Serious unexplainable Performance Differences - got proof

[Helmut_AUT 0]

Trooper, the wall is not made of continuous polygons. That is a single large flat house wall (likely a flat polygon with a texture), not a fence with holes in it. There's no question that such an object blocks view 100%.

So you are happy if the engine renders "most objects in front of you all the time" even if that means a crappy framerate? Why then, tell me, can other games do graphics as good with 50% more FPS?

Maybe because they are not rendering stuff you can't see, could that be it? Oh, and you forgot to explain why looking at the ground on a stretch of road (absolutly NOTHING in your view then, not even a blade of grass) doesn't significantly push framerate.

And no, until we hear from the devs about the next patches I'm certainly not going to shut up and pretend there's nothing to report. I have a rig that is well within the recommended specifications (not minimum - recommended specs...), I think can expect a game that gets playable framerate, and not 15 FPS slowdowns even at low detail settings with a handfull of soldiers around.

If that sounds harsh, then because it's frustrating that they created a most excellent game (core system, AI, editor, functionality) but slapped an engine on top that makes it painfull to actually play it.

[Steinfisch 10]

They not only have to draw the scene every frame. The physics, the scripts and KI calculations also depend on the units which are inside the view frustum, if they are friendly or not. If you change the view direction your sight of enemies change too and this leads to a range of actions. I don't know if they update these things every frame, but it has influence.

Make a little test: build a mission (editor) with only you as actor and no scripts. Add this textured object above and test the framerate. This would give a more realistic impression on what the graphic engine performs and what are other influences.

Edited June 10, 2009 by Steinfisch

[dvolk 10]

You won't spend much time looking at walls in the game, so what does it matter how high FPS go when you do?

BIS could modify their algorithms to get you 1000fps when you're looking at a wall, but if it dropped overall performance by 1fps, it wouldn't be worth it.

[Helmut_AUT 0]

Steinfisch, my initial post already states that this is an empty editor map with only the player on it.

---------- Post added at 11:01 AM ---------- Previous post was at 10:58 AM ----------

You won't spend much time looking at walls in the game, so what does it matter how high FPS go when you do?

BIS could modify their algorithms to get you 1000fps when you're looking at a wall, but if it dropped overall performance by 1fps, it wouldn't be worth it.

When you are moving into a village (where performance drops massivley) you will by definition often move along a wall or between houses. If houses block the rendering of objects behind them then that would offset the increased performance cost for the higher LOD on the stuff you actually CAN see. Right now, as you get closer to buildings, their increased LOD adds rendering strain, but the decreased object count in your view doesn't offset this.

Sorry, but this is a very relevant test. As is the "look at the earth" test, since such things can be used to see how much object density in player view actually influence FPS.

[Steinfisch 10]

Of course you are right. They could optimize their code.

[Helmut_AUT 0]

Maybe the reason for not culling objects out of view is indeed that it would slow down the CPU too much, in which case I get why they keep objects rendered.

Wouldn't it be sweet if the 1.02 patch brings FSAA and a 20% performance increase? Arma's 1.09 patch gave some people 40%.

[IronTrooper 0]

I guess they could optimize this a little but as dvolk pointed out, it wouldn't really matter as you don't look right into walls or ground most of the time.

[Helmut_AUT 0]

Is the concept really that hard to grasp? If you are inside a town or village, you are looking towards walls ALL of the time. Not towards ONE specific wall, but you are basically encircled by walls, and if the engine renders stuff behind those walls it's absolutely pointless for gameplay, bad for performance.

And if the engine always renders an overhead even when NOTHING is in your view (ground-look) then that matters EVERYWHERE on the map.

Are you guys really not seeing the purpose behind those tests, or are you purposely obtuse since you feel you need to defend BIS Honor or somefink? Please explain to me why other games do this kind of culling, improve framerate considerable, but Arma2 somehow shouldn't?

[dale0404 5]

Gents, lets not turn this into a slanging match please.

People have their own opinions and ideas about the graphics engine.

What I would ask is to post some testing results that you have done yourself. Lets hope that BIS are taking notice of what we have discovered here.

[SANGEKi 10]

Maybe the reason for not culling objects out of view is indeed that it would slow down the CPU too much, in which case I get why they keep objects rendered.

Wouldn't it be sweet if the 1.02 patch brings FSAA and a 20% performance increase? Arma's 1.09 patch gave some people 40%.

