ARMA2 performance on 146 cpu!

[StefX 0]

ARMA2 1.03 (not 1.05? :confused: ) PERFORMANCE ON 146 CPU! From the last Intel Cor i7-980x to the pentium D 805..

For AMD is from A64 X2 6000 65n 512 to the A64 X2 3800 90n 512

this test bench has been done with the hardware base of :

- GeForce GTX 280 + GeForce 190.62

- Raptor 74 Go + Raptor 150 Go

- Creative Audigy

- Windows 7 64 bits

ARMA 2 settings:

resolution: 800*600 (to avoid graphic card bottleneck)

everything to the MAX

View distance : 10km

HT off for I7

enjoy:

link to review:

http://www.behardware.com/articles/778-1/giant-roundup-146-intel-and-amd-processors.html

Edited March 31, 2010 by StefX

[11aTony 0]

Im sorry but enjoy what?

[StefX 0]

Im sorry but enjoy what?

you were too fast I did not have time to upload the image...:j:

[Quincey 10]

Very helpful chart. How come HT was disabled for the i7 processors?

Edited March 31, 2010 by Quincey

[Pulverizer 1]

$1000 CPU for 23 frames per second :D

I wonder what the test was...

[Leon86 13]

I've seen something like this before only then it was 1.01 version. Now to find one with 1.05 :)

18 cpu's 1.01

good thing to see the i5-750 so high in the list, I recommend it to everyone in the "what cpu to get etc. topic"

It also shows the gulftown doesn't do anything for arma2, the only thing that really matters for i7's (and the i5-750) is clockspeed.

Edited March 31, 2010 by Leon86

[Quincey 10]

It seems rather strange that the gulftown doesn't perform that much better. Nothing like other charts, anyway.

[galzohar 31]

resolution: 800*600 (to avoid graphic card bottleneck)

Of course this means the results of this test are useless. If you actually played on a real resolution then your graphics card would obviously the bottleneck and thus the CPU would really not matter much at all, especially if it's any kind of i5/i7 quality, as long as you don't run hundreds of AI soldiers of course.

HT is useless for Arma 2 since Arma 2 does not run enough threads to utilize it and you will get better results with it turned off. In fact the only real way for Arma 2 to give an i7 a hard time is by having lots of AI running around, and AI only use 1 thread, so again it would run faster with HT off in that situation.

[Leon86 13]

resolution: 800*600 (to avoid graphic card bottleneck)

Of course this means the results of this test are useless.

No it doesn't, if you buy 2 5970's and put them in quadfire there's no way the gpu will be the bottleneck. besides, maybe the gpu wouldn't be a bottleneck at 1920x1200, no way of knowing unless you run all the tests agian. There's a lot of people who prefer viewdistance over shaders, anti-aliasing and anisotropic filtering.

Edited March 31, 2010 by Leon86

[Adobe.F 10]

:D i7$20,I will buy it.

[SteveJA 12]

Great looks like the PC im gona buy with the i5 750 is a good buy, id say its a good compromise between price and performance.

[walker 0]

Hi all

Always useful to read the original text 10km view distance!

New to our test protocol, Arma 2 is configured with all settings at a max including max visibility (10 km), which brings the configurations to their knees. Resolution stays at 800x600 to avoid the graphics card impacting on performance. To gauge performance we measure the framerate during a well-defined movement after having loaded a saved game. Note that you must disactivate HT or force the affinity for this game, as you’ll lose 15 to 20% of fps otherwise and it will suffer from rather disagreeable lags.

http://www.behardware.com/articles/778-11/giant-roundup-146-intel-and-amd-processors.htmlAs always follow the link for the original context

I wonder how many entities because that is the real CPU killer. Set this test up to 1000 entities, two sides of a 500, as all arms battle groups, with groups set to guard waypoints and a pair of guarded by triggers to induce a meeting engagement and you will probably have to halve the view distance to get similar performance.

As other have said I would be interested in this test performed with 1.05 of ArmA II.

Kind Regards walker

[Leon86 13]

You can see here that the processors with higher fsb speeds perform a bit better. If you overclock see if you can get a high fsb speed with a bit lower multiplier.

I run my q6600 at 400x8. 425 is still stable but 433 won't even post unless I start doing nasty things in the bios, like cpu pll voltage.

