The people worried about things on a global scale aren't the people building strike packages or figuring out what weapons need to be hung on the wings. They're more worried about politics, schmoozing allies, and figuring out how to translate policy goals into things the military can actually do. So... yeah... I suppose in theory you could, but by the time you finished planning one air strike in a realistic manner, and you've still got hundreds to go, and that's just one day, you probably will have found other ways to use your time.



DMAN -> RE: Limitations curiosity (3/23/2019 7:39:55 AM)

So, any technical information or replies to my query?

Hoping the developers can respond, as they'd know best.



kevinkins -> RE: Limitations curiosity (3/23/2019 1:24:45 PM)

I agree with SeaQueen. For one person to manage even a entire carrier strike group is sort of a stretch. But we do so in Command (and other wargames) because it's a fun and interesting exercise. I suppose Command could handle a "worldwide conflict". But it depends on how you define that type of scenario. Pretty sure that there is a technical limit to the number of units the software can "seamlessly" handle on the map. But that number is well beyond what a player would want to have in the scenario to begin with. DMAN, let us know the size of your world wide scenario. For example, some of the LIVE scenarios are very large. Are you thinking of combining 2 or 3 of that size into one?

Kevin



DWReese -> RE: Limitations curiosity (3/23/2019 2:40:12 PM)

I'm jumping on-board with SeaQueen and kevinkins as well.

It appears that you are trying to micromanage the entire thing, which would be a huge feat, but it seems to me that it would take all of the fun out of it. There are things that we don't need to really jump into and still keep it realistic. When a jet is set to take off in two minutes, that is good enough for me. I don't need to know why it was two minutes. I just accept it. Just like if something is trying to defend an attack using AI, you can only do so much with WRA to make it happen. You can decide what range they SAMs should start shooting, but who is to say that they wouldn't actually shoot later, or earlier?

Perhaps you aren't actually expressing what you are trying to do in a way (or scope) that we are understanding. From what you describe, given the actual size, it may take you a week to program everything that would take place in a 10-minute period, just to get it close to being the way that you want.

I doubt that the devs will be able to respond until they know precisely what you are trying to do, and why.

Your question: "Could the sim run an actual world-wide conflict seamlessly in 1:1 time with hi-fidelity mode on?"
, might simply be answered as "yes", but it begs for some kind of follow up as to what your definition of "world-wide conflict seamlessly in 1:1 time" actually is. I suppose that a supercomputer could do all kinds of things, but game devs need to create a game that works on regular home computers (which we all have), as opposed to supercomputers, which none of us have.

So, until then, we are looking for some definition and clarification.

Doug



Zanthra -> RE: Limitations curiosity (3/24/2019 1:36:36 AM)

Technically I don't think there is any special limit on the size or scale of the battle in the engine. On the other hand, there are performance problems that start showing up as the number of active units increases that are at least linear - O(n) or greater. Although improvements have been made recently that help especially if the scenario has a lot of active units not actually being very active, there are still some issues, in particular, unit damage or destruction, and AI unit and mission decision making with large numbers of valid targets come to mind. There are also some UI performance problems, espeically related to the Weapons Release Authorization window which run some long running code on the UI event thread and hangs the program for the duration, and the Aircraft window for ships and airbases which can be quite slow to populate with all the aircraft available, but is pretty fast when it finishes populating.

As for running on a supercomputer. Most supercomputers link many computer together and dispatch work to be done over a longer period of time and then collate the results. The latencies involved are much too high for interactive stuff like video games. Multiprocessor systems will probably lose any performance benefits with loss due to latency in interprocessor communications. Consumer desktop processors have some of the highest clock speeds and lowest memory latencies available, so you are likely to see the best performance from a high end gaming machine for almost any interactive application like this one.



