Screenshots from tutorial

[LJFHutch]

Oh wow, thank you so much Wilhelm! I had completely forgotten about that program, I used to use it for Live For Speed, CMBN feels so fresh now!

How do you export a profile, I found a good one and was going to post it up but they don't seem to be stored in the .ini.

[DaveDash]

Back on topic, yes that gapa tool is actually very cool.

[Guest]

Yeah, no need to argue, folks. Different hardware will play the game differently.

Using ArmA II as comparison, which runs smooth on pretty much max, and is also quite a CPU intensive game:

How many troops are in that CMBN screenshot? Put that many troops in view in ArmA II, even from a distance, and see what happens to frame rates. I would venture that the frame rates would suck, since the few ArmA II missions I played with anywhere near that many troops (especially on one screen) were on occasion slideshows.

That's not to say that CM doesn't need (and won't get) work to improve performance. But it's not a straight comparison by any means. Also, with the way that ArmA works (and quite a lot of other FPS-oriented games work), BIA can make some assumptions about camera placement and other things that we can't. I've definitely noticed dips in frame rates when up in helicopters and the like in ArmA, and it's largely because those assumptions they can normally make are no longer valid.

[Erwin]

I downloaded gapa, but am a bit unclear on where to put it, and how to use. Scared to mess up something else in my pristine system. Any reassurance, hints etc?

[para]

I downloaded gapa, but am a bit unclear on where to put it, and how to use. Scared to mess up something else in my pristine system. Any reassurance, hints etc?

What is Gapa Erwin?

[DaveDash]

Yeah, no need to argue, folks. Different hardware will play the game differently.

How many troops are in that CMBN screenshot? Put that many troops in view in ArmA II, even from a distance, and see what happens to frame rates. I would venture that the frame rates would suck, since the few ArmA II missions I played with anywhere near that many troops (especially on one screen) were on occasion slideshows.

That's not to say that CM doesn't need (and won't get) work to improve performance. But it's not a straight comparison by any means. Also, with the way that ArmA works (and quite a lot of other FPS-oriented games work), BIA can make some assumptions about camera placement and other things that we can't. I've definitely noticed dips in frame rates when up in helicopters and the like in ArmA, and it's largely because those assumptions they can normally make are no longer valid.

I knew you'd bring that up. I've had sometimes hundreds of troops, tanks, vehicles in map at once, flying around, on the ground, and on this computer it runs like a charm. Not only that, I ran various mods which are running AI scripts for each unit in the game.

I used ArmA 2 because it is the closest game I can think of to CMx2 in terms of demand on hardware. It's also not strictly FPS (in command view and with various mods) and can be a hybrid RTS/FPS with no noticeable slowdown. Obviously yes, camera angles can be assumed and the game isn't strictly the same, CMx2 has complex LOS calculations and so forth and pushes the boundaries in many ways that other games don't. I understand and appreciate that.

The point is, there is something very noticeable about the combat mission CMx2 engine that is unoptimized. I am not complaining, but I have spent a lot of money on this computer and it is the only disappointment with the game. It's the only game on my machine that does not run in an excellent manner, despite what hardware I throw at it.

HAVING SAID THAT, I applaud the effort that has been made to make combat mission what it is, given the resources that BFC have. The game is still perfectly playable at 20-40 FPS, and given Steve's benchmark of 20-30 FPS then I guess the 80/20 rule probably applies to squeezing out that extra optimisation.

[LJFHutch]

ArmA handles a lot of troops quite well:

I'm pretty sure that was a pretty hefty overclock but I've personally had battles with 800 or so AI at frame rates comparable to what I get in CMBN and I have a cheap AMD processor from 2007 or something. However, you try to run even a few hundred troops in a city ... BOOM! The micro AI will tear your processor apart

[Guest]

I knew you'd bring that up. I've had sometimes hundreds of troops, tanks, vehicles in map at once, flying around, on the ground, and on this computer it runs like a charm. Not only that, I ran various mods which are running AI scripts for each unit in the game.

Hmm, strange. ArmA 2 ran like hell for me on a computer that actually plays CM quite well. I was supposed to be doing a review of ArmA 2 for Armchair General and couldn't finish it, the lag was so bad. Whenever more than 15-20 men were on screen at once the framerate dropped into the toilet. And yet I can play 1-2 battalion sized CM battles on the same machine. Go figure.

I used ArmA 2 because it is the closest game I can think of to CMx2 in terms of demand on hardware. It's also not strictly FPS (in command view and with various mods) and can be a hybrid RTS/FPS with no noticeable slowdown. Obviously yes, camera angles can be assumed and the game isn't strictly the same, CMx2 has complex LOS calculations and so forth and pushes the boundaries in many ways that other games don't. I understand and appreciate that.

