ArmA II Comprehensive i7 Benchmark XP vs. Win 7

[Scorpio9 10]

When I got ArmA I was really disappointed with the performance on my i7 system so I upgraded the GPU from a EVGA 8800GT to a EVGA GTX 280 SC which helped matters some but I felt I should have seen a much greater improvement so I set down again and started tweaking and I have to tell you Im a self confessed tweaking junkie. I have compiled the results of my tweak session and the results are in.

Using ArmA Mark II I compared the i7 stock clock to a 42% overclock and then Windows XP to Windows 7. As you will see I was able to achieve a AmaA Mark score of 5387.76 with Windows XP and 5059.61 with Windows 7 see how I did it and how the numbers break down vist me at www.armaman.com

[mcvittees 0]

As I've said elsewhere, Vista = Windows 7, hence your results don't suprise me - you'd get the same bench marking results against Vista.

Don't believe me? Have a look at this benchmark text between the two OSs: http://www.tweaktown.com/articles/2869/windows_7_final_oem_vs_windows_vista_vga_performance/index.html

[galzohar 31]

With such a score you should seriously check if you didn't cap your FPS at 60 for a significant portion of the time. If this is the case, then your score is biased towards the weaker system (that is, in reality the difference is bigger than your score shows). Please specify if this is the case or not, and if it is the case, increase graphic settings so that it doesn't happen anymore and repeat the tests.

Alternatively you can also make sure that you can go over 60 FPS, but I don't know if checking whether the game runs at 100 or 80 FPS matters as much as checking how well it runs at high settings. After all, nobody buys an i7 and GTX 280 to play on low.

[Scorpio9 10]

With such a score you should seriously check if you didn't cap your FPS at 60 for a significant portion of the time. If this is the case, then your score is biased towards the weaker system (that is, in reality the difference is bigger than your score shows). Please specify if this is the case or not, and if it is the case, increase graphic settings so that it doesn't happen anymore and repeat the tests.

Hi galzohar I believe ArmA 2 does have a cap at 60 FPS. I haven searched to no avail as of yet to up my FPS cap which i would like to see where the top end limit is for this game is. If you haven any idea please let me know.

I do think you have pointed something out that I should have touched on or stated a little better. The reason for the lower settings is to provide a base line for others to compare their rigs too since most people are having to play at these settings, be it another comparable i7 or some variation of it ie. i7 with a 8800GT, 4870 or 4890 or even a slightly lesser processor. Because sometimes we are comparing apples to oranges here. This is just one of the thousands of configurations that people could have here. Its just the raw numbers of my benchmarking to find what role the processors plays in the FPS with a fairly high end GPU. As to the settings i play with Im listing them below. If you or anyone else would like to benchmark their system I would be more then glad to post it on my website.

Visibility: 4071

Texture Detail: Very High

Anisotropic Filtering: Very High

Terrain Detail: Normal

Objects Detail: Very High

Shadow Detail: Very High

PostProcess Effects: Disabled

http://www.armaman.com

[ICE-Raver 10]

I did a bunch of tests on my i7 dual boot system, Vista 64/Windows 7 64 and I can tell you without a doubt.......................Windows 7 is NOT a reskinned vista.

I had major performance gains on Windows 7 vs Vista with using the same settings on Arma Mark.

My results are here http://forums.bistudio.com/showthread.php?t=73610&page=55

Just an FYI, You must Force off Vsync to get accurate results. With V Sync on I cannot break past about 5400 with windows 7. With it off I get over 7000. You do the math.

Edited August 14, 2009 by ICE-Raver

[ArmedAnarkie 0]

I tested Arma II mark over XP SP3 and Windows 7 x64 RTM; it scores almost same rate. But I wonder frame rate is also same. In some place of map, XP runs smoother and in other place, Windows 7 RTM runs better.

[Scorpio9 10]

Hay thanks a million ICE-Raver when i changed my driver i must have forgotten to check my Vsync so thinking i was in the ArmA config I assumed it was there and changed it too 100 but no change and when you said Vsync it hit me like a ton of bricks. So my scores up to 6089 now but I need to rerun all my benchmarks now. Also I see you have the Coolermaster V8 besides looking as heck how do you like it and what are your temps on your i7 920 and 4.2 GHz running and 100% load with somethink like prime95 or folding at home, also what have you set your GTX 280 clock too?

http://armaman.com

[ICE-Raver 10]

Also I see you have the Coolermaster V8 besides looking as heck how do you like it and what are your temps on your i7 920 and 4.2 GHz running and 100% load with somethink like prime95 or folding at home, also what have you set your GTX 280 clock too?

http://armaman.com

As far as the V8 goes, it's decent. Wish I had got Meglahems after reading all the praise and seeing all the user temps in the evga forums..........but anyway. My biggest issue was that the V8 was so big I had to take my side fan off (that blows on the video cards) because it hit, and I have an alienware case. I had to add a fan loose in my case to blow on my video cards lol. I also added 155 cfm intake and exhaust fans as well as some other case fans for added cooling. BTW- all the fans run at 100% manual. Still yet, I may end up water cooling the cpu so I can put a fan back on my side panel properly.

