Content without Performance isn't playable.

[HKFlash 9]

TL:DR:

 

ARMA 3 performance is a mess. RV engine is a mess. Each patch its band-aid fixes. It's time BI sits down and crunches the RV engine for CPU performance. Don't even get me started on GPU usage. But that goes way deeper than the standard A3 users know of RV engine...

 

BI gets a pass and excused greatly sometimes on these topics, because at end of the day, we all love and adore ARMA, and it gives us endless pleasures. But, everyone has a limit. For some, the limit / line is being crossed now. BI can continue to ignore it too sadly. The sales and hype around A3, and its "Life" missions alone, make sales out the window. Sales to A2 compared to A3 is insane. It went from selling only to Military Enthusiast and Milsim, to DayZ kids, to Life kids now.. Why fix something when 90% of userbase doesn't care and only plays Wasteland, Exile or Breakingpoint? Flashback's of other devs flashing in my eyes.....

 

/love you BI, but its' time you wake up and smell the flowers and give RV engine some love and actually crunch with it for weeks. Esp when you now charging $30 for a DLC, when base game is $9 more..... Yea, it's time you care about Performance more. 

 

Your arrogance is not helping you neither is your ignorance. BI is already developing a new engine, code named Enfusion.

[jw custom 56]

"Go buy a $500 CPU" ... not my kind of advice for most of the Arma3 players, but for a player spending all those bucks on an "i7 5820K/Titan X 12GB/32GB DDR4" based rig ... why not?

 

:D  i like it :D

[kilrbe3 37]

Your arrogance is not helping you neither is your ignorance. BI is already developing a new engine, code named Enfusion.

Your arrogance is not helping you, neither is your ignorance.

 

Enfusion isn't being made for A3 or ported to A3. So A4. So unless real changes and performance shows, and BI adpats Enf to ArmA series, is still yet to be known. For now at least.

 

Please, be more of a rude boy next time. I love throwing things back in people's faces ;) 

 

Go buy a beer and relax.

[I give up 152]

Honestly, the performance is not a issue (or issues).

The issue  is some guy that made a trip to a game studio and now he see himself  as the master guru of the sound and consequently the sound of A3 is totally destroyed.

The issue is some guy that probably lost is glasses in some war, and now he cant see how the things should look, in matter of graphics.

[HKFlash 9]

Your arrogance is not helping you, neither is your ignorance.

 

Enfusion isn't being made for A3 or ported to A3. So A4. So unless real changes and performance shows, and BI adpats Enf to ArmA series, is still yet to be known. For now at least.

 

Please, be more of a rude boy next time. I love throwing things back in people's faces ;)

 

Go buy a beer and relax.

 

Sure BI will change A3 engine after releasing the game just for you. They even have a record of that right?

 

Your words touched my heart and mind to the point I don't need a beer to relax. But thanks for the suggestion.

 

Also, if you love throwing things back at people's faces you won't stay around for long.  :)

[R0adki11 3949]

Ok that's enough, any more flame-baiting or throwing insults at each other will result in users having their forum access suspended. 

[req 10]

I have a pretty damn good rig and even I get some performance issues here and there...

[chrisb 196]

 

 

 

 

@oldbear,

 

I'm calling 100% bs on those pics or they are doctored. ARMA DOES NOT use 80-100% CPU usage, unless you have other task open in background also using your CPU. ARMA 3 at MAX will use 70% MAX, usually 50-65%. Your CPU clock is almost stock, and you complain, yet don't OC either? Shake my head...

 

 

 

 

Test pics from some time back.

 

http://i1122.photobucket.com/albums/l538/triopalite/A3/arma3_2015_05_16_15_52_33_724_zps0rsiz0bl.jpg

http://i1122.photobucket.com/albums/l538/triopalite/A3/arma3_2015_05_16_15_54_56_794_zps9djto2hs.jpg

http://i1122.photobucket.com/albums/l538/triopalite/A3/arma3_2015_05_16_15_57_30_078_zpsicgvnxjj.jpg

http://i1122.photobucket.com/albums/l538/triopalite/A3/arma3_2015_05_16_11_00_03_000_zps9s8oyyw1.jpg

http://i1122.photobucket.com/albums/l538/triopalite/A3/arma3_2015_05_16_10_00_22_836_zpsucrsgbbz.jpg

http://i1122.photobucket.com/albums/l538/triopalite/A3/arma3_2015_05_16_10_00_37_735_zpswikdvbox.jpg

http://i1122.photobucket.com/albums/l538/triopalite/A3/arma3_2015_05_16_10_02_58_473_zpscvqjgcxo.jpg

[skidmatic 1]

Haven't all the ARMA games been bad when it comes to performance?

