Arma 3 : Operation Make Faster Game

[RushHour 11]

It does make difference until there's too much calculations for the CPU or the server is running slow.

Nope doesn´t do sh*t frankly.

And it´s in offline mode not on a server where what specs you have on your computer is rendered useless for the majority of the time.

[galzohar 31]

Well, if there is a lot of crap on your mission that needs to be calculated, it will make both server and client FPS lower. But it's not the server's FPS that makes your client's FPS low. You can, for a fact, show an example where the server has 7 FPS and the client has 60. This is achieved by creating a mission with a large number of patrolling AI infantry units (enough to bring the server to its knees) and nothing else, while having the player at a remote location looking at the sky. Try it yourself. Even if the server dies, your FPS will remain the same (except everything will continue to move more or less in the same direction it was moving the last time you received an update, but you will still be able to move around just fine, at least as far as you'll be able to see).

If your mission is a CPU killer, it will kill the CPU on both the server and client. But if your server's CPU is dead, it does not necessarily mean your client's CPU will be dead too. Of course, if the server has a decent machine and the mission kills its CPU, it's likely to kill client's CPU too (though not necessarily, as shown in the "only infantry AI" example).

As for a frame, it's practically the same on the server as it is on the client in terms of what it means. It's just that the server has to do a lot less every "frame", as it skips all graphical-related calculations. What the server calculates every frame is the same as what it would need to calculate before it can render the next graphical frame. So yes, you can safely use the term "FPS" when talking about the server's performance. You just need to realize the server doesn't actually draw the graphics every frame, but it does still do everything else non-graphic-related same as you (plus handling network traffic, of course, but that part has been shown already to not affect client FPS).

To me the problem is obvious - In addition to the game having extreme graphics which bring the most modern GPUs to their knees (which is OK, since you can always just lower your settings a bit and be fine with an average GPU), most missions that people run nowadays are simply too much for modern mid/top-end CPUs to handle when played on the current version of the engine. You can blame the missions, you can blame the engine, or you can blame both. In the end we don't really have a basis for comparison, as no other game really runs these kinds of missions to allow us to compare performance of different engines. I just hopes BIS lets us know what we can do to make our missions put less stress on the CPU rather than just wait until they optimize stuff some more and/or better CPUs are produced.

For now, you can just play missions that don't have 1000 vehicles all over the map. Stick to smaller PvP (or even COOP, if they're small enough) missions that periodically clean up the trash and your performance should be fine if your CPU isn't an aged piece of crap.

Edited October 5, 2013 by galzohar

[clydefrog 3]

Well, if there is a lot of crap on your mission that needs to be calculated, it will make both server and client FPS lower. But it's not the server's FPS that makes your client's FPS low. You can, for a fact, show an example where the server has 7 FPS and the client has 60. This is achieved by creating a mission with a large number of patrolling AI infantry units (enough to bring the server to its knees) and nothing else, while having the player at a remote location looking at the sky. Try it yourself. Even if the server dies, your FPS will remain the same (except everything will continue to move more or less in the same direction it was moving the last time you received an update, but you will still be able to move around just fine, at least as far as you'll be able to see).

If your mission is a CPU killer, it will kill the CPU on both the server and client. But if your server's CPU is dead, it does not necessarily mean your client's CPU will be dead too. Of course, if the server has a decent machine and the mission kills its CPU, it's likely to kill client's CPU too (though not necessarily, as shown in the "only infantry AI" example).

As for a frame, it's practically the same on the server as it is on the client in terms of what it means. It's just that the server has to do a lot less every "frame", as it skips all graphical-related calculations. What the server calculates every frame is the same as what it would need to calculate before it can render the next graphical frame. So yes, you can safely use the term "FPS" when talking about the server's performance. You just need to realize the server doesn't actually draw the graphics every frame, but it does still do everything else non-graphic-related same as you (plus handling network traffic, of course, but that part has been shown already to not affect client FPS).

