Bohemia Interactive's ambitions are always set too high.

[NeuroFunker 11]

Do we need 300 sq. km maps though? I love big maps, I love lots of area's to explore, but do we need these sized maps in ArmA for ArmA to be interesting, and I mean at the cost of performance, playability and stability? Wouldn't it be better to have smaller but detailed maps that fulfill the both the trouble area's as well as being interesting and functional within the scope of ArmA? What's the fascination with increasing the size of the maps if the engine can't handle it, beyond the fact that you're increasing the size of the map for the sake of increasing it's size and that WOW factor. In that regard, I definitely feel like BI's ambitions were too high in a negative way. I'm talking from the perspective of the road we are on right now which is heading in the direction of ArmA 4 in the same boat as ArmA 3 currently or whatever sequels/DLC/Expansions that are coming after ArmA 3.

do we need maps the size of a coin too? Look at stratis, how long would you want to play there? I was sat already before altis came (played since early alpha), and did not really want to touch it again, untill first campaign episode. Altis is already 1/3 downscaled of real island size, so BI actually though of to largy terrain would make it impossible to play. Altis still didn't tire me out, there are lots of places i've not bean, and untill next island comes, i might just get tired shortly before. I don't think missions like warfare, altis life, would be much fun on smaller islands. There might be a question, if you can't opzimize the engine/island well, so it performce good on medium/low end rigs, and for some highend rigs even, why would you bother making such huge island then?

Still to remember, when arma 2 came, what nightmare it was, was not playble for nearly anybody.

Also, since many missions, not global ones, never use the whole island, maybe it would be wise, to split Altis to like 4 parts, so the most players would have good fps?

[Opticalsnare 12]

The size of the islands is NOT the issue. The problem is that the engine does not fully utilize the other aspects in its design.

[mercenar1e 10]

hopefully they use the $20 million in sales from DayZ to make another engine... but unfortunately i don't see that happening as the devs are hardcore penny pinchers lol

[Instynct 1]

what about map size? I saw BF sized maps only. Maybe BIs could make similar if they wanted, editing map size utes, pumping it up with lots of details.

Map size shouldn't mean squat if your engine was built properly unless you're going to have view distance maxed. There should not be major performance changes between a giant map and a small map if view distances are the same. Loading a larger map should only increase memory usage.

[antoineflemming 14]

Map size shouldn't mean squat if your engine was built properly unless you're going to have view distance maxed. There should not be major performance changes between a giant map and a small map if view distances are the same. Loading a larger map should only increase memory usage.

Honestly, the large map isn't the thing that affects FPS the most. I don't understand why we're debating the size of the map. It really doesn't matter that much. What affects FPS the most are the trees, and even more, the amount of AI and what they are doing. If you had a large, Altis-sized, empty desert, it wouldn't do much at all to your FPS. But put a lot of trees on there, or put a lot of AI there, and have them fight each other, and you'll see FPS drop. It's not about the engine properly handling the terrain. It's about the engine properly handling AI and vegetation. A BF-sized map full of trees (let's just say one of the forests on Altis) will drop FPS more than Altis in one of its more open areas. A BF-sized map full of opposing AI will do the same. Not about the size of the map. It's about what's within your view distance (trees, buildings, or nothing) and about how many AI and what they are doing. For example, for me Takistan runs much better than Altis, not because it's smaller, but because there's less clutter. For me, Chernarus runs worse, because there's arguably more tree clutter.

[dnk 13]

I get far worse lag on Stratis specifically because of all the high detailed vegetation (oh, to be able to FORCE OFF the highest detail LOD of trees!). Sucks being GPU-limited (not for long).

do we need maps the size of a coin too? Look at stratis, how long would you want to play there? I was sat already before altis came (played since early alpha), and did not really want to touch it again, untill first campaign episode.

Stratis was boring because it was empty primarily, secondly because it was small. I played on Zargabad a ton in A2 because it was interesting, though small. Detail counts for a lot more than size. Stratis just lacked both.