"Camera angles can be assumed" is HUGE. It's not a "yes, but" thing at all. Nor are complex LOS calculations. You're talking about two ENTIRELY different calculation and rendering loads. And I mean different, as in not even remotely the same.

ArmA 2, in terms of the way the engine works, is definitely a strictly FPS thing. The hybrid RTS portions that I played used the same restricted camera.

The point is, there is something very noticeable about the combat mission CMx2 engine that is unoptimized. I am not complaining, but I have spent a lot of money on this computer and it is the only disappointment with the game. It's the only game on my machine that does not run in an excellent manner, despite what hardware I throw at it.

HAVING SAID THAT, I applaud the effort that has been made to make combat mission what it is, given the resources that BFC have. The game is still perfectly playable at 20-40 FPS, and given Steve's benchmark of 20-30 FPS then I guess the 80/20 rule probably applies to squeezing out that extra optimisation.

Eh, like I said, we do need to work on getting more performance out of the engine. We actively *want* that. Arma's never going to be a fair comparison for a number of reasons, though. Just like Total War isn't because they can do a couple dozen tricks that we can't, by dint of CM's structure. If by "unoptimized" you mean "doesn't use the same tricks as ArmA" then, yes. But we can't. So your point, while the general gist of it is something I agree with, is based on an incorrect assumption.

It doesn't matter either way, honestly. We'll keep improving performance, you'll keep getting a better ROI from CM on your fantastic gaming machine, and we'll ALL be happier.

[Erwin]

para: see post a couple pages ago on Gapa Gamma.

I was hoping for some advice on installing to improve CMBN performance before I go screw up my machine.

[Rocky Balboa]

@Erwin,

Its not going to improve your performance but the tool is supposed to improve your visuals.

.... Of course your mileage my vary .....

[AstroCat]

I just wanted to quickly throw my voice in here, the performance even on a super clean, perfect shape high end PC is terrible with the CMxx engine. I've tested this like others extensively and there is no way around it. The engine for this game is way below average, and I am a huge supporter of the series and Battlefront group as a whole. I noticed this the day CMSF first arrived at my house long ago.

I really wish they could "fix" their engine to get the kind of performance out of it that is expected by many of us. I will just throw this out there, would they have been way better off supporting Direct X, even version 9?

[poesel]

No, they wouldn't be better off. If they would support DirectX they would at least loose one customer: me. Why? DirectX is not cross platform so no CM for the Mac.

[AstroCat]

I know that native OSX support would be lost with DirectX but I wonder what the OSX user base is vs. base gained by a better and more modern looking and performing game engine? Macs can dual boot, its the OS that's the issue, not the hardware, and/or the game could have been ported later to regain some of that lost market share.

[Wilhelm Heidkamp]

Well, returning to more funny/productive things/thread...

Shermans at big resolution (1920*1200):

[poesel]

I'm no expert on that matter but AFAIK OpenGl (which CM uses) is as capable as DirectX - at least for the things used here. ID (of Quake and Doom fame) use OpenGl and they are not known for bad visuals.

Dual booting? Nope, wouldn't have bought the game. And porting later? Later when you have time? Time you could spend on creating new content? Great business plan.

[AstroCat]

I know Rage used OpenGL and a few others, L4D 1 and 2 I believe, I wonder how the id Tech 5 engine would work for a game like CMxx, my gut is telling me might not be the best match, but that would take a lot more research. I would explore using a 3rd party for the port if the numbers made sense. OSX(and previous versions) have never had a relatively strong gaming market. Slower hardware options and non-existent or outdated drivers being one of the issues.

My overall point being that I believe you could get a lot better performance out of the terrain/model/texture rendering than the current engine is doing. I know the 580gtx I have is not being used in an efficient manner with the CMxx engine.

[Erwin]

Rocky: Yes, tech performance is not my problem, but anything that improves the visual "performance"...

But, like I said I would like to see a clearer explanation of where to install it etc, (how to uninstall if needed) and assurance that it won't do anything nasty to my clean system.

[Guest]

I know Rage used OpenGL and a few others, L4D 1 and 2 I believe, I wonder how the id Tech 5 engine would work for a game like CMxx, my gut is telling me might not be the best match, but that would take a lot more research

No FPS-optimized engine can do what the CM engine does. I can put 8-900 fully animated, AI-bearing troops into a field and rush them over a large, well-terrained map at a line of opposing machine guns supported by mortars. I can watch them all run (at once) through the hellstorm. And it'll be playable. I know, because a scenario just like that was one I used to do benchmarking while the Mac version was in process. Do that with id Tech 5 and see what the frame rate is. I'm guessing it'd be "seconds per frame" rather than the other way round.

