MaD_DoG

That's true. This game has to run smooth to be fully apreciated !

Guys, Report status. I just received my new i7 3770k replacing my old i5 760, and I just won 15 fps average ! Now the game is incredibly smooth anytime no matter what's going one, single or multiplayer ! I can even turn the AAX8 on ! I'm f**king happy about that, I knew the cpu was so important. So to give you some comparaison between both, here are the results of the previous tests I did earlier in this topic, but with the new cpu. The test lasts 60 secondes and use 2456km of view distance: cpu | min | avg | max i5 760 | 31 | 40,75 | 53 i7 3770k | 44 | 53,9 | 63 So in addition to give better performance during larger scale battles, changing the cpu also gives higher minimum and average fps. Hope you'll find this useful :)

Guys, I've made an interesting discovery. I did 2 tests with 2 scenarii each. In the first scenario, I was looking at the sea horizon at the end of the map. 1) Alone on the map => Great performance 2) Me+100 idle ennemy troops at the other side of the map => Great performance (even though around 10 fps drop or something) In the second scenario, I was looking to the other side of the map, with a dune in front me, hiding the horizon, so not seeing much thing. 1) Alone on the map, same performance as the first scenario 2) Me+100 idle ennemy troops at the other side of the map => HUGE fps drop (around 20- 30 compared to the first scenario) That means that, fps seem to go down from the moment where you look in the direction of the ennemies, even if you don't see them or have something between you and them. If you look to the opposite of them though, you won't see big fps drop. How the hell is that possible ???

Here are some benchmarks I did on the docks at Agia Marina / Marina Bay to figure out the impact of view distance. I simply ran alone through the town following a simple path with ultra settings, vsync off, AA off except FXAA to Ultra, and trying different view distance with a resolution of 1920*1080 I have an i5 760 @2.8ghz and a hd7950 @950Mhz. Each test lasts 60 secondes. Distance | Min | Avg | Max 500 | 47 | 55.8| 66 2456 | 31 | 40.8| 53 4982 | 23 | 30.2| 43 12000 | 12 |19.5| 27

Sure ! I have a 2.8ghz so getting a 3.5ghz I guess I'll have some better performances :) ---------- Post added at 18:42 ---------- Previous post was at 18:41 ---------- No, the K just means it is open to overclocking.

I see, so cpu and gpu are really bound to each other. But in any case, cpu is prior to gpu since even with the lowest settings, the cpu still has to handle every events and threads (like AI) all around the map because it is a simulation. That's why I'm pretty sure upgrading my cpu is a good idea to have better fps anyway.

directX ... cpu ? You mean gpu right ? View distance has a lot to do with cpu aswell because Arma is using some pre caching stuff when having a long view distance. It's easy to notice it: Beyond some distance, no matter what video settings you change, you won't have any fps differences.

Ok then. Are they planning to make it run on the gpu ? Anyway, I don't mind because I have an AMD card. But that's good to know. I'd even be more accurate about your sayings, I would say "because of the insane number of objects in the world". I did the test, no matter where objects are on the map, since they are somewhere, it will costs cpu. That's specific to "Simulations", unlike other classical games.

Somewhere in the video settings you will find the PIP settings

That's intersting, I will try this out this evening.

I don't think it will run smoothly.. You'd need a i5 or i7 at l east, and a better graphic card. I think it may run ok if there isn't many people on the island. As soon as you will add some troops, it will get choppy due to your cpu.

And what does autodetect says ? For me, the autodetect worked pretty good as they improved it compared to Arma 2. However my cpu is the one to blame. So try with the autodetect and reduce the view distance. The view distance has a huge impact on performance.. I goes from 17 to 60 fps between 12km and 500 meters.. Also, make sure the VSync is off, it has a big impact too. ---------- Post added at 11:11 ---------- Previous post was at 11:06 ---------- As Gliptal says, PhysX (not Physix) works on the GPU only if you have a NVidia. For the single thread performances, just check the link I wrote ... Thing is kinda the same for multi-threading. Of course I'm talking about gaming performances. It's been said that AMD cpus are not really optimized for video gaming, just some benchmarks around. I won't search for the links again so I let you Google it.

Guys, even though AMD cpus have 8 cores and more than 4ghz, they are not as good as any intel. They are pretty bad at single threading so.. Be careful of AMD cpus. Just look at some cpu benchmarks in gaming around and you will see. (for exemple check this link http://www.cpubenchmark.net/singleThread.html and search for the first AMD in the list to see how down the list it is). Try to put the distance view to the minimum and I'm scared you'll have to stick to low details, or try autodetect.

I have the exact same problem with this server. Going down to 25-30 no matter what settings. I'm getting a new cpu, hope there will be some differences. Don't forget that on multiplayer, your performances will depend on the host..

Link ? If you're talking about i5 on Sandy bridge, I disagree: http://arstechnica.com/information-technology/2012/08/inside-the-second-gaming-performance-with-todays-cpus/8/