The game is not that CPU bound anyway.

There is hardly a real world difference in performance between a Core 2 Quad at 3GHz and a Core i7 at 4.2GHZ.

Even a moderately complex culling algorithm won't slow the game down much or at all.

Even if it would make the game slower on slower CPUs it wouldn't matter because no one who owns the GPU(s) required to run this game with acceptable performance on high settings is still on a Pentium 4.

The devs (as dedicated as they might be) don't have the money/time and/or knowledge necessary to really (if even still possible) optimize the engine, let alone create a new one.

While I'm sure that future patches will indeed improve performance, it will never run GREAT.

Gents, lets not turn this into a slanging match please.

People have their own opinions and ideas about the graphics engine.

What I would ask is to post some testing results that you have done yourself. Lets hope that BIS are taking notice of what we have discovered here.

Thing is that this is not about opinions.

The engine is doing a lot more work than necessary.

This game is slightly more complex than your regular AAA shooter no doubt but most of that complexity has it's place in the CPU.

This problem here, at it's core, is a different matter.

Edited June 10, 2009 by SANGEKi

[Helmut_AUT 0]

Sorry, didn't want to "raise my voice" here, and don't want a flame war, but it would be nice if people didn't respond with "you're not gonna look at a wall" type of postings, since I think other and I have already explained in detail what these tests are supposed to show.

Sangeki, I have an AMD 5000+, and the Warfare mission on Chenarus throws me to 16FPS, which are CPU limited. I can't agree that the game isn't CPU bound, and the 5000+ is actually above recommended specs.

[SANGEKi 10]

No, the game is obviously more CPU bound than "normal" games.

That comes with the territory of having so much possible viewing distance and AI but my comment was only in relation to the required GPU performance and the question if a culling algorithm would possibly slow the game down to much.

This is not FSX where GPU power just stops making any difference at some point and it's only CPU, CPU, CPU.

Edited June 10, 2009 by SANGEKi

[Leopardi 0]

The engine obviously has a long way to go in optimisations. Another test, I look through the forrest and spin I, get say 30 fps.. I look at the floor and get 30fps, i look at the sky and get 30fps.. Its like I am limited to 30fps no matter what I do. Clearly there are engine problems, and hopefully BIS is working hard to correct them. Unfortunately I am sure BIS does not have the resources at their disposal that a big studio has so will have to be patient. But you never know!

thats called cpu bottleneck which it seems you have.

[dmarkwick 261]

It's most likely to be related to transparency code. There's the possibility that the wall might have a transparent texture on it, like a window, and so the engine needs to know what's behind it.

[Helmut_AUT 0]

I will check that, but as far as I know that house has no "see trough" windows.

[FraG_AU 10]

I don't have this problem. I can look at the ground 30fps, spin between 30-40fps and at the sky up to 70fps. Not to be an ass but maybe your hardware is the limiting factor?

I hope not, at the moment I would say I have a state of the art PC. While I know that each game is created differently, and each engine works differently I would expect this game to perform better then it does currently. My system is

i7 @ 4.2Ghz

6Gb RAM 1600

GTX295 OC'ed

Xfi Xtreme Gamer Fatality ed.. Ie pretty much the best system money can buy (aside from going quad sli). I never had expecations of running at 100FPS etc, but I was hoping to run more then 30-35 FPS @ 1920x1200. My ArmaMark II score is approx 4400-4700 on my detail setting, but in game it just suffers a breakdown.

thats called cpu bottleneck which it seems you have.

If I have a CPU bottleneck I pitty 99% of other users.. Sys specs as above.

Don't get me wrong I appreciate the game, the size etc, but clearly there are plenty of optimisations that need to be done to the engine aside from the mission bugs. I knew when I bought it that there were a lot of bugs, but i thouht performance was optimised for multi core/high end gpus and perform well on said hardware. I know at the moment my SLi is not working properly, so card is really a gtx275. I will loan my mates GTX285 and 4890 to test see which works best and i think the 4890 will beat out my gtx295

Edited June 10, 2009 by FraG_AU

[Helmut_AUT 0]

Frag, what fillrate and settings are you using?

It's a bit shocking that your system suffers frame drop, since I wouldn't know what else short of a cray we could use instead.

[householddog 10]

I had a bit of a play around with this in the editor.

The method I used was to hide behind corner, then strafe into the open slightly. Keeping my orientation the same way.