[Vigil Vindex 64]

Still waiting on an official benchmark like in A2. I think it's interesting BIS was confident enough to have benchmark in previous games but not in A3. And how many years has Dwarden been pushing out performance binaries now? What is the results of these performance binaries? Are they in the stable branch yet or are we still testing? I think a benchmark would help give us a consistent test to gain insight to how useful the performance binaries are and how to move forward.

[CANADAVE 14]

How sad...

 

just makes it even more sad that a dual core from literally 10 years ago, can beat a quad core of modern day... Gee, let me go pull out my trusty E8400 from 2007 and put it at 4.5Ghz , just to get best out of ARMA! While I suffer in everything else that has moved on with the times and actual has common sense when comes to coding for newer processors! /s

 

how sad BIS, how f'ing sad.

 

So I can play arma 3 today for the cost of 10 year old hardware?   I don't need expensive new hardware?   I'm just confirming before I start to build my son a machine .   I don't know the point of this thread  but are you saying this is a bad thing?   Seems awesome for me and by bank account!  thanks for the tip!

[kocrachon 2]

We need evidence and benchmarks to diagnose the issue, regardless of all of the arguments here. He says he has poor performance and displayed specs. We need a good 2 minute recording of him doing simple things like idling, running, running and shooting, and surrounded by targets. If we see all of this in a single player environment and no issues, we can blame Multiplayer. But without any benchmarks or evidence, theres not much support to be provided. 

[empleat 1]

There's this amazing FPS improving tweak called "don't play badly scripted missions on shit servers administrated by kids" they you should try. Why is it that every time I see a ludicrously expensive rig, the guy using it knows absolutely nothing about how computer games actually work? It doesn't matter how amazing your PC is if the server is the thing lagging, and Bohemia can't do anything about serverside issues.   

It is not true, that it is only serversided.

I have intel 3770 with geforce 780.

Arma is using max 50% of gpu and 30% of cpu.

And there aren't any good servers for tdm and even battleroyale.

60fps is minimum for good gameplay.

Game engine is bottlenecking cpu hard.

[jw custom 56]

60fps is minimum for good gameplay.

 

LOL...

 

30fps+ and ArmA is awesome.

 

...and to those with high end computers complaining about performance issues... atleast set SAMPLING at max 100% before crying here <_<

[calin_banc 19]

Why 30+ when 24 is more cinematic? 

[calin_banc 19]

@oldbear,

 

you could play tactical if the AI were better suited for that and even then you need good, steady fps. I was ironic about the 24fps, since some see movies as proof that low fps is fine.

 

Low input lag, predictable experience is what you need when you do an action that requires finesse and shooting at someone is definitely such an action. I can play an RPG no problem at 30fps, if it doesn't involve precise moving or aiming, just clicking in a certain direction, but not here. 

 

With that said, I do appreciate that the latest build (on dev branch) does show a nice boost. It's not where it should be and it still doesn't use the available resources as it should, but at least it runs better.

[Arkensor 96]

To express my point of view, I just will say that I start overclocking my pc this afternoon.

 

On ~ 230 players active on our server, in kavalla where many things happen and alot of people are present I get around 60-70 FPS - On Altis

Same amount of players, but far out in the slat lake or other less populated ares I get up to 90 fps - On Altis

 

Now on Tanoa IN SINGLEPLAYER / EDITOR my fps do not rise above 53 fps, at no given time.

If this framerate would be constant in MP that would be fine, but witgh 200+ players in MP the frames will drop from ~50 to ~35/30, which is close not not playable.

 

There are too much objects on the map. Every 50 meters a garbage can, man road signs, and a railway which is not usable in any way which alone makes alot of obejcts ...

 

BI need to either reduce the detail amount, that is still looks good, and is playable, or needs to give everyone a 1500$ 5,X ghz cpu with custom water cooling, to run Tanoa the same way they do run Altis right now.

 

All FPS test were made with the same grafics, under the same hardware and software settings. 

 

My pc is build for arma, and the fact that is runs that amount of players in MP with ultra grafics etc, should proove this. Now how can that kind of pc not handle Tanoa?

 

@BI did you internal QA people even test Tanoa on medium and week pcs? What framerate do you as developers consider playable and "good". Are there any plans to get more performace out of the map?

