Jump to content
Greenfist

Yet Another Arma Benchmark

Recommended Posts

view distance is defined by the mission itself, i'm just saying ...

29.6fps on mine . it's a good benchmark , as it seems to be on part with multiplayer performance.

ps: with blastcore addon !

Edited by griffz

Share this post


Link to post
Share on other sites

38.0 fps 4690k@4.5ghz, 2x HD7770, 1920x1080, Win10 64

 

With what settings??

Share this post


Link to post
Share on other sites

Is there any way to get this outside steam workshop?

Oh crap. I forgot to add the new direct download when I updated.

I'll get it back up when I get back to my computer. Soon.

 

Btw, you can also subscribe to it and copy the pbo manually to your missions folder.

 

edit. Link added to the first post.

Share this post


Link to post
Share on other sites

Oh crap. I forgot to add the new direct download when I updated.

I'll get it back up when I get back to my computer. Soon.

 

Btw, you can also subscribe to it and copy the pbo manually to your missions folder.

 

edit. Link added to the first post.

Thanks for the link.

 

WOW, this is intense, almost looks like a stress test.

https://youtu.be/8IX74Dj68k4

Share this post


Link to post
Share on other sites

31.4... Well, I think it's very hard to find a game as CPU-intensive as ArmA nowadays.

 

Ultra@1080P

i7-6700K@4.00GHz

DDR4 2400 8GB*2

R9 380 4GB

SAMSUNG SM951 NVMe 256GB

Share this post


Link to post
Share on other sites

I was playing  a bit with your tool, trying different hardware configurations with game settings in very high, view distance 2500, AA x4.

With graphics, tried with single, double and triple GPU and got basically the same bench result
It seems that your benchmark tool is not made to get results based in graphics performance.

I went to CPU, I have tried different clock speeds from 3.0 to 4.5 GHZ and also got quite similar results.

So I have to ask, how many units and modules you have placed in your benchmark tool?

Share this post


Link to post
Share on other sites

I was playing  a bit with your tool, trying different hardware configurations with game settings in very high, view distance 2500, AA x4.

With graphics, tried with single, double and triple GPU and got basically the same bench result

It seems that your benchmark tool is not made to get results based in graphics performance.

I went to CPU, I have tried different clock speeds from 3.0 to 4.5 GHZ and also got quite similar results.

So I have to ask, how many units and modules you have placed in your benchmark tool?

Yes, the mission is very CPU intensive. Any decent GPU along with reasonable graphics settings shouldn't have any effect on the result.

 

If I recall correctly, there's about 32 active units at the start and then around 12 are spawned/enabled during the test. Everything else is mostly just a scripted camera and some effects.

 

Your results are a bit strange, going from 3 to 4.5 GHz should have a clearly visible difference, like maybe as much as 10 fps increase. I haven't seen many comparisons between clock speeds, but all of them have had some sort of increase in fps. Are you running any mods, by the way?

 

On the other hand, this is Arma. The performance might not obey the laws of physics, computing, and reason.

Share this post


Link to post
Share on other sites

Yes, but are there some situations that I am having hard troubles to understand.
For instance I have several missions made with ALiVE where around 200 individual units (around 30/35 groups, including land and air vehicles) are spawned at start and I can play these having stable 60 fps, even when I load on my computer.
Or, when I load this mission:
https://forums.bistudio.com/topic/176544-coop-dynamic-combat-generator/
No matter if is dedicated server or in my computer and I have 60 fps.
Howvever when I load this one:
https://forums.bistudio.com/topic/175097-spcoop-04-contention-zone/
I have 30 fps no matter what settings or where I am loading it.
These things I am unable to understand.

Maybe is there something that struggles with the game engine and under this situation the hardware has little or none relevance.

Share this post


Link to post
Share on other sites

Bratwurste, check your vsync, it has to be disabled.

 

Most people have the experience of dependency between cpu/ram clock and fps in ai-heavy situations.

 

My bench:

 

30.2 fps @ 4.0 ghz cpu clock

36.2 fps @ 4.8 ghz cpu clock.

Share this post


Link to post
Share on other sites

As you can see patches for a year did not improve the performance.

 

Arma 3 is still not playable ...  :unsure:

 

Let us hope for ARMA 4 

 

orw4oi7.jpg

 

 

dh8YoUi.jpg

 

 

spec: AMD FX 8320 OC @4.0gHz, geforce 660 gtx, 16 gb ddr3 1886 mHz.