What I found was interesting. Some buildings' walls do actually raise my fps higher by around 10-15 frames. Other ones do not. Particularly wooden buildings, I believe do not increase your fps when they block your sight. Concrete ones do not, as far as I can tell.

Some sort of penetration/los madness? Not sure yet. Just an early observation. You guys/gals, might want to try it, and see if you get the same results.

[SANGEKi 10]

I also have a Core i7 at 4.2 GHz with HT enabled.

I "only" have GTX280 at the moment and my FPS are OK...ish as long as I stay low with the resolution (not so easy on a 30" monitor) but there is only so much more hardware out there that you can put into your computer.

[maddogx 13]

I also have a Core i7 at 4.2 GHz with HT enabled.

I "only" have GTX280 at the moment and my FPS are OK...ish as long as I stay low with the resolution (not so easy on a 30" monitor) but there is only so much more hardware out there that you can put into your computer.

Can you post a screenshot of your gfx settings screen, including the advanced options?

[SANGEKi 10]

Can you post a screenshot of your gfx settings screen, including the advanced options?

Not really. ;)

I don't have any set settings right now and I don't really play the game at the moment.

As I said the game runs OK as long as I don't exactly run it on 2560x1600.

On everything maxed out, at 1680x1050, fillrate at 100% and visibility at about 4000 I get between 60 and 25 or so on average.

I'm not asking for support, the game runs how it's supposed to on my system.

I'm just not satisfied with the way it actually is supposed to run.

[FraG_AU 10]

Frag, what fillrate and settings are you using?

It's a bit shocking that your system suffers frame drop, since I wouldn't know what else short of a cray we could use instead.

1920x1200

Fill rate 100% (set manually in .cfg)

Settings

Viewdistance = 3.5k

Textures - Very High

Text memory - Very High

AA - High

Terrain/Object/Shadow - Normal

PP - LOW

Now I have played with all settings one by one and FRAPs, doing a little benchmark by running around a set course (not 100% accurate but it gives you and idea)

I run once to load the textures into memory, then record it.

Basically Textures drop my AVG by 1 going from Low to Very High

texture memory, Like wise Anistroptic filtering does not affect my performance at all.

Draw distance is negligable performance diff say going from 2500-4000 may lose 2 fps..

Like I said I am having difficulty in getting SLi going, I am just about to try after cleaning out drivers (I tried 4 drivers to see if any diff).. Going back to 186.08 now.

EDIT:

WOW FINALLY: Got SLi working!! Things I tried before but worked after using driver cleaner and re-installing drivers.. or could be new profile I made using nhancer?

Anyway what i did is clean drivers, install 186.08

I run game with -winxp -cpucount=8

Nhancer profile is copied from Arma 1, but I enabled multi core optimisation and of all things I used AFR-4 (Quad sli) * I have previously tried evry single setting in nhancer and didn't work *

Anyway that fixes one problem, no doubt nvidias next driver set should encompass a native profile for this game that works properly. I know I have been trying for days to get SLi going and when i almost gave up it seems to have kicked in.. I will do more testing and see what was the cause, as I know a lot of others were having SLi woe's, some managed to fix via "crysis.exe" trick, some used -winxp etc..

Edited June 10, 2009 by FraG_AU

[Helmut_AUT 0]

Yep, that's the magical question: Why does changing detail settings have basically zero influence on frames at high resolution.

What I do notice is that people with this kind of problems generally run more than 1280x1024, which leads me to believe it is somewhat shading related. Pixelshading should be the one demand that really scales with resolution, no?

[dmarkwick 261]

I will check that, but as far as I know that house has no "see trough" windows.

It might not matter that there is not a window, the issue might be that the possibility of the wall having a window/transparent texture exists (even if it does not actually exist) and so the geometry behind the wall is at least calculated, if not rendered.

As another test, you might try manually placing 4 buildings all around your position in the editor, with no gap between them. Then do a slow 360 and noting the FPS at each point. My feeling is that stuff that gradually leaves your FOV, or that goes behind you, is culled as irrelevant to transparency concerns.

[Helmut_AUT 0]

Okay I get what you are saying - but then that's still something unnecessary, isn't it?

Just because you could theoretically look trough a hypothetical window in the house doesn't mean they can't cull if there isn't a window.

It goes a bit with above tests - maybe material is not only used for penetration but also for culling, so wood - often seen on fences and other "see-trough stuff" might have different culling than concrete.