We couldn't have something half the size of Altis but twice as detailed? Or a quarter/quadruple? Takistan was a bit large in my opinion (164km2), Chernarus just about "right" (~150 taking out ocean). Altis is needlessly large at 270km2 (excluding ocean again, which does actually add to this due to underwater whatevers). If they took all the resources that went into Altis and made an island the size of Chernarus with more detailed terrain, I'd be far more pleased. Especially since they haven't given us much to use on all that space (limited armor/aircraft). I really hope they stop with this "bigger is better" attitude for the announced island DLC.

Still to remember, when arma 2 came, what nightmare it was, was not playble for nearly anybody.

A flaw in that game as well...

[roshnak 41]

Honestly, the large map isn't the thing that affects FPS the most. I don't understand why we're debating the size of the map. It really doesn't matter that much. What affects FPS the most are the trees, and even more, the amount of AI and what they are doing.

Yeah, map size shouldn't really matter very much at all in terms of performance, given the streaming nature of the engine. Vegetation, on the other hand, has been a huge performance killer since Arma 1. I don't understand why this is the case when there are so many other solutions out there that impact performance so much less.

[NeuroFunker 11]

so you say: small but detailed map=large and detailed map? I thought poly count counts at the end. Bigger map. more polys, more performance hit.

Btw. about "19 players" and stuff before, here is my yesterdays recording of another 100 people game, where like at lest 60 people were battling eachother at the begining. We were protecting the nato general, from surrounded us csat troops.

www.youtube.com/watch?v=NDGPWjMnYxA

had like 30 fps there

Edited January 26, 2014 by NeuroFunker

[nikiforos 450]

The main problem is the AI and not the size of the map.

I have great FPS in an empty editor but as soon as AIs gets involved I get huge frame drops if the AI amount is moderate.

[Sealife 22]

so you say: small but detailed map=large and detailed map? I thought poly count counts at the end. Bigger map. more polys, more performance hit

No its often the case larger map based on Larger cell sizeand less details , smaller maps are often more densley populated , like cityscapes and such , BIS attempt at large populated but small terrain was Zargabad or more popularly known as Laggabad , other Engines outperform here by far thats why Arma is not so popular with people who like to play dense object filled maps.

whilst BIS have reduced to 7.5 m for Altis , this was apparent from day 1 a massive problem and all fixes subsequently have been a process to counteract the small cell size effect on average PC , this has lead to less Entity being configged , reduce texture size and other things , however some may say a larger cell size would have been better to start with and the reduction of 2.5m reall added no massive benefit to gameplay and only a little to the Pretty looking element.

had like 30 fps there

Again Its greta if your happy with TvT and No Ai but if Arma3 is only good for that then it certainly begs the question is the 18 month + development period to achieve a goal of stable play for all Arma game modes most specifically Co-op really too much , should BIS maybe set out with 2 goals and a less resource terrain that AI and players can enjoy be in parallel to a lifelike pretty terrain that only PvP or TvT can enjoy at 30FPS ?

Again BIS will getthere but the whole question is is it a measurable Aim this time ? time will tell .

Edited January 26, 2014 by Sealife

[NeuroFunker 11]

I'm playing BECTI sometimes solo, on my pc, whith 10 officers each side, 120x2=240 AI, can live with 30 fps, when the world feels so liveable.

[tortuosit 486]

Or how about this way: If RV is so much more amazingly powerful and capable than all of those other engines out there, why aren't companies breaking down the gates trying to license it?

They wanted to visit Bis exactly for licensing reasons; but I heard rumours that due to problems with the action menu, BIS' office doors could not be opened.

[tortuosit 486]

We know view distance and trees kills a lot Fps. It seems like the engine will completely draw and animate every far away tree branch. Can't far away stuff be drastically simplified? Instrad that we have to leave it out completely,.. we all know how ugly empty hills are, especially when you look through binoculars a lot... I'm sure its not that simple :-P

[duffbeeer 10]

It wont matter until BI decides to rework the server into a real multithreaded application. Right now theyre just trying to mask the issue with minimal effort.

[dnk 13]

tortuosit: it's the object quality setting. Set that to low or very low and you can punch out VD to 4000 without much of an impact.

I have an i5 3350P 3.3GHz, hardly a beast, and there's next to no difference in FPS between 800m and 3500m. But my IQ settings are low, so past 800m I'm only seeing very low-complexity models (and few of them) and terrain. Doesn't matter if it's urban or whatever. Only killer is when I get close to vegetation.