 

zd3C5uf.jpg   TrvL5IT.jpg

 

 

dUfn4sW.jpg   rzF5iAf.jpg

 

HHsE6M0.jpg   TEfdKBs.jpg

Share this post


Link to post
Share on other sites

grenfist: this looks very interesting. In what ways is this better than A3's in-built Altis & Stratis benchmarks? more representative of real-world performance? less CPU-bound? easier to analyse? to compare?

d_lukin: A3 runs MUCH better on Intel (due to higher IPC), setting Lower options won't necessarily increase benchmarker & this benchmark doesn't seem to reflect reality of in-game performance.

Share this post


Link to post
Share on other sites

grenfist: this looks very interesting. In what ways is this better than A3's in-built Altis & Stratis benchmarks? more representative of real-world performance? less CPU-bound? easier to analyse? to compare?

d_lukin: A3 runs MUCH better on Intel (due to higher IPC), setting Lower options won't necessarily increase benchmarker & this benchmark doesn't seem to reflect reality of in-game performance.

There's no built-in benchmarks in Arma. You're probably thinking of Helo's ArmA3Mark?

In my opinion, YAAB is somewhat more representative of real world conditions, because it's very CPU bound. Whereas Helo's missions have only handful of AI units, no action and high fps.

On the other hand, my mission isn't as consistent because of the randomness of AI. So comparing effects of small system/settings tweaks is not that easy.

Share this post


Link to post
Share on other sites

There's no built-in benchmarks in Arma. You're probably thinking of Helo's ArmA3Mark?

In my opinion, YAAB is somewhat more representative of real world conditions, because it's very CPU bound. Whereas Helo's missions have only handful of AI units, no action and high fps.

On the other hand, my mission isn't as consistent because of the randomness of AI. So comparing effects of small system/settings tweaks is not that easy.

My bad. I was referring to these 2 missions which seem to be the most common benchmark used in A3 for the past 3 years:

https://forums.bistudio.com/topic/142875-arma3mark-benchmark-your-arma-3/

I'm going to give your YAAB a try.

Thanks

Share this post


Link to post
Share on other sites

We were messing with benchmarks on the 3CB forum and someone showed me this - brilliant mission btw !

 

So I thought it worthwhile to load my results here.

 

First run is with - everything at Ultra  / max apart from oversampling which was set to 100% (1080p)

 

jsfSekh.jpg

 

Then did the same with an OC on the GPUS 

Ql2WyGD.jpg

 

Then dropped back to my vanilla settings with Oversampling at 200%

 

 

BoMbsqE.jpg

 

 

Looks like 4k is costing 5 frames ! Amazingly CPU bound

 

 

CPU 5960x at 4.5

32gb at 1600

2 x GTX980 OC at 1496Mhz core and 1903Mhz on the mem at OC

Share this post


Link to post
Share on other sites

17 fps max on mine, 1920x1200 high setting. Ultra I drop to 15-16fps. Painful.

Share this post


Link to post
Share on other sites

17E06656E0903E633F11DBE6FE783A5C9F9B73D0

 

I7 6700K

GTX 980 TI

ASUS Z170 Pro Gaming

32 GB HyperX Fury DDR4 2666MHz

Samsung 850 SSD

 

Settings:

757979A20BE3DC174DE68B0FDE44BB04F12962E7

 

D58F06B667F89D088E9C90AAFEFFAA003DB1729A

Share this post


Link to post
Share on other sites

I5 4690K

32Gm 2666 RAM

980TI

Arma3 installede on M2 disk

Custom water loop

Settings:

 

ULTRA

2560*1440

144hz

Share this post


Link to post
Share on other sites

I am wondering about the way it works, for the story I am getting on screen is slightly different if I am testing a Pentium G 4500 with a GTX 750Ti or with a GTX 770

With the GTX 750TI, the 2 choppers always went down and crash in town

With the GTX 770, most of the times, the 2 choppers always fly over town and are engaging the T-100 and the Kamish.

Of course I understand the variations due to AI movements, but here it seems a bit different ... weird.

Share this post


Link to post
Share on other sites

I am wondering about the way it works, for the story I am getting on screen is slightly different if I am testing a Pentium G 4500 with a GTX 750Ti or with a GTX 770

With the GTX 750TI, the 2 choppers always went down and crash in town

With the GTX 770, most of the times, the 2 choppers always fly over town and are engaging the T-100 and the Kamish.

Of course I understand the variations due to AI movements, but here it seems a bit different ... weird.

 

I have never seen this when I switch ON and OFF CFX with my AMD 6950s. Totally weird. DEV Branch or Stable Branch?

Share this post


Link to post
Share on other sites

Last test was on "Sneak Preview"

 

I must add that without any discrete graphic card using only the Intel HD Graphics 530, I am getting 20 FPS on "Low" ... and only one chopper is going down on town!

 

I will test again with 1.60 (Main) but it will take some time.

I will have to rebuild some paths due to switch from SSD 840 to SSD 850 EVO M.2

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

×