Yes, and what you will see is the effect of the low server fps on the AI. They will move at a very slow pace (their animation will be normal speed but they will move a lot slower, same as when you see a player with bad fps doing the "slow run") and be very unresponsive.

Edited October 5, 2013 by clydefrog

[galzohar 31]

Yes, the effect of low server FPS is that when the AI need to react they will be more dumb and/or slower in their reactions, and when you shoot them there might be an additional delay until the hits actually register.

But it will not affect your displayed FPS at all.

[NeuroFunker 11]

But why is that? That's what I never understood.

don't ask me mate, i'm not the netcode coder :)

[sxp2high 22]

Our server never drops below 48 FPS, even with a 12 hour domination session. However the clients FPS will degrade over time regardless. The more players the worse it gets. I just experienced that yesterday. :)

[fabio_chavez 103]

imho, beside making assumtions about what those mythological figures at BIS are bound to do about these mysterious performance witchery that is plaguing us.

some instantly applyable realworld approach at hand would be: server hosters will have to engage in some kind of self censorship. eventhough it would require for them to comprehend the long therm damage that their shortsighted mission choices are doing to the GLOBAL server population and therefore for the gamecopy sales and eventually to BIS. The problem (in my view) is, the servers seem to receive positive gratification instantly, because eventhough the misisons they host arent playable for maybe half the population, they still accumulate the critical mass of players to keep their server running. This causes a drain of players from more broadly accessible servers that struggle to seed the critical mass of players to get the ball rolling. I think i can speak for a lot of people if i say this was the central source of frustration, right from the beginning of the alpha, when 90% of the populated servers didnt allow you to have acceptible performance (apparently despite your hardware or gfx setup) and you had to search long for a suitable server, sometimes in vain. Im convinced that if this problem had been recognized in the alpha/beta and something would have been undertaken, the global server population would be 3 or 4 times as much as it is right now (remember the playercount drop in the alpha when people realised that there where no indications that the multiplayer lameness would dissolve any time soon?).

What would you say, is it possible that it all basicly boils down to a instant gratification vs delayed gratification issue for servers?

If the community was able to undertake a collective and systematic approach to create a more accessible and fun server landscape the game would bloom instantly. (sometimes freemarkets, demand and supply just dont do it.)

Also BIS should have a stake in promoting such an attempt, are they hoping that the problem dissolves into thin air by itself or are they somehow scared to apply restrictions upon their free and innovative but sometimes counterproductively chaotic community?

p.s.

http://feedback.arma3.com/view.php?id=12658

http://feedback.arma3.com/view.php?id=1264

Edited October 5, 2013 by Fabio_Chavez

[LoneCrow66 10]

Yeah I did like planetside but it was a dog ... was a good video to see them dedicating resources into looking at what they had and how to improve it.

[dayglow 2]

That´s great but AA settings does exactly ZERO difference in Arma.

Neither does Ultra to Standard. We are talking if you are lucky, about 2-3 fps.

In other games that gap is 20-30 fps.

You don't? I see a very distinct drop going from none -> x2 -> x4 -> x8

I leave it off and rely on SMAA

[Beagle 684]

Incorrect.

A servers performance in Arma will have an effect on the clients FPS. Always has done.

Especially when the server starts to get below 20fps on the #montor.

See it all the time.

You play on a server thats running at max fps of 50 on #monitor and everyone will be fine. All getting the client fps they expect.

If your on a server thats running at #monitor fps 7 everyone will complaining about how they only have exceptionally low client fps.

I realise that what your saying should be true but it isn't where Arma is concerned.

Unfortunately you're right, a problem that came with ArmA II and is still not solved. But the biggest problem hereby is that servers admins often refuse to realize the connection beetween too big missions and low client fps making missions unplayable for all but the high end hardware clients.