[galzohar 31]

If you buy 2 5970s and have any i5/i7-level CPU you will most likely run this game 100% maxed out with very very high FPS, and a better CPU or overclocking will basically do nothing except increase the number of AI you can handle (as high number of AI seems to be the only way to load up such CPUs). That is, assuming you're not trying to get hundreds of FPS and consider 60 more than satisfactory (and 60 is really more than plenty).

Sure though, if you have 2 5970s then a better CPU might bump your FPS up, but you won't notice it in any way whatsoever without actually displaying the FPS number...

[Leon86 13]

Viewdistance is also an fps killer, and here the cpu is certainly the limit. Doesn't matter if you buy a gulftown and clock it up to 5 Ghz (which you can do with a good watercooling setup I believe) With 10km viewdistance and high-detail models and terrain you wont get more than 35-ish framerates, although it'll certainly be a sight to behold.

edit: of course this is all largely speculation, I just tested something with arma2, if I lower 3D resolution the trees on the horizon disappear. So when you lower resolution there is also less load on the cpu because some objects dont have to be included in the render thread. A better way might have been if they tested it at a normal resolution and then check if it was gpu limited by putting antialiasing on. I bet the game isn't gpu limited at 1920x1200 if you put postprocessing, anisotropic filtering and anti-aliasing off on a fast gpu like an 5870.

Edited March 31, 2010 by Leon86

[galzohar 31]

Basically if you're running the game at settings where it's not GPU limited you should either increase settings (or you've got a powerhouse overkill GPU and are already at max settings).

Also 10k VD is not *that* heavy. With 10k objects only draw to something in the area of 3000m. When I change VD from 3500 to 10k (with rest of the graphic settings as usual - AA/AF maxed, PP off and rest HIGH) I do get an FPS drop but not a massive game-stops-moving kind of drop, while it should be almost 9X more strain on the CPU/GPU if it really rendered almost 3X the distance, but it's not even close to that as the game doesn't scale the rendering distance on objects nearly as much, and the low detail terrain that is shown up to 10k probably doesn't require all *that* much.

Overall, I'd much rather see benchmarks test things at the settings at which you would play with such a system, as that's eventually what people truly care about.

[NeMeSiS 11]

Overall, I'd much rather see benchmarks test things at the settings at which you would play with such a system, as that's eventually what people truly care about.

I dont, i just need a new CPU* for this game, and this helps me decide whats best by eliminating all factors.

(Though personally i would have set viewdistance lower and just dropped more AI's in the game. :p )

*And thus in my case mobo and ram as well.

[MavericK96 0]

Also 10k VD is not *that* heavy. With 10k objects only draw to something in the area of 3000m. When I change VD from 3500 to 10k (with rest of the graphic settings as usual - AA/AF maxed, PP off and rest HIGH) I do get an FPS drop but not a massive game-stops-moving kind of drop, while it should be almost 9X more strain on the CPU/GPU if it really rendered almost 3X the distance, but it's not even close to that as the game doesn't scale the rendering distance on objects nearly as much, and the low detail terrain that is shown up to 10k probably doesn't require all *that* much.

I don't know about anyone else, but going up to 10k VD, I don't really get a framerate drop per se, but I do get MASSIVE hitching/freezes as I rotate the view around. Like, on the order of a second or two per instance. Additionally, if I move forward a bit and/or zoom in and out again, I will get the same massive hitching. Obviously this is due to having to load and then dump TONS of models/AI/whatever over and over in an instant, but it's certainly not playable this way, regardless of the framerate when it's not hitching.

[galzohar 31]

Yeah, makes me really want to get this game running with a ramdisk (look up the appropriate thread, I forgot where that was).

[MavericK96 0]

Sadly, I have been considering buying another 6 GB of RAM to do the same. :P

Oh, the things we do for games...

[Quincey 10]

resolution: 800*600 (to avoid graphic card bottleneck)

HT is useless for Arma 2 since Arma 2 does not run enough threads to utilize it and you will get better results with it turned off. In fact the only real way for Arma 2 to give an i7 a hard time is by having lots of AI running around, and AI only use 1 thread, so again it would run faster with HT off in that situation.

I had no idea, but I'll remember to disable it when I'm playing ArmA2. Thank you.