[carlostex 38]

There is no engine that does all the things Arma series does, and with the detail Arma does, period. People have to understand that this is not an easy thing to do, and that the only highly successful virtual military simulators are based on the RV engine.

Of course there is room for improvement, i believe the code could be optimized and they could implement and favor a low level API (like mantle) to take full advantage of what current hardware can do.

However, i would like to have Altis fully populated with NPC's, civilians, OPFOR's, BLUFOR's, traffic, ambient life, etc... The beauty of RV engine is that everything can happen at real time. The problem with this, is that it takes far too many resources as the nature of AI is too complex compared to other games. So BIS opted for a more player centric campaign, so that things only happen if player enters an area or does a certain thing to save on performance and avoid some dead ends. Of course this is not as realistic, but is more like a "game" and less like a "simulator".

Until we get 100GHz transistors, BIS can only improve so much. There is room for improvement, but as more is added the more performance is sacrificed.

A breakthrough in hardware is needed to provide a complex engine like RV all the performance for it too deliver a more realistic world like we know.

[windies 11]

There is no engine that does all the things Arma series does, and with the detail Arma does, period. People have to understand that this is not an easy thing to do, and that the only highly successful virtual military simulators are based on the RV engine.

Of course there is room for improvement, i believe the code could be optimized and they could implement and favor a low level API (like mantle) to take full advantage of what current hardware can do.

However, i would like to have Altis fully populated with NPC's, civilians, OPFOR's, BLUFOR's, traffic, ambient life, etc... The beauty of RV engine is that everything can happen at real time. The problem with this, is that it takes far too many resources as the nature of AI is too complex compared to other games. So BIS opted for a more player centric campaign, so that things only happen if player enters an area or does a certain thing to save on performance and avoid some dead ends. Of course this is not as realistic, but is more like a "game" and less like a "simulator".

Until we get 100GHz transistors, BIS can only improve so much. There is room for improvement, but as more is added the more performance is sacrificed.

A breakthrough in hardware is needed to provide a complex engine like RV all the performance for it too deliver a more realistic world like we know.

Why do we need a breakthrough in hardware when RV can't scale to the hardware we currently have?

As for island size, I simply asked as a hypothetical, IF it's a major cause of performance problems then do we really need islands of that size?

[bad benson 1733]

The beauty of RV engine is that everything can happen at real time. The problem with this, is that it takes far too many resources as the nature of AI is too complex compared to other games.

exactly my point earlier. the game not being able to handle its own size is not something to do with exclusively the size of the terrain mesh. it's how it's used in actual gameplay. i get way better FPS when i look at altis in the editor. that's hardly actual arma gameplay though. that's why Neurofunker's examples are pretty useless.

as i said before. there are many engines that already do PvP on a large scale (MMOs, Just Cause MP, etc) or could be scaled to arma's size to do so. so arma not being able to use its whole scale with stuff actually happening like you describe there is exactly the problem.

if BI's streaming system was as genious like people make it sound then the game would run perfectly (other games use mehods like LOD too, nothing outstanding here). instead we have complex problems that come from object count AND size. if there's more space to put things and far view distance you will quickly end up with a lot of objects in your view. that's also the cause of the main problem arma has. inconsistency. i remember when i got my graphics card how i went back to arma 2 to see how performance was. i placed myself on chernarus. i stood in one place (above Novy on the hill). when i looked towards the forest i got 60 FPS, when i looked towards the village i got fast changes between 20 and 35. Novy! not Kavala or something. so yea. the streaming and everything related to scaling smoothly is not optimal at all.

same goes for Altis. FPS differs way too much between the overkill towns and barren nature. now someone will say again "but it's big and many objects, it's supposed to not run good". to that i say and said before. exactly my point. this shows that arma's technology is not special when it comes to size and streaming. it's like any other engine scaled to something it can't handle well.

about AI/NPCs. i was always advocating for BI to work on engine side caching (like Alive but not squeezed into the game via sqf) to make the game finally handle what it promises/implicates. maybe it will happen some day.

that's what i'm trying to say.

II

\/

Oh, and btw, i am not sayng other engines would work better because there is no proof of that. But as it seems, RV engine doesn't do all that well either in it's current form.