[smiley_ie 7]

Update Part 3 and Part 4

Update

Part 3

http://www.youtube.com/watch?v=6kUyl...Mm-XrfT0ldALjg

Interesting stuff with part 4 and PhysX

[smiley_ie 7]

I was a away for a few weeks and got to try PS2 after all the "Make game faster" and have to say wow they done a great job i hang around 57 - 56 fps in large battles with there new fps smoother. General feedback from it seems people got better performance, So BIS how about you hold the DEV patch's and focus on optimization ?

[jiltedjock 10]

I find Arma runs fine as long as you keep realistic view distance and AA settings. These 2 can bring a system to its knees.

I have a i3820 w/ sli 660ti's. It's a high end system, but I do push 3 1080P monitors with it. With SMAA normal and 3000m view distance I can max all other settings and still get 30+ fps. If you're a 60fps or give me death type person you'll never be satisfied.

So what is the point of any of the weapons or vehicles that can engage > 3000m ?

[bvrettski 10]

Is the problem here that BIS has never figured out the relationship of:

Poor core game code + scale of the map(s) + Physx effects + AI overhead + netcode X numbers of players +/- quality of players PC = Barely playable game

Barely playable game + little official content = "Welcome to the A3 beta program!!"

"Welcome back to the A3 beta program!!" = Nothing to see here. Move along....

LOL I'm venting...sorry

I do love how we all offer our opinions, debate the situation around the game and never really get any acknowledgement or insight on these issue from BIS. For a company relying so heavily on the community to fix / fill in all the missing and broken parts you would think they would communicate more openly with their non paid interns (those who freely pour their time and energy into trying to make the game better). I guess they would risk alienating anyone hasn't already figured out how messy things get around an Arma release.

[timmo1982 10]

Is the problem here that BIS has never figured out the relationship of:

or is the problem that PLAYERS are still expecting to turn up all settings to the Maximum and get 60 FPS? Arma, like FSX (Flightsim 10) was, have ambitions beyond the 'here and now' of computer hardware. They have rendering capability that extends beyond current hardware and instead of players just pulling their sliders back to accommodate their particular set up, people expect to be able to go all out.

Now, BIS could probably just pull back their ambition, artificially restrict view distance to a max of xxxxx for now and release a patch later that just extended that.....but that doesn't get to the root of the issue in my mind. You may still have people complaining about performance on their 3 year old machine not being able to max out the settings even then.

Now, granted, ArmA can be optimised further, but I think a lot of the problems of 'poor performance' are simply related to the HUGE amount of rendering that the engine is capable of and players automatically expecting that their current hardware can crunch it. No amount of optimisation will get around the fact that if you have more data to load, more objects to display, more objects to move and track in XYZ space etc etc then you will simply need more computing power to do that.

[jgbtl292 0]

or is the problem that PLAYERS are still expecting to turn up all settings to the Maximum and get 60 FPS?

? singelplayer is ok - the muliplayer is the fps desatster

[Vibrant_Viper 0]

? singelplayer is ok - the muliplayer is the fps desatster

Yes and I think you should look to what .kju mentioned. The singleplayer is running well point says that there is a problem in the syncing. Not saying that it's problem like a bug, but the data transfer between server and clients bog down the fps.

At least that's how I understand it. Now I don't have any technical knowledge of the game, but I do see that when playing solely on TvT servers the FPS is amazing. When AI is introduced it quickly drops in performance.

[Alabatross 1]

I was a away for a few weeks and got to try PS2 after all the "Make game faster" and have to say wow they done a great job i hang around 57 - 56 fps in large battles with there new fps smoother. General feedback from it seems people got better performance, So BIS how about you hold the DEV patch's and focus on optimization ?

this

BiS needs to stop everything and just focus on performance, it'd be so great

[adanteh 58]

or is the problem that PLAYERS are still expecting to turn up all settings to the Maximum and get 60 FPS? Arma, like FSX (Flightsim 10) was, have ambitions beyond the 'here and now' of computer hardware. They have rendering capability that extends beyond current hardware and instead of players just pulling their sliders back to accommodate their particular set up, people expect to be able to go all out.

