Jump to content
Sign in to follow this  
squirrel0311

Arma 3 Engine - What would have been a better option and what can we learn?

Recommended Posts

Besides, in Arma 2 I had huge amounts of remote battles taking place in CTI and it was not a performance killer.

I tried placing 400 AI units (200+200) to fight against eachother in a bunch at one place, and myself about 20 km away.

At first I got 15-20 fps. Then after less than a minute, 25-30. With only one unit on the map, I would get 45-50 fps in the same area, near Kavala.

Then after a few minutes, teleported myself in the middle of the battleground, to find some fighting still going on and a couple of hundred bodies lying around.

Then I scattered 20 smaller fights with 10+10 AI all around the Altis.

20-30 fps.

Not bad.

Well, I tried with 800 AI too...that was maybe too much, less than 10 fps. But it's possible.

Edit: I scattered the 800 to 40 differend fights around the island. 15-20 fps. Not bad at all.

Edited by Azzur33

Share this post


Link to post
Share on other sites
I tried placing 400 AI units (200+200) to fight against eachother in a bunch at one place, and myself about 20 km away.

At first I got 15-20 fps. Then after less than a minute, 25-30. With only one unit on the map, I would get 45-50 fps in the same area, near Kavala.

Then after a few minutes, teleported myself in the middle of the battleground, to find some fighting still going on and a couple of hundred bodies lying around.

Then I scattered 20 smaller fights with 10+10 AI all around the Altis.

20-30 fps.

Not bad.

Well, I tried with 800 AI too...that was maybe too much, less than 10 fps. But it's possible.

Unit caching anyone?

Share this post


Link to post
Share on other sites
Unit caching anyone?

yea and also removal of dead units. i played on 8 versus 8 pvp servers that started lagging really bad until the mission was restarted and all the "garbage" was removed. just a few human players :rolleyes:. did anyone use the simulation manager in an actual mission yet? i'm curious if it's useful.

some time ago i said somewhere that i'd like to see BI research more unit caching stuff because it makes sense considering the size of arma but i got shot down by some crazy guy waving utilizaion charts around asking for a new engine.

i still think it would be a great investment to make more vanilla caching stuff (sites are a start). i mean i'm sure there is some going on in the engine but scripted caching like removal and recreation of units or dynamic "hideobject" and "enablesimulation" are way more effective. who knows. maybe it could be more integrated into the engine.

Share this post


Link to post
Share on other sites

I guess Arma had a limit of 256(?) groups at once on the map? So I could have 127 separate fights going on out of sight around the map... oh well. I had to try this too.

Share this post


Link to post
Share on other sites

i think the it was 144 or something. you could have huge groups though. not sure what the limit is/was there. itS' impressive that it works but to me it'S like viewdistance. there's testing settings and practical ones for everyday playing.

Share this post


Link to post
Share on other sites
thanks for the clarification.

Except he's wrong. Most major engines have been around for years and have seen many itterations with progressive sequals and games licensed to use the engine just like the Real Virtuality engine has. There are very few really new original engines coming out and those new engines that aren't developed by a developer that is backed by a major publisher hardly ever get to a stage where it can be used to run a game on let alone an industry standard for engine design. A good engine is modular so that parts of it can be redesigned and updated if needed. The problem is that BIS lacks the funding and manpower to upgrade every part of the engine for every itteration of ArmA.

Share this post


Link to post
Share on other sites

Hmm.. I tried with 240 groups (plus me) having 120 fights, but "only" 4 AI in one group. That alone made 960 fighting units on the map, so fps was 8-10.

Editor doesn't allow preview if there are more than 256 groups, but having basic 8-man infantry units, that would be 2048 units.

I've had at least 20 vehicle unit groups (20veh+plus crew), I wonder what is the group limit.

Too many numbers, I need some sleep.

Edited by Azzur33

Share this post


Link to post
Share on other sites
And on the DCS forums you'll see endless bitching about under utilizing of CPU and GPU, stutters, etc. Strangely familiar.

Yes I left DCS early on after A10 never bought another product, and I enjoyed A10 but couldn't get past all the bugs and crashes in a completely unfinished game. I'm so tired of our simulation developers doing a piss poor job, releasing things before they are ready.

---------- Post added at 12:39 AM ---------- Previous post was at 12:35 AM ----------

BIS don't want to change engines because one it costs money to licence a new engine and two they have to re-learn a bunch of new code and how it works.