DMAN -> RE: Limitations curiosity (3/25/2019 5:41:34 AM)


quote:

ORIGINAL: Zanthra

Technically I don't think there is any special limit on the size or scale of the battle in the engine. On the other hand, there are performance problems that start showing up as the number of active units increases that are at least linear - O(n) or greater. Although improvements have been made recently that help especially if the scenario has a lot of active units not actually being very active, there are still some issues, in particular, unit damage or destruction, and AI unit and mission decision making with large numbers of valid targets come to mind. There are also some UI performance problems, espeically related to the Weapons Release Authorization window which run some long running code on the UI event thread and hangs the program for the duration, and the Aircraft window for ships and airbases which can be quite slow to populate with all the aircraft available, but is pretty fast when it finishes populating.

As for running on a supercomputer. Most supercomputers link many computer together and dispatch work to be done over a longer period of time and then collate the results. The latencies involved are much too high for interactive stuff like video games. Multiprocessor systems will probably lose any performance benefits with loss due to latency in interprocessor communications. Consumer desktop processors have some of the highest clock speeds and lowest memory latencies available, so you are likely to see the best performance from a high end gaming machine for almost any interactive application like this one.


Thanks to all who answered, but this is the response and information I was seeking.



Anathema -> RE: Limitations curiosity (3/25/2019 11:06:16 AM)

Is CMANO even coded to take advantage of multiple processors? I know when I have used a multiple CPU workstations for work probably 95-99% of desktop software will only ever used one CPU with the other sitting idle because it was never designed to use more than one processor.

As far as I know only very specialised software meant for servers or high end workstations and things like huge databases, 3d rendering or complex simulations etc will ever take advantage of more than one CPU.



DMAN -> RE: Limitations curiosity (3/26/2019 3:18:28 AM)


quote:

ORIGINAL: Anathema

Is CMANO even coded to take advantage of multiple processors? I know when I have used a multiple CPU workstations for work probably 95-99% of desktop software will only ever used one CPU with the other sitting idle because it was never designed to use more than one processor.

As far as I know only very specialised software meant for servers or high end workstations and things like huge databases, 3d rendering or complex simulations etc will ever take advantage of more than one CPU.


Great observation and something I didn't even consider.

Hence why I'd like someone who understands the coding/development of the software, to chip in.



SeaQueen -> RE: Limitations curiosity (3/26/2019 2:17:10 PM)

I can handle a few CSGs, but it's a situation where I probably spend a couple hours each night setting up missions, carefully planning DCA, figuring out where I need to be, how to weaponeer my targets, etc., and then a weekend of actual play. It's not a "pickup" game at that point.

The advent of the 1000-mile ranged cruise missile or TBM, along with globally ranged aircraft and mid-air refueling, makes the distinction between a tactical level wargame and a strategic level wargame blurry. For example, when a B-52 flies halfway around the world, links up with a tanker over the Indian Ocean, and then provides CAS to a platoon of Marines in Afghanistan is that a tactical movement or a strategic movement? In some ways, the distinction has to do with intent (or maybe the effects regardless of intent?), and sometimes the intent might be purely tactical (e.g. suppressing SAMs), and others there might be many layers to the onion (reassuring allies, deterring aggression, shaping other nation's policy decisions), which Command doesn't really deal with.

Even so in the contemporary era, it's possible to have a purely tactical game, which has aircraft and ships moving very long distances. I think that's particularly the case in theaters like the Western Pacific where there's just not a lot of places to land. In that case, a lot of the game can end up being not just about shooting missiles and dropping bombs, but also making sure they have the gas so that they can Winchester out before they Bingo out. If that's what is meant by a "global" game, Command actually handles that kind of thing fairly well, although there's a bit of an art to it.

quote:

ORIGINAL: kevinkins
I agree with SeaQueen. For one person to manage even a entire carrier strike group is sort of a stretch. But we do so in Command (and other wargames) because it's a fun and interesting exercise. I suppose Command could handle a "worldwide conflict". But it depends on how you define that type of scenario. Pretty sure that there is a technical limit to the number of units the software can "seamlessly" handle on the map. But that number is well beyond what a player would want to have in the scenario to begin with. DMAN, let us know the size of your world wide scenario. For example, some of the LIVE scenarios are very large. Are you thinking of combining 2 or 3 of that size into one?

