How Long And How Much Work To Build Future New Engine?

[HorbeySpector 164]

If they just add rivers and lakes, it would satisfy me for a long time.

Why do people like you try to turn every thread about Arma's engine into a feature request thread?

Once someone starts doing that, more follow and eventually the thread will get closed. :/

[roshnak 41]

I certainly wouldn't dismiss that as there's always a possibility - albeit remote. But there seems to be a general consensus that Arma isn't optimised for newer hardware. I've read it from numerous sources that people with hi-spec builds are getting on average 20FPS. It's just ironic that I was getting better performance on my older rig. I'm naturally disappointed and I regret that after many years of great gameplay on the Arma series I'm now looking to play games to take advantage of the gaming rig, otherwise it's a waste.

Well, as someone with modern hardware (4690k, GTX 970), I can only say that I am not experiencing the same issues you are, and neither are others that I know with similar hardware. It seems highly unlikely that "Arma just runs worse on modern hardware" is the explanation for your problems.

[dayglow 2]

People seem to forget other engines aren't built from the ground up each iteration. ID Tech, Unreal, Cryengine, frostbite, etc are just as old as the Arma.

To to see how long a rewrite takes look at Eagle Dynamics and the Edge engine that they are getting close to releasing for DCS. It has been around 4 years in the making to make better use of CPU cores. I have no doubt BI is working in the back ground improving things.

[vran. 13]

To to see how long a rewrite takes look at Eagle Dynamics and the Edge engine that they are getting close to releasing for DCS. It has been around 4 years in the making to make better use of CPU cores. I have no doubt BI is working in the back ground improving things.

It's a bit surpising to me that many CPU intensive games seem to utilize multiple cores rather poorly. This doesn't apply just for ArmA but also Total War series, DCS, Flight Simulator, ect. Either the developers of such games haven't yet quite gotten to grips with new hardware (or 'new', since quad cores are on the market for several years now) or something is really 'bottlenecking' the overall performance. I've recently heard a lot about how GPUs only communicated with one CPU core under Direct X 11 and how DX12 is supposed to change all that. Maybe that's why BI already expressed big interest in DX12.

[inlesco 233]

It's a bit surpising to me that many CPU intensive games seem to utilize multiple cores rather poorly. This doesn't apply just for ArmA but also Total War series, DCS, Flight Simulator, ect. Either the developers of such games haven't yet quite gotten to grips with new hardware (or 'new', since quad cores are on the market for several years now) or something is really 'bottlenecking' the overall performance. I've recently heard a lot about how GPUs only communicated with one CPU core under Direct X 11 and how DX12 is supposed to change all that. Maybe that's why BI already expressed big interest in DX12.

Regarding your observations, it might be the case. Either way, developers have one more excuse to come up with :D.

I'd like a large-studio or BIS' programmer elaborate on the issues of CPU-intensive / high AI count on screen / scalable games about CPU utilization to clear things up once and for all.

Perhaps there's blogs such programmers constantly write we're not even aware of.

[calin_banc 19]

I liked how Dslyecxi went on about blaming the users for using mods and how "advanced" the gameplay is for the shortcomings of ArmA 3 and ArmA in general or Day Z. Just no, the game cannot sustain 60fps even on low settings and at times, even at lower settings than low (yes, you can go beneath with settings in menu than the preset "low" used by Bohemia). We're talking about a 2500k@4,5GHz, 8GB RAM and a R290 in and with Bohemia own made assets and mission in SP! Good luck to all AMD CPU users or other lower performing PC users in keeping a good enough FPS (and I'd say a rock solid 60fps is paramount for this kind of game) at a decent quality and that without actually going for a full scale conflict!

Perhaps we'll be lucky and get some nice FPS boost via DX12 for the render at least, if will ever be implemented into RV (although, Mantle with it's theoretical 9x more draw calls seems more appealing, especially for AMD CPU users). How much time to fix/build a new one? As much as it needs, because an improvement is a must!

[Sandy* 10]

It's a bit surpising to me that many CPU intensive games seem to utilize multiple cores rather poorly. This doesn't apply just for ArmA but also Total War series, DCS, Flight Simulator, ect..