So they are sticking to what they know and how to use it, which is fair enuff but imo is now actually beginning to harm them than actually improve their overall product. You can only go so far with an existing product before it reaches its limitation of usefulness. Sure you can upgrade it and slightly increase its lifespan but as soon as you start reaching that limited threshold for performance and functionality then to me thats a good sign to start looking elsewhere for a more suitable replacement.

This was made fully aware in A2's lifespan. And with A3 we still have the same core engine and its issues, but with ALOT of make-up plastered on it and features that were implemented poorly to mask its age.

Its like a 60 year old woman trying to make herself look 25 again and try and run the 200m hoping nobody will notice that she only gets 50m and ends up having a bloody heart attack. But hey its the taking part that counts right?

But personnaly imo i think thats it, A3 will be the last of the breed. Perhaps this is why we never got a replacement engine because BIS knew that the franchise was done after A3.

It was too much cost to them to invest into a different engine and far too much risk involved, and they are now insted moving onto something different.

- Game Over - :(

Well said. That is why I hope they are taking the money they raped from people for an unfinished product and using it right now to get a few guys working on a new engine to use 4-5 years down the road.

Share this post


Link to post
Share on other sites

My oh my, I return to Arma after a year or so and the first thing I read in the forums is this thread... never seen so much crap in one place.

I cannot spend any time quoting every piece of crap I found written here, so I will just answer a generic answer - if you think you can do a better job than Arma 3, well... go ahead, grab this book, and after some dozens (hundreds?) of thousands of man-hours, and many many head-bashing nights and weekends looking for that race condition in your inevitably buggy multi-threaded code, come back here and post your amazing product so we can all enjoy it.

Share this post


Link to post
Share on other sites

somebody mentioned Outerra Alpha engine in this thread.. they have a tech demo available if anyone is interested.

Share this post


Link to post
Share on other sites
its crazy how a game like GTA 5 can have as much view distance as Arma 3 and actually have better looking terrain, im guessing that is because its cell drawn anywho I wouldn't be surprised if BIS is completely burnt out on Arma and even more so because it got 2 of their guys held in prison over the project...

Or just cause 2 with an even larger island than altis and now matter how much crazy stuff is going on never seems to drop a frame. People even modded in MP

We keep getting improved graphics and arma 3 does look great but armas always looked good enough its just never run good enough.

Share this post


Link to post
Share on other sites
My oh my, I return to Arma after a year or so and the first thing I read in the forums is this thread... never seen so much crap in one place.

I cannot spend any time quoting every piece of crap I found written here, so I will just answer a generic answer - if you think you can do a better job than Arma 3, well... go ahead, grab this book, and after some dozens (hundreds?) of thousands of man-hours, and many many head-bashing nights and weekends looking for that race condition in your inevitably buggy multi-threaded code, come back here and post your amazing product so we can all enjoy it.

and BIS is suppose to do what? continue robbing Peter to pay Paul? ...we don't have the TOH flight model because it caused to much FPS loss....we cant shoot the shutters out of windows because it causes a further 10-%15 FPS loss. The thing that should really bring the future into focus for you is the fact that the new consoles are sporting a low clock hex core from AMD...how do you think engine's that cannot scale across multiple cores are going to fare in the environment?

There comes a point of diminishing returns which means evolve or die! business 101...keep up or pack up!

Share this post


Link to post
Share on other sites
Or just cause 2 with an even larger island than altis and now matter how much crazy stuff is going on never seems to drop a frame. People even modded in MP

We keep getting improved graphics and arma 3 does look great but armas always looked good enough its just never run good enough.

Not quite. Just Cause 2 map is 250km2 while Atlis is 270km2 of landmass and not including sea floor. The total map is 300km2.

Share this post


Link to post
Share on other sites

Also half of JC2 is very low detailed since it wasn't made to be playable (check those big mountains full of nothing) and all enemies and vehicles disappear into thin air 100m away and everything you destroyed gets deleted/respawned. How would that be suitable for ArmA where the average engagement range is twice the JC2's npc rendering distance limit.

RV is fine.

The horrible lack of programmers that are now forced to work on 2 RV versions at once is NOT.

Also FYI some of the "big" engines are about as or even older than RV.

Frostbyte - 11 years old

Unreal Engine - 15 years old

CryEngine - 12 years old

Quake3 engine used in CoD to this day - 14 years old

Obviously they have all the programmer force they need to make changes they deem necessary. Nobody starts from scratch nowadays anymore unless other engines cannot offer what's required for a new title.

