Jump to content
helo

ArmA3Mark - Benchmark your ArmA 3

Recommended Posts

After having waited so long for a benchmark mission, no one provided to the community, I took the initiative and created the ArmA 3 Version of the toad´s arma-mark to benchmark the MilSim we all appreciate. The benchmark is still in the early phases and will be updated later on to fully utilize its potential.

Unzip the mission file and extract it into the mission folder in your steam -> arma3 directory.

Happy benchmarking.

Download:

Stratis Benchmark:

https://docs.google.com/file/d/0ByhZc2l2kSOaX0czdDhaRnMtbW8/edit?usp=sharing

Altis Benchmark

https://docs.google.com/file/d/0ByhZc2l2kSOaQ2VJRTZjbFBoN1k/edit?usp=sharing

---------- Post added at 11:19 PM ---------- Previous post was at 11:11 PM ----------

Link is accessable for everyone in the internet :bounce3:

Current Version 0.51 for the Beta:

https://docs.google.com/file/d/0ByhZc2l2kSOaX0czdDhaRnMtbW8/edit?usp=sharing

Altis Benchmark v0.6

https://docs.google.com/file/d/0ByhZc2l2kSOaQ2VJRTZjbFBoN1k/edit?usp=sharing

Edited by Helo

Share this post


Link to post
Share on other sites

Yep... confirmed it is now public, just downloaded it :)

This was sorely needed. I did a run and got some variable script errors in some scenes, apparently nothing show stopping. put a "-showScriptErrors" to see what i mean.

Aside of that would suggest two things, one is for the results be copyToClipboard for easy access and sharing*, another is to settle some default settings to normalize results from different users/systems.

Many thanks for this. :bounce3:

* adding this in init.sqs under #scoreloop does the job but lacks good formatting (newlines go in plain text)

copyToClipboard format["- RESULTS -\n-----------\nTest One - %1\nTest Two - %2\nTest Three - %3\nTest Four - %4\nTest Five - %5\n%7's OFPMark is %6!\n\n - press 'ESC' to quit -",(scores select 0),(scores select 1),(scores select 2),(scores select 3),(scores select 4),(((scores select 0) + (scores select 1) + (scores select 2) + (scores select 3) + (scores select 4)) / 5) * 100,(name player)]

- RESULTS -\n-----------\nTest One - 33.1077\nTest Two - 23.714\nTest Three - 19.94\nTest Four - 21.4516\nTest Five - 38.2514\nGamma's OFPMark is 2729.29!\n\n - press 'ESC' to quit -
Edited by gammadust

Share this post


Link to post
Share on other sites

Point of this is to have on the one hand test scenarios with little to no AI active in the scene and on the other side showcases, where there is a lot of AI calculation done in the engine evironment, referring to test 2 and 3 here mainly with the amount of AI soldiers and vehicles.

Share this post


Link to post
Share on other sites

Wasn't sure of parameters in test so I used recommended Default as well as placing all settings to Standard/No AA.

Dev Build:0.53. 103478

Default Settings:

Resolution: Native 2560x1440

Texture: Very High

Objects: Ultra

Terrain: Very High

Clouds: Ultra

Shadows: Ultra

Particles: Very High

Score: 3829.01

All Standard/No AA:

Resolution: 2560x1440

All Standard

Score: 5384.37

Specs in sig. Just to note mid-range tex mod was running and when I disabled it I actually had lower scores.

Edited by froggyluv

Share this post


Link to post
Share on other sites
Yep... confirmed it is now public, just downloaded it :)

This was sorely needed. I did a run and got some variable script errors in some scenes, apparently nothing show stopping. put a "-showScriptErrors" to see what i mean.

Aside of that would suggest two things, one is for the results be copyToClipboard for easy access and sharing*, another is to settle some default settings to normalize results from different users/systems.

Many thanks for this. :bounce3:

* adding this in init.sqs under #scoreloop does the job but lacks good formatting (newlines go in plain text)

copyToClipboard format["- RESULTS -\n-----------\nTest One - %1\nTest Two - %2\nTest Three - %3\nTest Four - %4\nTest Five - %5\n%7's OFPMark is %6!\n\n - press 'ESC' to quit -",(scores select 0),(scores select 1),(scores select 2),(scores select 3),(scores select 4),(((scores select 0) + (scores select 1) + (scores select 2) + (scores select 3) + (scores select 4)) / 5) * 100,(name player)]

Give me a moment to implement your clever suggestion....

Share this post


Link to post
Share on other sites

thanks helo for the benchmark. Do we get more infantry-egoperspective-sequences in the future?

latest dev build, latest nvidia driver, AI=1 (expert)

jbzkknb7ojh.jpg3oonv94l4hyg.jpg6hus5pbnk2vj.jpg2hzbv2w6vjq.jpg

OFPMark: 4562

Edited by JumpingHubert

Share this post


Link to post
Share on other sites
OFPMark is 7011.21

Thanks!

please post the Settings you benched. The result only makes no sense.:o

Share this post


Link to post
Share on other sites
please post the Settings you benched. The result only makes no sense.:o

Check his sig ;)

Will update the results to this post later, gonna make the benchmark soon.