Kevin




HalfLifeExpert -> RE: Limitations curiosity (3/27/2019 4:21:04 AM)

I think the only way that one could simulate a 'world wide' conflict in something like a CMANO environment would require a professional LAN network with the core system operating off of a large, powerful, and well maintained supercomputer server, with the different operators interacting with forces on individual computers connected to the network.



DMAN -> RE: Limitations curiosity (3/27/2019 9:53:45 AM)

Interesting takes on it all. I like big.

I don't at all play small scenarios. Just not my thing.

SeaQueen alluded to a theatre-wide or hemispherical conflict. That would be nice. [8D]

Would multiprocessors and a tonne of RAM be able to handle the WestPac theatre seamlessy in 1:1 hi-fi mode?

And which IS the more important: processing power or RAM? Any hardware experts care to add? [:)]



SeaQueen -> RE: Limitations curiosity (3/27/2019 1:27:41 PM)

quote:

ORIGINAL: DMAN
SeaQueen alluded to a theatre-wide or hemispherical conflict. That would be nice. [8D]


Most American wars are theatre-wide or hemispherical. We're a global power. One of the challenges for the US is that we have to schlep all our stuff all over the world, most other nations are fighting in their neighborhood. In parts of the world where there's lots of bases that's not so much of an issue because you can preposition stuff. In other parts of the world the options are much more constrained either due to politics or geography.

In Command, though, you'd need to do a lot of very sophisticated LUA coding to move beyond the 12 to 24 hour timeframe, because you run into a problem with the AI and aircraft in particular (on the naval end things unfold more slowly, so it's less of an issue). After they've flown their missions, what do you do next? Do you reapportion them and rearm to strike new things and patrol different areas? Which areas do you patrol? How do you pick which weapons to strike which targets with? What supporting assets do they need? There's a lot of questions that need to be answered. Multiplayer would also potentially make it easier to do things beyond the 12-24 hour timeframe.

I think it's all feasible, but it'd be a fairly big investment of effort. Even then, though, I don't see doing entire conflicts in Command. I just see maybe longer duration missions over a wider geographic scope. Once you get beyond a certain level of warfare (and it's a continuous scale, so there's not a hard-and-fast cutoff) it isn't really about striking aimpoints at all (the subject of Command). It's about things like managing alliances, strategic messaging, logistics, and all the underlying stuff.

If you wanted to make a model of the entire Balkans Wars in Command, for example. To really do the WHOLE conflict, you'd be faced with a political/policy goal. In that case, "Stop the genocide in the Balkans." NATO does not have a "genocide stopping bomb." So how to stop it by military means isn't obvious. All that NATO can do is attack some of the obvious tools of genocide, in that case, they went after artillery sites as well as the military leadership. They'd also strike bridges with the idea that they could keep the parties separated and block lines of advance. Politically, they were being asked by the press about how many tanks they were killing even though tanks weren't really contributing to the problem, but they had to do it because they needed the electorate to feel like they were making progress. The thing is Command doesn't handle any of those decisions. Command is more down in the weeds. It's at the level of "Destroy the FRY XXXth MRB HQ located at Sarajevo." Then you weaponeer your aimpoints, develop your plans for all the supporting aircraft, ships and submarines" then go fly it over a day or so. Then you make a hole in the air defenses, fly your bomb trucks and cruise missiles in, do as much damage as you can and go home. It's a mistake to try to do the entire conflict in Command, because Command is about striking targets and that's only a portion (sometimes not even the most important portion) of warfare. Unless you script in what the political impact is (which is really just an expression of the scenario author's opinion, not a game because reasonable people might disagree) the game is completely agnostic what it is.



Tailhook -> RE: Limitations curiosity (3/27/2019 4:06:18 PM)

I think at that point you’re going to run into a lot more issues with the game itself. The AI is only as reactive/proactive as the scenario designer can make them ahead of time. It’s very difficult to predict what is going to happen 2-3+ days into a scenario, so there’s a certain point where it would just effectively stop working.



Page: [1]

Valid CSS!




Forum Software © ASPPlayground.NET Advanced Edition 2.4.5 ANSI
7.015625