Here's the thing - all videogames utilize multiple cores poorly and still rely on the power of a single core a lot. It's just the simming genre where things choke up due to having much more things to calculate. Multicore programming is hard, and it is much easier to optimize a corridor shooter to run at 60 fps with size x map and y amount of AI by using sleigh of hand tricks than it is to make a simulator with varying map sizes, varying numbers of AI and an open environment without compromising the accuracy of whatever it tries to simulate.

[maddogx 13]

I'd like a large-studio or BIS' programmer elaborate on the issues of CPU-intensive / high AI count on screen / scalable games about CPU utilization to clear things up once and for all.

The AI don't even have to be on screen. If enough of them are being simulated somewhere (even 15km away), they will bottleneck the main loop no matter what, causing framerate drops.

I posted a repro mission showing this a long time ago. :)

[inlesco 233]

Does RV4 engine iteration has its renderer separated (current DayZ SA primary task)? Think that'd somewhat increase performance, though I'm just speculating.

The AI don't even have to be on screen. If enough of them are being simulated somewhere (even 15km away), they will bottleneck the main loop no matter what, causing framerate drops.

I posted a repro mission showing this a long time ago. :)

Yeah, it's logical. You can say it's for Zeus to make the mechanics all the more flexible, so you can issue waypoints and other orders to any unit on the map without spawning it. Or you can disableSimulation.

[bad benson 1733]

i wish there was an inbuilt caching system that one could activate and manipulate on the fly. like a slider on each unit for "simulationDistance" or something much like the skill slider for which you then could change the value dynamically via script functions. maybe something that puts the AI in a simplified hold loop where it maybe does some dice roll stuff based on loadouts and stuff, similar to ALiVE.

it's a real shame the engine doesn't do such things by default. it's one thing to say that arma allows 100s of AI but making i work dynamically and scale intelligently is another thing. i don't even see the point in having stuff actually happen when no players are around in most cases. at least not as complex as when they are actually around. without stuff like that arma is sadly just a shooter that is too big for its own good. i get that there has to be a limit where one can't expect it to run well anymore since you can place way too much stuff with the editor (which is great) but the limit is currently way too low especially compared to arma 2.

[badluckburt 78]

Does RV4 engine iteration has its renderer separated (current DayZ SA primary task)? Think that'd somewhat increase performance, though I'm just speculating.

I think that the render / simulation processes have been separated as far as possible. There's an old article by Marek floating around that explains a few things about optimizations, I can't do a search for it right now but he also talks about the separating the render process from simulation in that.

[Sandy* 10]

Yeah, it's logical. You can say it's for Zeus to make the mechanics all the more flexible, so you can issue waypoints and other orders to any unit on the map without spawning it. Or you can disableSimulation.

It has nothing to do with Zeus specifically. AI units outside your draw distance are fully simulated because anything can happen - combat against other AI, artillery, realistically traversing terrain. Any interaction, you name it. This is why some missions are badly made, because the designers insist on filling them with useless units instead of spawning/deleting/caching them when needed.

[Greenfist 1863]

i wish there was an inbuilt caching system that one could activate and manipulate on the fly. like a slider on each unit for "simulationDistance" or something much like the skill slider for which you then could change the value dynamically via script functions. maybe something that puts the AI in a simplified hold loop where it maybe does some dice roll stuff based on loadouts and stuff, similar to ALiVE.

it's a real shame the engine doesn't do such things by default. it's one thing to say that arma allows 100s of AI but making i work dynamically and scale intelligently is another thing. i don't even see the point in having stuff actually happen when no players are around in most cases. at least not as complex as when they are actually around. without stuff like that arma is sadly just a shooter that is too big for its own good. i get that there has to be a limit where one can't expect it to run well anymore since you can place way too much stuff with the editor (which is great) but the limit is currently way too low especially compared to arma 2.

I thought the game already does some caching by default. If you make a 300 AI fight 3km away, you'll see the framerate drop gradually as you move closer to them.

[Sandy* 10]

I thought the game already does some caching by default. If you make a 300 AI fight 3km away, you'll see the framerate drop gradually as you move closer to them.