Share this post


Link to post
Share on other sites

http://cloud-2.steampowered.com/ugc/1082262203020257525/7CFBA2E5E7317A9C31B0959595AFADF299B96AD5/

On these settings:

http://imgur.com/a/2WJ06

With the "Landtex" and "Noblur" mods loaded, Maxframesahead on 1 in the config, latest dev build and latest 314.22 WHQL nvidia drivers.

Specs:

Win7 64bit

i5 2500k @4.7ghz

8gb 1333mhz ram

gtx680 @ 1320/3326mhz

OS and ArmA 3 installed on SSD

Share this post


Link to post
Share on other sites

~7500 with everything lowest

~5000 with 1000 object viewdistance and under quality all but particles set to high, rest of the settings should be gpu heavy and I'm not gpu limited, pip always off of course, but shouldn't impact the benchmark.

Core i5-750 4.2GHz, 8GB ddr3-1600

Share this post


Link to post
Share on other sites

Running:

Intel i7 2600 (non-K) OC'd to 3.8ghz

16GB DDR3 RAM Crucial Ballistix

Triple Crossfire (5870 + 5970)(1 + 2)

Arma 3 installed to SSD (Samsung 840 120gb)

3TB HDD (Raid 0, 3 x 1TB HDD)

V-sync switched on for screen tearing really, really bugs the hell out of me.

Using preset "Ultra" with 8 x AA, 1920x1080. V-sync on.

Arma3_Bench_1_Ultra1920x1080_AAx8_zps6a5ab9ad.jpg

Using preset "Standard" with 0 x AA, 1920x1080. V-sync on.

Arma3_Bench_2_Standard1920x1080_AAx0_zps556a4960.jpg

Using preset "Low" with 0 x AA, 1920x1080. V-sync on.

Arma3_Bench_2_Low1920x1080_AAx0_zps08aa3d15.jpg

MSI After burner GPU usage graph for standard:

1. Alt-tabbed into game.

2. Benchmark stated.

3. Benchmark finished.

StandardMSi1_zpse6ad1eb6.png

MSI After burner GPU usage graph for Low:

LowMSi1_zpsf3911e16.png

seems that it doesn't crossfire too well:

Using preset "Low" with 0 x AA, 1920x1080. V-sync on. Crossfire OFF (running only only on 5870)

Arma3_Bench_2_Low1920x1080_AAx0_Cross_Off_zps48b36cc8.jpg

MSI After burner GPU usage graph for single card:

LowMSi1Crossfire_Off_zpscf5f0044.png

Edited by rhyswilliams999

Share this post


Link to post
Share on other sites

with standard settings and VSYNC off

5276

1920x1080

4200 with Very High

--

2500K 4.5ghz 560ti

Edited by tremanarch

Share this post


Link to post
Share on other sites

I get a script error

Error in expression <stop>
 Error position: <stop>
 Error Undefined variable in expression: stop

Share this post


Link to post
Share on other sites
im not going to do this until you decide on a de facto quality setting to run it on, otherwise whats the point?

you can compare every setting providing that you adjust the same settings. Why a standarized setting? Everyone have other preferences related to his system and other factors.

Share this post


Link to post
Share on other sites

Custom settings might of course be useful for personal evaluation, but are absolutely pointless otherwise. Defined settings is the only way to check relative performance, which is where the most interest lies for the community.

EDIT: I am not totally sure but i suspect the Presets do not affect all possible settings, but one of these could serve has a basis i think. And then there are those which have tweeked the arma profile, which would further increase deviations.

Edited by gammadust

Share this post


Link to post
Share on other sites

the view distance is preset by the benchmark. Defined Settings are of course more useful than customs.

Share this post


Link to post
Share on other sites

I think that RAM speed is crucial in this test. On the first run I get terrible FPS in the first test due to the textures still loading, gets better with 2nd and 3rd run.

Share this post


Link to post
Share on other sites
Custom settings might of course be useful for personal evaluation, but are absolutely pointless otherwise. Defined settings is the only way to check relative performance, which is where the most interest lies for the community.

EDIT: I am not totally sure but i suspect the Presets do not affect all possible settings, but one of these could serve has a basis i think. And then there are those which have tweeked the arma profile, which would further increase deviations.

Ahhh Internet forums. Where we all get told what we need.

Drives me nuts. I mean who wants an open benchmark to see which settings make the most individual difference for ones own playing experience and to maximize ones own fps. No no. We need something totally locked so its only a competition between rigs. Ie how much one is willing to spend vs other interests.

If that's such a great idea than post a set of presets you choose and people can use that if they like and post the results from the "presets" My god. Instead people just complain that its flexible and doesn't mean they can feel like a man based on a benchmark? Done my rant....

Interestingly mine showed no difference between high and very high but tanked on ultra

Edit- and by the way.... THANKS FOR MAKING THIS TEST!!!!

Edited by CosmiC10R

Share this post


Link to post
Share on other sites

Please sign in to comment

You will be able to leave a comment after signing in



Sign In Now

×