Edited by metalcraze

Share this post


Link to post
Share on other sites
and BIS is suppose to do what? continue robbing Peter to pay Paul? ...we don't have the TOH flight model because it caused to much FPS loss....we cant shoot the shutters out of windows because it causes a further 10-%15 FPS loss. The thing that should really bring the future into focus for you is the fact that the new consoles are sporting a low clock hex core from AMD...how do you think engine's that cannot scale across multiple cores are going to fare in the environment?

There comes a point of diminishing returns which means evolve or die! business 101...keep up or pack up!

Uh? I wonder what does the consoles architecture have to do with Arma 3 at all. The requirements are completely distinct from a PC game like Arma 3. Console games need more rendering power than CPU power, are constrained on heat generation and power consumption... like the nVidia vice-president himself said, there is no way a 200-watt xbox is going to beat a 1000-watt pc.

All the engines and titles mentioned in this thread would struggle in a next-gen console hardware. And all those same engines and titles still do not extract 100% cpu and gpu power theoretically available from parallel processing in the PC. Actually, not even the PCI express buses are bottleneck to any commercial title - it is easy to saturate the buses in a one-off demonstration or an academic exercise, but not when you put everything together in a real-world application.

I think the BI customers should focus on discussing improvements of the features they want instead of the technicalities of the RV engine, because they will always get it wrong when discussing something they have no domain knowledge. You end up with people demanding engine XYZ, multi-core processing, deferred rendering, or any <insert your regurgitated technical blurp here>, without pondering which practical consequences it will have to the gameplay itself. Leave it to the developers instead, the only people capable of determining whether something needs to be rewritten or not are the people who have the source code.

Share this post


Link to post
Share on other sites
Not quite. Just Cause 2 map is 250km2 while Atlis is 270km2 of landmass and not including sea floor. The total map is 300km2.

Actually the map of JC 2 is 32km x 32km = 1024 square km.

Share this post


Link to post
Share on other sites
All the engines and titles mentioned in this thread would struggle in a next-gen console hardware. And all those same engines and titles still do not extract 100% cpu and gpu power theoretically available from parallel processing in the PC. Actually, not even the PCI express buses are bottleneck to any commercial title - it is easy to saturate the buses in a one-off demonstration or an academic exercise, but not when you put everything together in a real-world application.

I think the BI customers should focus on discussing improvements of the features they want instead of the technicalities of the RV engine, because they will always get it wrong when discussing something they have no domain knowledge. You end up with people demanding engine XYZ, multi-core processing, deferred rendering, or any <insert your regurgitated technical blurp here>, without pondering which practical consequences it will have to the gameplay itself. Leave it to the developers instead, the only people capable of determining whether something needs to be rewritten or not are the people who have the source code.

Well thanks to this thread and the cumulation of of known major engines I am able to see now pretty much exactly what you said in the first part. As for the second part...people are going to complain and demand all sorts of stuff regardless... The reason I wanted to know what engine might have worked better is because of all the talk here and other places about how out dated the current one is.

As I lack the necessary knowledge I will refrain from addressing these issues as well.
Well I'm interested in hearing suggestions, up and downsides, 100% correct or not so much. Since you have no idea then clearly this thread isn't for you, but thanks for your time. :P
More so is frustrating when the AI cheating is second to Stalker's. It's a pain in the bum to be killed through grass, bushes and all sorts of cover that has perhaps a small hole in it. For me that's the biggest issue at the moment for the single player experience. They are either dumb as a brick or super agent Smith like entities.

I completely understand what you’re saying because I too have fallen victim many a time to the laser precision of an AI super soldier firing from 800 meters away in the standing position while he was completely concealed in thick foliage.

However I would just like to clear up something really quick... Cover is considered anything solid that a bullet will not penetrate or will at least stop bullets for a while. Concealment is anything that hides your position but is not bullet resistant. (Grass, bushes, sheet metal, things like that.) Cover can be concealment but Concealment IS NOT COVER.

I don’t know the details of your situation but if you were already in a firefight then their method of engagement would be pretty appropriate, especially in terrain with thick vegetation. There is a saying for anytime you know the enemy is out there but you can’t see them… When in doubt empty your magazine. Complete and total area saturation by fire is an appropriate response when engaged with a concealed enemy as opposed to sitting tight and waiting for them to creep up on you and get a more advantageous position. It is done so in the hopes that you will achieve one or more of three things, suppress and stop the enemy’s movement, drive the enemy back, or best of all…Kill them.