My overall point being that I believe you could get a lot better performance out of the terrain/model/texture rendering than the current engine is doing.

Not with an FPS engine. Not with an RTS engine. Both can make optimizations that CM's engine can't. Cannot. Full stop. You talk about CM's engine being "way below average", but there are no engines that I'm aware of in the world that do what CM's does. Unlimited troops, unlimited camera freedom, relatively detailed (as in, play-changing detail) terrain.

I play plenty of current games. I've talked to plenty of other game developers (including folks who work on games very similar to ours) and observed what their engines do. And nobody does what we do, for better or worse. You can't be below average in a field of one. (If you can offer a counter-example, I'd be willing to bet I've a) played it, and have a good idea of how it works, before we get into tit-for-tat.)

I'm glad you guys like the game enough to keep asking these questions, but other engines (or DirectX, goodness) aren't the answer. Your best bet is for us to do what we do, and make performance improvements. Like I said earlier, this is something we want to do. You can count on it happening over time.

[para]

No FPS-optimized engine can do what the CM engine does. I can put 8-900 fully animated, AI-bearing troops into a field and rush them over a large, well-terrained map at a line of opposing machine guns supported by mortars. I can watch them all run (at once) through the hellstorm. And it'll be playable. I know, because a scenario just like that was one I used to do benchmarking while the Mac version was in process. Do that with id Tech 5 and see what the frame rate is. I'm guessing it'd be "seconds per frame" rather than the other way round.

Not with an FPS engine. Not with an RTS engine. Both can make optimizations that CM's engine can't. Cannot. Full stop. You talk about CM's engine being "way below average", but there are no engines that I'm aware of in the world that do what CM's does. Unlimited troops, unlimited camera freedom, relatively detailed (as in, play-changing detail) terrain.

I play plenty of current games. I've talked to plenty of other game developers (including folks who work on games very similar to ours) and observed what their engines do. And nobody does what we do, for better or worse. You can't be below average in a field of one. (If you can offer a counter-example, I'd be willing to bet I've a) played it, and have a good idea of how it works, before we get into tit-for-tat.)

I'm glad you guys like the game enough to keep asking these questions, but other engines (or DirectX, goodness) aren't the answer. Your best bet is for us to do what we do, and make performance improvements. Like I said earlier, this is something we want to do. You can count on it happening over time.

My main gripe is that you drip feed a teaser and then nothing for months

[AstroCat]

Overall, love the series and fully support the company behind it, just frustrated at the performance of the game system even on high end hardware. I only wish the best of luck and success for you guys.

What about something like The Creative Assembly's Warscape engine?

[DaveDash]

Phil - I appreciate your comments. I understand that comparisons to other games might be unfair, and I appreciate that CMx2 does things that other games do not.

But the fact remains that no matter what hardware I through at your game it doesn't run in a satisfactory manner. The net result is I find large battles a chore playing at 20fps when every other game I have ever played runs like a dream.

You get used to a certain standard and maybe our expectations are wrong, but I am glad to hear this is something you guys will constantly try to improve with your limited resources.

As long as there is some sort of acknowledgement that there is room for improvement and it is something BFC would like to work on then that is enough for me, because it may seem fussy, but I spent so many years as a poor student with a crappy computer that now I really enjoy games that run smooth.

[Guest]

Overall, love the series and fully support the company behind it, just frustrated at the performance of the game system even on high end hardware. I only wish the best of luck and success for you guys.

What about something like The Creative Assembly's Warscape engine?

Ooh, they've got some serious tricks at their disposal. There are a lot of crowd-based optimizations (bone / animation sharing, etc.) that they can use to keep their rendering overhead down. Also their troops don't have to look perfect close up (although they do look pretty good) so they can keep texture and overall model sizes low. And their individual soldiers don't have their own TacAI. Our 8-900 guys is probably ten times the load, AI- and rendering-wise, as a few thousand in Warscape.

A scenario in our game can be micro-managing a squad of men through an ambush in a village, or an assault by a couple battalions of infantry with supporting armor. And it has to look decent at either end, and the soldiers have to handle themselves reasonably. You'd notice if a squad of troops were simply carbon copies of one another, or if individual soldiers didn't have the AI they do. That kind of flexibility has a price - we can't make nearly as many assumptions as most engines can.

If we were doing an era where fairly rigid formations or at least close order were more common then we might benefit from some of the optimizations that Warscape can use. So if we did, say, a Napoleonic game. But then they've already done that.

[DaveDash]

The thing is that I am not really buying though Phil is that it seems that shadows seem to be the main culprit and not AI. With all due respect I've got loads of CPU taxing games on my machine (mostly Paradox games) which do number crunching that makes the numbers in CMx2 seem pitiful. Victoria 2 for example manages population movements of a simulated world that is extremely processor intensive.