The closer you get, the more there is to draw (LOD).

[calin_banc 19]

Even if there isn't much to draw, the FPS tanks. Occlusion is fubar in RV.

[Greenfist 1863]

The closer you get, the more there is to draw (LOD).

What if you walk backwards?

I could be really wrong (have to test it), but I'm pretty sure the fps drops even if the action is offscreen.

edit. Made a video:

Note the dropping FPS as I get closer to the 200 units fighting. (Don't mind the music. I started the game with -nosound) Edited January 26, 2015 by Greenfist

[I give up 152]

Arma 3 does not need a new engine, Arma 3 needs to have the engine adapted to reality.

Whats the point of having support for over 100 players if cant handle with it? Cut it to a value that the engine can handle.

Why having a huge island filled with snakes and fishes when the engine cant handle with it? Reduce the size to something that the engine can handle.

Whats the point of having nice looking towns if we cant have there a decent firefight with some AI? Reduce towns size and buildings amount.

Why having shiny SUVs when in matters of hardware it cost the same as rendering 3 armored? No point when A3 is a military game.

And the list goes on ...but basically what we have is game bigger than the Engine itself, adapt it to reality and A3 will become what was intended.

[calin_banc 19]

There is no need for Bohemia to cut anything as it stands now; just server owners keeping it real, but since they stand by those high numbers slots, they must enjoy that FPS as well. :)

[esfumato 75]

Some reasons to have a new engine.

The lightning issues about lights traspassing walls and buildings... only for that I think it needs a new engine.

I hate how the engine interact with objects added to the terrain. Enterable buildings, towers, stairs, have a lot of clipping, players trasspass the walls a lot, or you can get the whole body of the player stuck in the floor of a building.

Another thing that annoys me too much is the LOD, game lets you use artillery, tanks that can shoot 3 kms away, Missiles that can hit targets kms away. But you can't properly difference at 1 km the diference of a grey rock and a tank. You must use optics to zoom in!!!

[thenach 10]

Even thou a new engine will be good for issues with clipping and lighting... I don't care for eye candy.

We don't need a new engine, we need this in ARMA:

https://bisimulations.com/virtual-battlespace-3

Bring back ARMA to what is supposed to be.

[spanishsurfer 58]

Overall, Arma runs pretty well for me. I understand the complexity of the game and know that there's nothing else out there that does what Arma does, so I don't bitch when my FPS dip. Those of us who played Op Flash, ARMA, ARMA2, ARMA2 OA, etc...know the engine/series has come A LONG WAY. I remember during ARMA/ARMA 2's release we suffered from desync that was so bad it made PVP gameplay almost impossible. The only real way to have a decent gaming expereince online was by playing domination or Team vs AI, and that's why that gamemode thrived for a long time. It wasn't until late in the Arma 2 series (After OA released) that the desync was addressed with a miracle patch. After that CTI, I&A, etc all was tolerable. Those of us that have been around for a while have dealt with WAY more bugs, desync, and FPS issues than many of the new wave of posters I see on here and I think are more tolerant on issues related to the game.

Many of you want a new game engine, to that I say....fuck that. It would take several YEARS just to develop the engine and several more YEARS just to make it stable/playable. Many of you will prob finish highschool, college, and/or start your careers and prob have a family by the time a new engine shows up.

I say stick with this engine and keep working on it, it's already polished. There are a lot of exciting new techs out there like DX12 that may help ARMA a lot. Maybe someone can finally figure out the heating issue with CPUs and we can see another age where CPU speeds double every few months....or maybe some other dev out there will create a competitor to ARMA, who knows. A new engine from scracth is wishful thinking and not very realistic.

[Jackal326 1158]

Overall, Arma runs pretty well for me. I understand the complexity of the game and know that there's nothing else out there that does what Arma does, so I don't bitch when my FPS dip. Those of us who played Op Flash, ARMA, ARMA2, ARMA2 OA, etc...know the engine/series has come A LONG WAY. I remember during ARMA/ARMA 2's release we suffered from desync that was so bad it made PVP gameplay almost impossible. The only real way to have a decent gaming expereince online was by playing domination or Team vs AI, and that's why that gamemode thrived for a long time. It wasn't until late in the Arma 2 series (After OA released) that the desync was addressed with a miracle patch. After that CTI, I&A, etc all was tolerable. Those of us that have been around for a while have dealt with WAY more bugs, desync, and FPS issues than many of the new wave of posters I see on here and I think are more tolerant on issues related to the game.