Edited January 26, 2014 by Bad Benson
thx PuFu

[PuFu 4600]

@CarlosTex: what a bunch of bullshit^^

Future of processors is parallel computing, not 100GHz single core on a silicon transistor. That being said, i cannot agree with the view on the fact that BI releases their games for future hardware. And as previously said, an argument is arma2 in itself. All the ones saying that they are fine with 30Fps can argue as much as they want. It is as simple as that: this game does not take full advantage of the hardware it is being run on. No matter of the CPU or GPU, A3 cannot sustain a decent average FPS, at least in MP, wherein theory most of the CPU lifting is done by the server, and not by the clients.

Yes, i have been around long enough to see this thing forming some sort of pattern. Yes, a game can try and push the boundaries of hardware, Crisis 1 did it back in the day, but surely the visual options meant you were able to adjust your settings in order to maintain a healthy framerate. With A3, no matter of the settings one chooses, the fucking stutters and dips are sure to occur more often and not. For a game that has the GPU underused (maybe because it cannot have RT soft shadows and multiple light sources) and is single core oriented, there needs to be a lot more work done under the hood to keep it "future oriented".

Oh, and btw, i am not sayng other engines would work better because there is no proof of that. But as it seems, RV engine doesn't do all that well either in it's current form.

[Sealife 22]

There is no engine that does all the things Arma series does, and with the detail Arma does, period. People have to understand that this is not an easy thing to do, and that the only highly successful virtual military simulators are based on the RV engine.

The problem is it cant do it , thats what the thread is about , other Engines are overtaking everything Arma does and only lagging behind on terrain size , however with current world streaming even the very large mmos do it better than BIS imo .

Bis need no advice i`m sure they are fully aware but its no shame for them to look to reduce and stabilise sometimes rather than set out with Maximum achievement and band aid for 18 months

This chap will alledgedy recreate Chernarus style map in Cryengine (read 3rd comment ,vid is not relevant) , it will be great to test it someday , till then i myself am having fun making my own first Cry Terrain after 12 years of Making OFP and Arma ones . i may be back when Ai can handle and A3 tools are released.

Edited January 26, 2014 by Sealife

[jiltedjock 10]

There is no engine that does all the things Arma series does, and with the detail Arma does, period. People have to understand that this is not an easy thing to do, and that the only highly successful virtual military simulators are based on the RV engine.

Of course there is room for improvement, i believe the code could be optimized and they could implement and favor a low level API (like mantle) to take full advantage of what current hardware can do.

However, i would like to have Altis fully populated with NPC's, civilians, OPFOR's, BLUFOR's, traffic, ambient life, etc... The beauty of RV engine is that everything can happen at real time. The problem with this, is that it takes far too many resources as the nature of AI is too complex compared to other games. So BIS opted for a more player centric campaign, so that things only happen if player enters an area or does a certain thing to save on performance and avoid some dead ends. Of course this is not as realistic, but is more like a "game" and less like a "simulator".

Until we get 100GHz transistors, BIS can only improve so much. There is room for improvement, but as more is added the more performance is sacrificed.

A breakthrough in hardware is needed to provide a complex engine like RV all the performance for it too deliver a more realistic world like we know.

What a load of garbage.

If you run your missions as multiplayer on a dedicated server on your quad core, and run your client on the same quad core, you get far better hardware utilization and better, more stable client performance, because the AI is offloaded to a core which is otherwise largely unused by the Arma client.

If you load the PBOs from a Ramdrive rather than disk, you get better client performance because you are using far more RAM than Arma will use as it is a 32 Bit app.

These are two simple things to do which utilize CURRENT hardware far better than Arma does out of the box.

The software and its use of the hardware is what needs to change.

[antoineflemming 14]

There is no engine that does all the things Arma series does, and with the detail Arma does, period. People have to understand that this is not an easy thing to do, and that the only highly successful virtual military simulators are based on the RV engine.

Of course there is room for improvement, i believe the code could be optimized and they could implement and favor a low level API (like mantle) to take full advantage of what current hardware can do.