The i7 gets a little toasty with prime 95. High 70's-about 82 to be exact.

However, the only thing I do with my system is game and check my email and such.

Running back to back to back 3d Mark tests and gaming for many hours at a time I have yet to see my temps go above 58-60 in the worst case scenarios. Usually it is in the high 40's/low 50's. I've been running it like this for a couple of weeks now so it seems to be very stable.

As far as the GPU's go here is a screen. Both cards are clocked almost identical. This also shows my CPU info.

And here is another that shows my typical temps.

The coolermaster in all it's glory.................Excuse the mess, this was during the upgrade.

158 kb

Had to take a side fan off to clear the V8 cooler to get the side back on my old alienware case but here she is.

Edited August 30, 2009 by W0lle

[p75 10]

Grab the Windows 7 Ultimate retail off MSDN and be very surprised....it runs alot better than xp :)

[hybrean 10]

Don't compare 32-bit with 64-bit, theres automatically a performance drop when a 64-bit system is running 32-bit application which in this case Arma 2 is. It makes even bigger difference which such low settings, the more memory is in use the bigger advantage a 64-bit system has.

[Scorpio9 10]

Hybrean are you quoting a Vista 64 bit drop in performance compared to a windows XP 32 bit or what combination are you comparing? I never heard people with 64 bit XPs never complain about this.... but i could be wrong. If you have a source I really would be interested. Thanks

[hybrean 10]

I mean overall, doesn't really matter which OS, running 32-bit applications on 64-bit system takes some extra resources to make it all compatible and by default a 64-bit system takes up a little more memory than a 32-bit for using 64-bit blocks.

The performance difference shouldn't be 11% but it could be around 3-4% depending on the application. Look at this Benchmark http://64-bit-computers.com/windows-vista-32-bit-vs-64-bit-benchmark.html if you want to know how well 64-bit and 32-bit perform different operation against each other (higher value is better) heres another one with xp32 vs xp64 with games http://www.tbreak.com/reviews/article.php?cat=cpu&id=295&pagenumber=4.

Also if Arma 2 was written in 64-bit instead of 32 the 64-bit system would beat it hands down. All im saying is that benchmarking a software on a 32-bit OS and on a 64-bit OS and then draw a conclusion about which OS performs best is not the way to go, instead try 32 vs 32 and 64 vs 64 to get a more exact result.

[galzohar 31]

Who cares which OS performs best, what matters is which OS runs Arma 2 better ;)

[Easun 10]

Grab the Windows 7 Ultimate retail off MSDN and be very surprised....it runs alot better than xp :)

I second that.

[bascule42 10]

I don't know about benchmarks, or test on uber mega balls out in front gaming rigs. I only know what my own 2 peepers are telling me, and that is...Windows 7 RC 7100 runs ArmA 2 significantly better than Win XP (sp3).

OK, 1st installation of ArmA 2 was on Win Xp, (with the hard drive formatted about a week or so beforehand). It ran OK, was just about playable, I tried so many different settings, I probably was on the option screen as much as playing it the first week. 2nd installation of ArmA 2 was on the same machine, (formatted drive of course), but with Win 7 and ran like a bugger. Smoother frame rate, higher settings.

I used to use vista, tried it for about 2 weeks, hated it. It looked good, but was just crap, but really for no real reason that I can tell you, just didn't like it. Win 7 however...I love it.

I dont have a brilliant set up and I belive its on the verge of being out of date, (if not some of allready), but the main bits are as follows:

AMD 64x2 Dual Core 2.7GHz - 2GB memory - ati 3780 (512MB)

And this, me sitting here playing it, is really the only kind of test that matters to me at least for another 18 months when I Upgrade my machine.

[Scorpio9 10]

Visibility Benchmarks as to FPS penalty.

This benchmark was done in a open field looking northwards towards a wooded treeline and mountains. Testing the impact that trees have on FPS. I carefully balanced the sweet spot to 60 FPS to give me a good scale and then increased in value until I reached a 2 FPS drop and then recorded the results. As you can see I found a penalty between 500 to 690 with 695 to 1260 being optimal and the drop in FPS from there. How this will scale with other systems and locations I have no idea. It was only to study the impact of Visibility on trees, how ever there does seem to be some changes to the vegetation in the foreground up to 50m I could not really tell if more was added or the vegetation was just turning or rearranging slightly.

Vis. FPS Range

500 55 penalty

546 53 penalty

598 57 penalty

635 58 penalty

695 60 1260

1266 58 2370

2374 56 2925

2928 54 3300

3301 52 4070

4071 50 5020

5021 48 5659

5660 46 6375

6380 44 7410

7411 42 7865

7869 40 8609

8609 38 9415

9419 36 9999

[Scorpio9 10]

All im saying is that benchmarking a software on a 32-bit OS and on a 64-bit OS and then draw a conclusion about which OS performs best is not the way to go, instead try 32 vs 32 and 64 vs 64 to get a more exact result.