 

 

Thank you for reading

 

 

Regards Arkensor

[chortles 263]

Any advice for settings if I'm on a i5-3570K (not OC'd, Hyper 212+) and 660 Ti?

[Mirek 166]

Why 30+ when 24 is more cinematic? 

I have some  fluctuations in my FPS. Normaly i run 50+  sometimes even 60+ i drop to 45 in cities and sometimes below 25 in heavier fights. I only notice it when i look at the FPS counter. Never noticed it before i enabled the counter. So i gues there is no visible difference. 

[Vigil Vindex 64]

We need evidence and benchmarks to diagnose the issue

 

I really hope BIS get around to making an official benchmark like A2+OA. It will really help with troubleshooting and testing if BIS provide an officially supported and consistent test that we can use to give meaningful feedback and data. DxDiag.txt and a text dump from the benchmark of the results and graphics settings would be really great for submitting to threads like this, and probably help define the minimum and recommended hardware requirements in reality.

[brightcandle 114]

I have no doubt Tanoa is slower, up to 30% slower at times and its heavier on the CPU than Altis is. I did a comparison video a week back which shows when the game is CPU and GPU limited and also the difference in performance between the two terrains is quite dramatic:

 

[I give up 152]

There is no surprise.

Foliage always have been the gpu killer with arma 3. Just for instance, the gpu has more usage in Stratis forest than in Kavala.

About the called cpu bottleneck in Towns, here I disagree.

While in forest what we have is heavy foliage, basically there are no textures loaded in to ram, this means that the gpu (and consequently the cpu) do not need to wait for ram, here the graphics are rendered on the fly.

In Towns we have several Gigabytes (textures) loaded in to ram, being refereshed, being flushed, loaded again and repeat this in to infinite. This takes time and since the cpu is considerably faster than whatever ram,  the cpu ( and consequently the gpu) has to wait for ram, that's why the operation of these 2 pieces of hardware (cpu and gpu) is limited.

There you have your bottleneck. 

Memory management.

[PuFu 4600]

 

There is no surprise.

Foliage always have been the gpu killer with arma 3. Just for instance, the gpu has more usage in Stratis forest than in Kavala.

About the called cpu bottleneck in Towns, here I disagree.

While in forest what we have is heavy foliage, basically there are no textures loaded in to ram, this means that the gpu (and consequently the cpu) do not need to wait for ram, here the graphics are rendered on the fly.

In Towns we have several Gigabytes (textures) loaded in to ram, being refereshed, being flushed, loaded again and repeat this in to infinite. This takes time and since the cpu is considerably faster than whatever ram,  the cpu ( and consequently the gpu) has to wait for ram, that's why the operation of these 2 pieces of hardware (cpu and gpu) is limited.

There you have your bottleneck. 

Memory management.

the above is partly true. The issue is that a lot of the textures are loaded directly from the HDD in a continuous manner. There is little buffering going on for textures, hence there is really no reason to have A3 in 64bit (allowing the use of my 16 or 24gb respectively ram available). this is also the reason why a SSD will improve not only the stuttering but also FPS, more than any other game out there

[I give up 152]

Textures are buffered frm HDD when we have pagefile enabled (which is also memory). With pagefile disabled its all about ram.

HDD is only used when we load the island for the first time.

 

Edit,

 

About 64 bit, not sure if would do any better. Basically we already have it with "32 bit breaking barrier", still and probably because with 64 bit would have to be the executable to handle with memory load and management, some improvement could come, but would not solve the thing. Arma due to its architecture will always have this situation, in my opinion.

 

In fact was due to heavy load on memory (and the consequences of it) the reason for "32 bit breaking barrier". 

The thing is with Arma 2 and  because DirectX 9 the textures were considerably "smaller" and consequently the loading, flush, refresh, processing timings, etc, also were considerably smaller, so there was no issue.

With Arma 3 and because DirectX 11, these become "huge" requering a large space on memory (either pagefile or ram) and consequently the tinings for its management are largely increased and there is no way for a management under acceptable timings in a way that has no impact cpu and gpu operations.

 

If you want to see the difference between in memory load between A2 and A3, just use "cup" (or similar) and load Takistan (or Chernarus) using Arma 3, check memory (pagefile or ram) load and usage.

Then load Takistan (or Chernarus) using Arma 2 and  check again.

You will see that Arma 3 (under the same scenario) will double the memory usage.