aop

IMHO best solution would be not to buy their products in future. I lost all respect I ever had for BIS because of ArmA 3 and DayZ. Neither has improved the #1 problem with their games; piss poor performance. They gained large following with ArmA 2 and Dayz mod and choose to blatantly exploit it and sell subpar products with no intent to fix them. BTW. If someone says SSD improves the performance of the game don't fall for it. I have Intel 910 400GB PCI-E SSD which is significantly faster than any SATA SSD and there is ~0 difference in performance of ArmA 3 when compared to using plain regular HDD. Loading screens are like 10 times faster but that's it. No FPS increases since the game is so CPU limited. Now the same snake oil marketers are trying to push ramdisk as solution to shitty ArmA 3 performance...

http://koti.mbnet.fi/aop13/arma3notsim.png

Or the fanboy brigade could admit this game is considered a simulation and it is missing some pretty important features.

Are there any huge performance improvements expected in near future? The problem with this game is that no such hardware exists that could run it even at 30fps in almost all situations. No matter how much money you have, you just can't even buy your way to smooth fps. Considering how this game is completely bottlenecked by single-threaded performance it might be that there will not be hardware capable of getting smooth 60fps out of ArmA3 during this decade. The reason why people are asking you devs about multicore support is not because they would like their quad or octa core CPUs to be used 100% all the time but because they can see that the game is bottlenecked by one or two cores and it runs so bad there might never be hardware capable of running it smooth so going parallel seems to be the only option. Problem seems to as you devs have said many times that it's much easier said than done and perhaps even impossible for ArmA series. It makes the future of this game look very bleak IMHO, I'm pretty sure even less people are willing to put up with 15-30fps when next gen of consoles come out and set the standard to 60fps even for casual players.

If you don't think ArmA 3 is a military simulation, then maybe BIS should stop advertising it as one? http://koti.mbnet.fi/aop13/arma3notsim.png (357 kB)

It could use a thousand threads but it wouldn't matter if one of them was bottlenecking the entire engine and every other thread was waiting that one thread all the time.

Only if you use AA/ATOC. Those two features are GPU intensive but going to more powerful video card is not going to fix the low fps dips caused by CPU bottleneck.

This game is incredible CPU dependant. I just switched from GTX570 to GTX680 and on the same settings I get the same fps in this little test scenario: http://koti.mbnet.fi/aop13/bench1.zip I don't move during the test, just press benchmark key in FRAPS and it logs the fps for 1 minute. Here's the results: GTX570 Min: 28 Avg: 31.050 Max: 34 GTX680 Min: 28 Avg: 31.450 Max: 36 edit: Other components used: Core i5 2500K@4.3GHz 8GB 1600MHz DDR3 AsRock Z68 Extreme 4 Kingston V200+ 120GB SSD

That is exactly what happens when the engine is bottlenecked by one CPU thread. Everything else stalls when the engine is waiting that thread to complete it's tasks.

Not true, 1600MHz and 2500MHz DDR3 have exactly the same performance in ArmA 3: Note the completely linear scaling with CPU clock. 26% CPU clock increase gives 26% framerate increase. This combined low CPU utilization points to the game having one extremely CPU intensive thread that is stalling the entire engine and GPU and other threads are waiting for that one thread to complete it's tasks before a frame can be rendered resulting in low framerate and low GPU/CPU utilization %.

Do you do voodoo magic too? Because on modern PC your advices do absolutely nothing. Maybe 1-2% percent performance gain max. We are not running 200MHz Pentiums anymore. For example on my rig background processes take 0-1% of available CPU resources when the CPU is in idle state (1.6GHz clocks). When I start gaming the CPU clock goes to 4.3GHz. Background processes have no impact on my rig. Some people really want to defend companies making games by shifting the blame on customers and their PCs.

30fps average is horrible. 30fps minimum is passable, but I really prefer 40fps+.

Lot's of text but no solution to the problems discussed in this thread. Try spawning some BLUFOR and OPFOR squads in Agia Marina or go play 40-60 player TDM and come back with your fantastic optimization results. ArmA 3 runs like crap. It's currently one of the worst running games bar some flight sims like X-Plane 10 or Prepar3D (atleast in them you can get good performance by lowering the graphics settings a bit). To get smooth 40-50 fps in ArmA 3 you need to have something like 5GHz+ Intel CPU. My i5 2500K@4.3GHz isn't enough.

