Sign in with
Sign up | Sign in
Your question

Battlfield 3 Multiplayer Core Scaling.

Last response: in CPUs
Share
a c 186 à CPUs
September 26, 2012 6:36:45 AM

Well, I still see these threads going around asking if a dual core would be able to play BF3 Multiplayer. Most of these threads are always talking about i3's vs Quad Core, so I decided to test this out with my i7-2600K.

I ran BF3 on Ultra Settings with V-Sync off, and benchmarked with a 120 second fraps run. I played on a 64 Player server called: Chuck Norris Training Camp http://battlelog.battlefield.com/bf3/servers/show/c96c3....

Test System:
i7-2600K @3.1ghz
ASRock Z68 Extreme 4 Gen 3
GTX 560ti 448 Cores @935mhz
16GB Patriot Sector 7

I tested the i7-2600K @3.1ghz in the following configurations:
4C/8T
4C/4T
2C/4T

The game was played with Ultra Details @1080P, meaning everything was turned on except for V-Sync.

120 Second Results:


Surprisingly the dual core setup did very well, in fact I was even shocked to find it performed 2FPS short of the 4 Core setups. :o 

Full Match Results:
September 26, 2012 6:43:36 AM

Can you test with only 2 cores active & without HT enabled?
a c 186 à CPUs
September 26, 2012 6:43:48 AM

Will test tomorrow.... ;) 
Related resources
September 26, 2012 6:47:57 AM

Thanks. :)  Not just FPS but I would like to know if you experience any micro stuttering/lagging.
a b à CPUs
September 26, 2012 6:49:24 AM

The minimum FPS is quite revealing, the 2 core config scores 6 FPS lower.

If you're a "hardcore" BF3 gamer IMO you should log a session or two in each config, not just some arbitrary 120 s sequence. That will give us more meaningful results regarding real game play and of course the more time/sessions you log, the less random variations affect your results.
a c 186 à CPUs
September 26, 2012 6:51:06 AM

Alright, will do that tomorrow^^
a b à CPUs
September 26, 2012 8:38:58 AM

Also need to tell us the map used and roughly what occured in the run, those figures are high enough that I'm guessing you were stood in spawn looking at the sky (basing that assumption on the fact a GTX 570 can only manage 59.9FPS on ultra on a single player bench so your figures appear high) this is why people don't like benching the MP its hard to create a proper test that can be repeated
a b à CPUs
September 26, 2012 9:16:31 AM

This is why you need to log lots of hours to even out the randomness.
a b à CPUs
September 26, 2012 9:56:31 AM

clicked on the link to that server, its showing nashar canals, 48 player.

looks like your running on a gpu bottleneck tho, run some lower res tests and see how they differ.



Anyone know wich map hits the cpu the hardest? That should be the main map to test.

According to sweeclockers MP test, caspian borders can bring a phenom X2 to its knees, pushing 9 fps through a 6990.
a b à CPUs
September 26, 2012 10:16:52 AM

noob2222 said:
clicked on the link to that server, its showing nashar canals, 48 player.

looks like your running on a gpu bottleneck tho, run some lower res tests and see how they differ.


Anyone know wich map hits the cpu the hardest? That should be the main map to test.

According to sweeclockers MP test, caspian borders can bring a phenom X2 to its knees, pushing 9 fps through a 6990.


Caspian or Firestorm are quite demanding but it depends on whats done in the run and here lies the issue.

Spawn at firestorm and run through empty desert for 2 minutes........ any CPU/GPU combo will get high FPS readings but its meaningless.... however this is the easiest to reproduce for multiple tests

Attempt to stage and then recreate an action sequence..... this is nearly impossible to do with multiple players and be considered reliable..... this would be the only way to really get accurate definitive data

Spawn in say caspian jump in a vehicle and roll about the map........ not very scientific and cant be repeated reliably...... although not really reliable this may be the easiest way to create data that actually means something (as in relates to gameplay though as a comparison tool between CPUs and GPU's it's obviously not reliable).

I think muffin began his run too early those high max FPS figures at ultra settings indicate spawning and looking at the sky, they inturn skew off the average figure
a b à CPUs
September 26, 2012 10:49:49 AM

the thing is multiplayer itself isn't reliable for equal comparisons, but it is necessary.

what we need to know is IF a certain cpu chokes for any reason at all, doesn't matter what caused it. Thats when you will get piZZED at your computer and start bashing the keyboard against the wall.

I know caspian border I would not run a 41xx BD cpu in crossfire 1080P ultra settings. I was getting 100% cpu usage and never dropped, fps from 40 to 80.

Comparatively, the 81xx ran 120fps 90% of the time, few drops to 90, cpu usage ~60%, gpu usage 100%.

This was like you said, ~5 minutes of just having fun, flying, driving, crashing into things on caspain border with 63 other people just reaking havok.

IMO MP testing is a pass/fail grade, running your fraps to monitor fps (especially minimum), if it drops to "your peronal threshold" it fails. For me personally thats 60 fps.


BTW muffin, try different maps also, see how much variance there is.
!