I totally understand your point hybrean and when i get the time I will bench it or someone else does the benchmark however I believe most people that will switch from Windows XP to Windows 7 will opt for the 64-bit to get at their extra RAM most i7 owners run at minimum 6GB and most MOBO's like mine support up to 24GB and there are instances were some games like Flight Simulator X can use 4 gigs if it can get at it. From what i have seen so far ArmA II can not I wish it could because it would smooth out some when entering a map and when in flight and the HDD has to load additional scenery, map area and models causing stutter and drop in frame rates. But this is a good point... it takes along time to run benchmarks and im sure you know how long it takes to install Windows of any flavor. And it does matter to those people what their loss in performance between all OS's will be.

Edited August 15, 2009 by Scorpio9

[mlenser 10]

I'd be interested to here which OS you prefer to play the game on Scorpio9? Do you find that visually the game appears faster on Windows 7 even though the benchmarks come out lower?

P.S.

Thanks for the post - interesting.

[Scorpio9 10]

I'd be interested to here which OS you prefer to play the game on Scorpio9? Do you find that visually the game appears faster on Windows 7 even though the benchmarks come out lower?

Hi mlenser as my results show a 9.62% difference between XP and Win 7 which im willing to let go of the 10% difference since I use my system for much more then ArmA II. Also with my OC it still will put me 10% over XP at stock clock on the i7 920. Also as Placebo states in his ARMA II Community Update post concerning patch 1.04

improving performance, especially in urban areas

Im really hoping for a 5 -10 % increase in performance which I believe they will be able to achieve, on top of that for NVIDIA owners (I dont know about ATI maybe someone can answer that question) I know NVIDIA is very good to optimize their drivers for new games as they come along. I have seen driver optimization yielded 5 - 30% increase in performance.

Visually faster, no a loss is a loss however I believe what your really getting at here is if it is smoother and my answer as of patch 1.03 is no. Short of a more optimized ArmA II a Raid 0 or SDD should really smooth game play out. I still see alot of stutter in ArmA II when it has to goto the HDD when moving, turning and flying across the map to load additional scene elements and also when Objects Detail is set to very high even with hyper threading turned off. Also the more you Benchmark with ArmA Mark you see the score fall off which i believe is due to poor memory utilization or caching.

So I will move to Windows 7 64-bit to get at all of my system resources. Looks like it will be up to the hardware guys to over come the performance issues in Windows Vista and Windows 7, we will just have to get used to it. We gave up performance in the DOS days to move to Windows 3.1 an 3.11 which piggy backed DOS up too when MS brought XP to market and introduced the NT Kernel to the consumer market. Hardware caught up and started meeting the demands of XP thus windows performance became transparent to us and we moved our focus to the demands the games placed our systems, which for many years now has fallen to the GPU makers. When MS brought Vista to market we all saw a huge drop in performance so most of us stuck to XP now that Windows 7 is arriving alot of us now have better systems and it looks like maybe MS has done a bit of optimization to Win 7 so i believe soon the OS will become transparent to us again and the focus will shift back to the hardware vendors, either way we will always be investing wads of cash just to play our games be it PC's Consoles or even iPhones.

Edited August 15, 2009 by Scorpio9

[funkee 0]

tested ArmA2 1.03 on both Windows XP SP3 and Vista Ultimate X64 SP2. performance is much better on Vista + no more mouse lag after installing latest beta patch :)

[koroush47 10]

The coolermaster in all it's glory.................Excuse the mess, this was during the upgrade.

LOL

LOL

LOL

You call THAT a mess??

You should see my computer.

4870x2 (HUGE) stuck into an antec 900. Barely fits and I got this corsair that has like a billion wires. I have no idea what to do with it.

Edited August 30, 2009 by W0lle
quoted image

[doakwolf 10]

Hello,

Has anyone done any Hyperthreading tests with the i7s and ArmA2?

I'm keen to know if the game runs better with HT disabled, as I've read about certain applications/ games.

Cheers,

Jero.

[ICE-Raver 10]

LOL

LOL

LOL

You call THAT a mess??

You should see my computer.

4870x2 (HUGE) stuck into an antec 900. Barely fits and I got this corsair that has like a billion wires. I have no idea what to do with it.

I switched to an antec 1200 this week. It has much more room than the alienware case and fantastic airflow. My cpu idle temps dropped 6 degrees with just the case swap and that is with stock fans on the antec.. I'm very happy with it.

[doakwolf 10]

^ It's incredible what a decent case can do for temps! :)

[purgi 10]

Hello,

Has anyone done any Hyperthreading tests with the i7s and ArmA2?

I'm keen to know if the game runs better with HT disabled, as I've read about certain applications/ games.

Cheers,

Jero.

Yes. At least for ARMA2 there is a significant boost from turning HT off. On my i7 it stutters like crazy with HT on and is smooth with it off. cpucount=4 makes no difference when HT is on.