Many of you want a new game engine, to that I say....fuck that. It would take several YEARS just to develop the engine and several more YEARS just to make it stable/playable. Many of you will prob finish highschool, college, and/or start your careers and prob have a family by the time a new engine shows up.

I say stick with this engine and keep working on it, it's already polished. There are a lot of exciting new techs out there like DX12 that may help ARMA a lot. Maybe someone can finally figure out the heating issue with CPUs and we can see another age where CPU speeds double every few months....or maybe some other dev out there will create a competitor to ARMA, who knows. A new engine from scracth is wishful thinking and not very realistic.

^ This. 100% this.

[foffy 58]

A new engine also risks the community having to relearn something all new, too. This is such a franchise where starting fresh may be more problematic than trying to refine. And who is to say that with a totally new engine it doesn't have some kind of issue? Saying "oh, throw it away and make a new engine" is a very simple-minded approach.

If any of you know of the company Square-Enix, they're a perfect example to this problem. This company in the span of seven years produced two new engines. The first of which - Crystal Tools - was entirely dropped for a new engine after only a few major titles were released under it, and what took its place was the Luminous engine. They've already scrapped support for the Luminous engine for their titles, save for the game they migrated towards it: Final Fantasy XV. They've since gone on to licenses Unreal Engine 4 for their internal projects. Making a new engine isn't easy, and you create a lot of trouble by trying to make something all new instead of refining your foundations. Crystal Tools was an engine made for open world titles, but they had to scrap it when they slowly learned it actually couldn't do open world titles very well. I'd rather see Bohemia refine what they have instead of risking making an all new engine that may or may not really address these issues.

[infiltrator_2k 29]

Overall, Arma runs pretty well for me. I understand the complexity of the game and know that there's nothing else out there that does what Arma does, so I don't bitch when my FPS dip. Those of us who played Op Flash, ARMA, ARMA2, ARMA2 OA, etc...know the engine/series has come A LONG WAY. I remember during ARMA/ARMA 2's release we suffered from desync that was so bad it made PVP gameplay almost impossible. The only real way to have a decent gaming expereince online was by playing domination or Team vs AI, and that's why that gamemode thrived for a long time. It wasn't until late in the Arma 2 series (After OA released) that the desync was addressed with a miracle patch. After that CTI, I&A, etc all was tolerable. Those of us that have been around for a while have dealt with WAY more bugs, desync, and FPS issues than many of the new wave of posters I see on here and I think are more tolerant on issues related to the game.

Many of you want a new game engine, to that I say....fuck that. It would take several YEARS just to develop the engine and several more YEARS just to make it stable/playable. Many of you will prob finish highschool, college, and/or start your careers and prob have a family by the time a new engine shows up.

I say stick with this engine and keep working on it, it's already polished. There are a lot of exciting new techs out there like DX12 that may help ARMA a lot. Maybe someone can finally figure out the heating issue with CPUs and we can see another age where CPU speeds double every few months....or maybe some other dev out there will create a competitor to ARMA, who knows. A new engine from scracth is wishful thinking and not very realistic.

Agreed. As long as the engine can eventually be adapted to support multi-core processors. If it cannot then it's destined for the scrapheap.

[KeyCat 131]

People seem to forget other engines aren't built from the ground up each iteration. ID Tech, Unreal, Cryengine, frostbite, etc are just as old as the Arma.

To to see how long a rewrite takes look at Eagle Dynamics and the Edge engine that they are getting close to releasing for DCS. It has been around 4 years in the making to make better use of CPU cores. I have no doubt BI is working in the back ground improving things.

AFAIK EDGE is not a new engine, it's an improved version of the current one that have it's legacy all the way back to LOMAC 1.00 released 2003.

/KC