As far as stupid AI goes, I have to agree. I usually avoid playing with AI, I much prefer shooting players. :p

I think if their visual detection system was tweaked a little more and if they were actually programmed with a few different random but tactically accurate remedial action procedures…that would help immensely. I’m not sure if they fixed it but I remember in the alpha it didn’t matter if you had a suppressed weapon, were completely concealed or not, after you fired a shot they knew exactly where you were.

Also, as far as I know they still have one basic immediate action (Lay down, return fire or just run) both of which are fine but I think it would be neat if you could pick cover points in the editor for AI who were defending areas.. I.E. If attacked, return fire, fall back to this position, and watch this sector. Rather than shoot a few and watch them run into a base and stand near the middle looking at a wall. (I do know that if they are near cover they will normally just lie down and start suppressing which I like and also the flanking movements are nice too.)

Now if we’re talking about AI drivers…. I think there should be a special option kind of like manual fire… If you’re the only player in the vehicle and everything else is AI you should be able to use the arrow keys and instantly override and stupid action the driver is trying to take, no matter what seat you’re in. However, if you have an AI driver, friend playing as the gunner and yourself as the commander, then the commander should have the override feature.

Ultimately what I would really like to see is a civilian AI system with as much random diversity and detail as Grand Theft Auto 5. Can you imagine a multi-player vs. player mission where you roll an armored convoy into a city like San Andreas? The mix of cautious bystanders who rush to return to their homes and the curious lookie loos who stupidly standby and peek around corners in hopes of seeing some action. You’d have it all! Planes, helicopters, tanks, cars, trucks, thugs, bums, hot chicks, fat people, skinny people, guns, of course... boats and hoes! (Note: This is probably what the Red Cross was talking about. Hahaha)

To quote Conan the Barbarian when he was asked by the Mongol general, What is best in life? - “To crush your enemies, see them driven before you, and to hear the lamentations of their women.â€

Anyway… this idea is what made me curious about the Forgelight engine since it claims to be able to handle up to a thousand players all doing different things at once. It seems to work fairly well for Planetside 2 but does anyone know much about it?

Share this post


Link to post
Share on other sites
Not quite. Just Cause 2 map is 250km2 while Atlis is 270km2 of landmass and not including sea floor. The total map is 300km2.[/QUOte]

The playable area is 1000km2,the real difference tho is i got 100+ fps in that game while arma its a struggle to maintain playable.

Those big engines have essentially been essentially rewritten over their lifetimes only thing incommon is the name, valve has a write up on the work it took to take source multicore http://techreport.com/review/11237/valve-source-engine-goes-multi-core

Share this post


Link to post
Share on other sites
The playable area is 1000km2,the real difference tho is i got 100+ fps in that game while arma its a struggle to maintain playable.

The other real difference is that the JC engine is player-centric, so it only needs to worry about what's going on around the player.

Share this post


Link to post
Share on other sites
Actually the map of JC 2 is 32km x 32km = 1024 square km.
The playable area is 1000km2,the real difference tho is i got 100+ fps in that game while arma its a struggle to maintain playable.

Those big engines have essentially been essentially rewritten over their lifetimes only thing incommon is the name, valve has a write up on the work it took to take source multicore http://techreport.com/review/11237/valve-source-engine-goes-multi-core

Playable landmass for JC2 is around 230 km2 from what I can find on the net. Atlis has 250km2 of landmass as well as a rendered sea floor.

Share this post


Link to post
Share on other sites

Its interesting, yes, but only thing its good for is flying and driving. What about AI, multiplayer? Syncing the world?

Again, discussions like these are pointless.

Share this post


Link to post
Share on other sites
BIS seem to be co conspirators with the hardware makers they keep luring us to buying new expensive CPU's and GPU's and selling us a new Arma game with alittle 'moar framez' as a selling point to each iteration as Jay Crowe the head honcho said himself.

Conspiracy Theory!

That or maybe they didn't click the optimize code button enough times perhaps they broke all their mouses???

not theory, not conspiracy ;)

it's pure fact :p http://jonpeddie.com/press-releases/details/global-pc-gaming-hardware-sales-shrug-off-pc-market-decline/

even analytics confirm :cool:

Share this post


Link to post
Share on other sites
Its interesting, yes, but only thing its good for is flying and driving. What about AI, multiplayer? Syncing the world?

Again, discussions like these are pointless.

Are you talking about Outerra? Lol it's still Work in Progress but they will eventually get to that point. Far better performance and the game world is Earth... Enter longitude and lattitude of any area in the world and it gets created in game.. What other game does that?

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now
Sign in to follow this  

×