Usage and allocation are two different things. Many games these days reserve almost all the available VRAM even if they are not going to use it all. Battlefield 3 does the same thing, if you have 1GB VRAM it shows almost 1GB VRAM usage in monitor software and if you have 2GB it shows almost 2GB usage.

Most likely mission specific change, since this setup provides exactly the same fps as in the last build: http://cloud-2.steampowered.com/ugc/882976114553809006/3CFBA4DB4C83B2CD576067036DF13442E7B69A3F/

I have done some testing in both ArmA 2 and ArmA 3 editor using Chernarus map and it seems the main culprit behind performance issues in ArmA 3 is the AI. How to prepare ArmA 3 for testing: 1. Copy ArmA 2 content for ArmA 3 by creating folder @ArmA2 in ArmA 3 folder and copy the AddOns folder of your ArmA 2 installation to the @ArmA2 folder you created. 2. Enable mod @ArmA2 in ArmA 3 Options ---> Expansions menu. How I tested it: First test: 1. Have ArmA 2 and ArmA 3 run at same or similar graphics settings. 2. Have FRAPS or other fps display overlay running. 3. Create missions in both ArmA 2 and ArmA 3 with only your character and some vehicle in Chernogorsk, I used the dock as starting location. 4. In this mission drive around Chernogorsk with your vehicle and watch the framerate displayed in the counter (you can record the framerate to .txt file in FRAPS) This test resulted very similar framerate in ArmA 2 and ArmA 3. Second test: 1. Use same settings in this test as the first one. 2. Make mission with 5 BLUFOR and 5 OPFOR infantry squads in Chernogorsk. 3. Observe as civilian. Now this test had huge framerate difference between ArmA 2 and ArmA 3. In ArmA 2 the framerate hovered around 30-45 with 120 soldiers killing each other in Chernogorsk but in ArmA 3 the framerate never went above 20, mostly staying at 15, sometimes dipping down to 10. My rig: Core I5-2500K@4.3GHz 8GB DDR3 1600MHz GTX570 ArmA 3 installed on SSD ArmA 2 installed on HDD Conclusions: Number of AI characters present on map has much higher impact on framerate than in ArmA 2. Currently having 100 active characters on map will bring the performance down to unplayable levels in ArmA 3.

Three squads is 30 infantry, each squad has 10. ---------- Post added at 10:53 PM ---------- Previous post was at 10:47 PM ---------- 15 HMG cars on both sides (30 AI on both sides, 2 in each car): Min fps: 24 Avg fps: 29.417 Max fps: 33

Same test with 2 HMG cars for both sides: Min fps: 28 Avg fps: 32.367 Max fps: 40

Here is another quick test with FRAPS, 1 minute benchmark. Watching Agia Marina from Southern hill, no AI: Min fps: 58 Avg fps: 62.517 Max fps: 64 Exactly the same place, 3 OPFOR and 3 BLUFOR squads in town: Min fps: 30 Avg fps: 33.600 Max fps: 38 Seems like AI is eating all my fps to me. edit: Mission setup: http://cloud-2.steampowered.com/ugc/882976114553809006/3CFBA4DB4C83B2CD576067036DF13442E7B69A3F/ For first test I just had no AI.

You are not helping anyone. You refuse to run the test described in the first post and yet you state that the AI isn't the problem. What is causing so big disparity between ArmA 2 and ArmA 3 performance in same map when multiple characters are involved? Higher polygon models?

So you are saying all the testing in the first post is wrong without doing the test yourself? edit: And your test sucks anyways, 20 AIs is really small number and you spawned them on airfield which means they are not going to use much resources for pathfinding like spawning them in city would do.

How about you run the test in the first post and come back with your results?

Here is some testing I did with both ArmA 2 and 3 using Chernarus map: http://forums.bistudio.com/showthread.php?148194-AI-kills-the-performance-completely-(instructions-how-to-test-inside) AI seems to be the main culprit behind piss poor performance.

I don't remember if it was some mod I used or if it was a feature in original Operation Flashpoint but I loved the ability to put live satchel charges inside vehicles. It could be used to simulate car bombs or just for fun in multiplayer, just rig a vehicle with satchel charge and wait for someone to take it and blow it up after he has started driving it. Will this feature be in ArmA 3? It would be great addition to all the mines ArmA 3 already has.

Seems like you didn't even bother to read the first post or give it any thought. The problem is that currently ArmA 3 can't handle anywhere near the same amounts of AI as ArmA 2.