Lol. You mean with top of the line hardware getting around 40 FPS as soon as you hop in multiplayer? You can turn all settings down and your FPS won't change a whole lot. Talking about graphic settings when it comes to performance never makes a lot of sense for me in Arma. Look at Arma2. The game is years old and with recent high-end hardware you still often get around 40 FPS.

I love the game, but multiplayer performance is a real problem. They give us a huge sandbox, but you're limited a lot in what you can do just from a performance perspective. Lots of crazy syncing and heck, weather doesn't even sync properly yet.

[UltimateBawb 1]

Lol. You mean with top of the line hardware getting around 40 FPS as soon as you hop in multiplayer? You can turn all settings down and your FPS won't change a whole lot. Talking about graphic settings when it comes to performance never makes a lot of sense for me in Arma. Look at Arma2. The game is years old and with recent high-end hardware you still often get around 40 FPS.

I love the game, but multiplayer performance is a real problem. They give us a huge sandbox, but you're limited a lot in what you can do just from a performance perspective. Lots of crazy syncing and heck, weather doesn't even sync properly yet.

Server FPS is highly dependent on the server hardware. Problem is, most moron admins choose massive missions to run on a spare PC. Horribly made and inefficient missions like wasteland (online daycare) or life (asshat simulator) don't help when you have 30+ players.

[Alabatross 1]

or is the problem that PLAYERS are still expecting to turn up all settings to the Maximum and get 60 FPS? Arma, like FSX (Flightsim 10) was, have ambitions beyond the 'here and now' of computer hardware. They have rendering capability that extends beyond current hardware and instead of players just pulling their sliders back to accommodate their particular set up, people expect to be able to go all out.

Arma 3 doesn't look revolutionary enough to warrant this. This isn't the crysis of 2013 like crysis was back in 2007.

[quadrono 10]

Server FPS is highly dependent on the server hardware. Problem is, most moron admins choose massive missions to run on a spare PC. Horribly made and inefficient missions like wasteland (online daycare) or life (asshat simulator) don't help when you have 30+ players.

Bawb bud, I have read a lot of your posts and you seem to be under the mistaken idea that the way YOU like to play Arma is THE way to play, it's not. "Real" Arma is not coop. It's whatever the individual player wants it to be for him, that's the beauty of the game. You can make it what you want it to be. If you can't hack playing against humans and want to do your little carebear coop (see what I did there, I tossed out a thinly veiled insult at your gaming style, just like you do) that's up to you, but don't you dare try and say those of us who's tastes don't fit in your little box are somehow marginalized or illegitimate.

[roshnak 41]

Pretty sure he's referring to the way the missions are designed, in that they are very inefficient and run a ton of poorly optimized scripts.

About graphics settings: It's been pretty much proven that on most computers, adjusting graphics options gives only a small boost to performance. I think most of us realize that Arma is very CPU limited right now.

[roni 11]

Real problem with CPU usage is that there is no difference in CPU usage when you put one soldier on map, and when you put 50vs50 AI. I mean it is clearly software-related.

For now this game don't use even 10% of real CPU computing power. It doesn't matter that you see cpu usage 40-70%. It is all about how the game uses this 40-70%...

[henri 10]

;2526385']It is NOT about server fps. It is about syncing data between server and client (most likely).

try this...

two people in MP

one low spec and one high spec

so they are both in the same chopper

then the high spec in 3rd person

sets his viewdistance to from 500 to max

and on the low spec machine

warp/desyc/teleport

and for everyone temporary a dip in fps.

it would be interesting to test mp with everyone on a 1000 max viewdistance and experiment with different viewdistance setups.

to see to what extent updating building state etc. to lowend clients like PvPscene to see how syncing data between server and clients with different viewsettings affects server fps. Maybe the client low end clients cause a wait for reply?