As demonstrated earlier I don't see problems with CPU load but rather strain on the GPU, especially with shadows. Maybe the calculations for their geometry is handled by the CPU? The shere number of them visible at once on a map?

And with all due respect while the unit details are great, the terrain is kind of average (and not exactly detailed while zoomed out either). And this is where I am having trouble grasping your explanation from a purely graphical standpoint. Ok yes, CMx2 has to deal with rendering loads of troops and terrain, but the sheer amount of objects being rendered in some FPS games is pretty astounding (with high res textures, complex shadows, etc) too. In CMx2, units may look detailed from far away but the terrain doesn't.

In regards to units, If the units themselves retain their high res textures from all angles, and high details the more you zoom out, which contributes to bad frame rates even on top end gear, then wouldn't that actually indicate the engine is unoptimised? Ie, you're not using little tricks of your own when players zoom out? I understand and appreciate that this task may be difficult given the resources you have.

[Guest]

The thing is that I am not really buying though Phil is that it seems that shadows seem to be the main culprit and not AI. With all due respect I've got loads of CPU taxing games on my machine (mostly Paradox games) which do number crunching that makes the numbers in CMx2 seem pitiful. Victoria 2 for example manages population movements of a simulated world that is extremely processor intensive.

They're not interacting with a 3D environment. Or performing LOS calculations. They're essentially altering a database. Apples and oranges. AND on top of that we need to render the world, which isn't cheap. Victoria 2 is a technical feat but for its calculations it does have more bandwidth to work with. All game development is a balancing act. CM has a particularly hard one to deal with, based on what we've chosen to allow.

As demonstrated earlier I don't see problems with CPU load but rather strain on the GPU, especially with shadows. Maybe the calculations for their geometry is handled by the CPU? The shere number of them visible at once on a map?

Shadows aren't really a "strain" on the GPU. However, they do require multiple rendering passes. This means they literally take a multiple of the normal frame time to work out. They do in every game with dynamic shadows (unless more expensive shadowing techniques are used). If you turn off shadows in other games you'll see the same effect. If you're *not* seeing that it means they're not using full dynamic shadows.

And with all due respect while the unit details are great, the terrain is kind of average (and not exactly detailed while zoomed out either). And this is where I am having trouble grasping your explanation from a purely graphical standpoint. Ok yes, CMx2 has to deal with rendering loads of troops and terrain, but the sheer amount of objects being rendered in some FPS games is pretty astounding (with high res textures, complex shadows, etc) too. In CMx2, units may look detailed from far away but the terrain doesn't.

You're talking about two entirely separate things here. The graphics in CM not being as detailed as FPS games doesn't mean they're cheap to render.

A "busy" FPS like your average Call of Duty is pushing about as much through the GPU as CM does, in various forms. And as we've established they can make assumptions and optimize much more heavily to get the best performance out of what they push through. They can even "optimize for quality" meaning they can decide how many troops, etc., they want on screen to be able to use a particular level of rendering ability. Most FPS engines have hard limits on visible object counts. We can't do that.

In regards to units, If the units themselves retain their high res textures from all angles, and high details the more you zoom out, which contributes to bad frame rates even on top end gear, then wouldn't that actually indicate the engine is unoptimised? Ie, you're not using little tricks of your own when players zoom out?

I feel like we're constructing a precarious strawman here in terms of how CM works. CM doesn't compare well to other games. We do the structural optimizations we're able to. You can't tell by looking at the game what is cheap, what isn't, what we can do to improve it, or what optimizations we have or haven't done. You're welcome to guess away, but your chances of actually hitting the mark aren't good, and we're getting into levels of detail about game operation and planning where I can't really confirm or deny (nor would I expect Steve to, if we get down to it).

You can rest assured that we'll keep trying to squeeze more performance out of your post-student gaming machine. Hopefully that's the takeaway for this. We definitely do acknowledge that more work is needed, and we plan on doing said work.

Trying to guess at precisely what that work is going to entail is speculation at best, and I'd much prefer to use my time on the forums chatting about the game than trying to keep threads like this from turning into a "well why don't they just do X like game Q!" conversation that doesn't do us or you guys any good.

So: I'm off to do some coding. Again I'm glad you guys like the game so much that you're willing to spend your off time talking about it with us. And thanks for keeping things nice and polite, too. I understand your frustration and confusion. I do prefer to remedy it by fixing things rather than explaining for the nth time why CM isn't like games X, Y, or Z, though.

[DaveDash]

Fair enough regarding the comparisons. I don't know enough about programming so I'll take your word for it.

We just want a great game to be perfect

I'm also very happy by the way with the release state of the game, given the trend these days to release first fix later.