However, i would like to have Altis fully populated with NPC's, civilians, OPFOR's, BLUFOR's, traffic, ambient life, etc... The beauty of RV engine is that everything can happen at real time. The problem with this, is that it takes far too many resources as the nature of AI is too complex compared to other games. So BIS opted for a more player centric campaign, so that things only happen if player enters an area or does a certain thing to save on performance and avoid some dead ends. Of course this is not as realistic, but is more like a "game" and less like a "simulator".

Until we get 100GHz transistors, BIS can only improve so much. There is room for improvement, but as more is added the more performance is sacrificed.

A breakthrough in hardware is needed to provide a complex engine like RV all the performance for it too deliver a more realistic world like we know.

No, what you mean is that there is no other developer, with another engine, that's chosen to do all the things that Arma does. I would expect, that if a AAA developer with a current gen engine chose to make a competitor to Arma, that they'd have more success. BI's engine is outdated, and their company is fairly small.

Why do we need a breakthrough in hardware when RV can't scale to the hardware we currently have?

As for island size, I simply asked as a hypothetical, IF it's a major cause of performance problems then do we really need islands of that size?

Yours is a hypothetical, but again it's not a major cause of performance problems. Nor is it necessary to have islands as large as Altis. There is no real advantage except for marketing purposes ("we have a very large island where others don't").

What a load of garbage.

If you run your missions as multiplayer on a dedicated server on your quad core, and run your client on the same quad core, you get far better hardware utilization and better, more stable client performance, because the AI is offloaded to a core which is otherwise largely unused by the Arma client.

If you load the PBOs from a Ramdrive rather than disk, you get better client performance because you are using far more RAM than Arma will use as it is a 32 Bit app.

These are two simple things to do which utilize CURRENT hardware far better than Arma does out of the box.

The software and its use of the hardware is what needs to change.

I agree. BI are doing the same thing that Infinity Ward/Treyarch are doing, using an old engine and try to make tweaks and adjustments to it to make it seem up to par. Problem is, the heavily modified id Tech 3 engine that they have been using and modifying since 2005 only features small maps and rather bland graphics. But they're doing the same thing. BI's engine is old, too old.

[froggyluv 2136]

No, what you mean is that there is no other developer, with another engine, that's chosen to do all the things that Arma does. I would expect, that if a AAA developer with a current gen engine chose to make a competitor to Arma, that they'd have more success. BI's engine is outdated, and their company is fairly small.

Really? I find these types of statements somewhat ridiculous akin to "I could beat all MMA fighters with my secret ancient Tiger Claw kung fu technique -I just choose not to". Surely AAA developers have noticed a little zombie game I won't name here that has steamrolled *cough* steam's top seller (arma2 included) for some time. I would also strongly expect an announcement of an Arma-type game being developed by one of the big boys would receive a HUGE ovation and interest. The financial reward is already there for the taking yet I see no takers and if anything, the trend moving in the opposite direction of what Arma does.

I'm a pragmatist. Until I see it, I don't buy it.

[roshnak 41]

I would also strongly expect an announcement of an Arma-type game being developed by one of the big boys would receive a HUGE ovation and interest. The financial reward is already there for the taking yet I see no takers and if anything, the trend moving in the opposite direction of what Arma does.

What? No, the financial reward is not there. This is a niche market. Games like Call of Duty and Battlefield generate far more sales than Arma does, and most of the people who are interested in playing this kind of game in the first place are already aware of Arma and purchasing it.

[froggyluv 2136]

What? No, the financial reward is not there. This is a niche market. Games like Call of Duty and Battlefield generate far more sales than Arma does, and most of the people who are interested in playing this kind of game in the first place are already aware of Arma and purchasing it.

1 hit kill, bean scavenging apocolyptic zombie fare was not a forefront either but look at it now. If you don't think AAA hasn't investigated how to better tap that vein then you are mistaken. Space sims are also niche yet look at the overwhelming $$$ response to Star Citizen because it's using Cryengine and authors pedigree. Many are tired of AAA yearly regurgitations and I'm quite sure if DICE stated they were releasing a new spin off of BF that included go anywhere, full squad control, full AI in the scale of hundreds if not thousands, full ballistics, full map editor and mod support -you'd see a wave of exhuberant glee. Of course we can disagree in this speculation -thats just the way I see it.

AAA isrun by suits and money men simply weigh risk, cost of production to projected sales and the above is risky as hell....