BF3 64 player MP CPU benchmarks!

Frito

Limp Gawd
Joined
Nov 7, 2005
Messages
254
Yes i know this is the video card section, i was going to post in CPU section but as there are 2 and most people looking to upgrade their systems for BF3 are asking here because naturally they are thinking their video card needs to be upgraded and this information should help out many that are experiencing less than 60 fps performance in game reguardless of settings, video card etc. so i figured i'd post it here

Test setup

Intel Core i5-2500k (sandy bridge)
8 GB DDR3 1600
2x 560 Ti's in SLI (im not testing video card performance here, just showing what none of the reviewers across the net are showing and that is your CPU's performance is key to good frame rates in multiplayer in this game)

in game settings are identical on both tests. 1920 x 1080 HIGH (i messed up on this, i meant to run low settings but had not realized until after i did this that the patch that happened a few days ago changed my video settings from low to auto so i will list my specific settings here, reguardless of this my CPU still became and is always the bottleneck on my system just not so much when its overclocked at 4.2ghz with all 4 cores)

texture quality - terrain HIGH
AA Defered OFF
AA POST HIGH
Motion BLUR ON (oh god how did i not notice this one for the last few days! must have been having too much fun hehe)
AF 16X
Ambient Occlusion HBAO




both FRAPs runs were timed benchmarks 5 min in length done on a FULL 64 player server on metro in conquest (you know how bad this meat grind fest is!!) i started the benchmark once i got to the famous choke point battles for B capture point and just played like i normally do from there and was in the middle of the grind both times rather than trying to get away/past the choke points.

first test i5-2500k with all 4 cores active, running an overclock of 4.2ghz
Frames
40855
Time (ms)
300000
Min
70
Max
201
Avg
136.183

second test i5-2500k with only 2 cores active and cpu downclocked to the minimum (3.4ghz no turbo mode)
Frames
19941
Time (ms)
300000
Min
32
Max
115
Avg
66.47



there you have it, half the frames were produced, my cpu use was maxed out during the 2nd test while during the first test my cpu use was very high but not 100% pegged. while the game was still very much playable it felt much slower than it did the first run, staying above 60 FPS all the time really is key to having the best multiplayer performance and your K/D Ratio will be better if you have higher frame rates not to mention the game is just more enjoyable :)

for those who are wondering why i have so much interest in this its because when i first got the game my system was a C2Q Q8200 overclocked to 2.8 ghz with 4gb of ram and 1 560 Ti, due to reviews on the net stating this game is not CPU depentant i bought another 560 Ti only to find out it was my CPU holding my GPU back even though it was a quad core. although i can no longer benchmark that CPU because i gave it to my brother i can tell you without a doubt that even my i5-2500k crippled down to basically what amounts to an i3 running at 3.4 ghz without HT the performance i experienced running that test above is still well above what my core 2 QUAD was giving me with the same 2 video cards in SLI.
 
Last edited:
here are in game performance graph screen shots, same settings as above, same environment full 64 player server on metro

2 cores 3.4 Ghz

bf32011-11-2408-35-03-05.jpg

bf32011-11-2408-36-22-40.jpg

bf32011-11-2408-38-07-71.jpg


notice the yellow line on the graph is constantly higher than the green line, the yellow one is CPU performance, higher is worse because it represents how long it takes to render a frame in ms

4 cores 4.2 Ghz

bf32011-11-2408-57-08-98.jpg

bf32011-11-2408-55-01-47.jpg

bf32011-11-2408-56-38-81.jpg


notice now my CPU roughly stays with my GPU's in performance, if i set everything to low the CPU will still become the bottleneck but the frame rates are so high it does not matter.

LOW setttings (to show CPU is still the bottleneck even when overclocked)
bf32011-11-2408-53-17-47.jpg
 
first test i5-2500k with all 4 cores active, running an overclock of 4.2ghz

second test i5-2500k with only 2 cores active and cpu downclocked to the minimum (3.4ghz no turbo

Would it not be better to run them at the same clock speed? ...your not really testing the difference between 2 & 4 cores unless all the variable are equal....
 
Would it not be better to run them at the same clock speed? ...your not really testing the difference between 2 & 4 cores unless all the variable are equal....

this is true, what i'm trying to show here is that its not really a core count issue in this game that determines your minimum frame rates, its raw FPU performance of your CPU. the sandy bridge architecture is even with only 2 cores much faster than core 2 architecture in this department so people still using core 2 quad's like i was and being told that a quad core is a quad core and you just need a better video card, like i was lead to believe are wrong. in these tests even with my 2500k crippled to the max (the game won't run on 1 core so can't cripple it that far hehe) it still performed visibly better than my Q8200 at 2.8 ghz did by a large margin too. 64 player servers like i was in during these test would bring my C2Q to its knees and my frame rates would rarely get over 40 fps in heavy action.

if people want me to do an apples to apples run 2 vs 4 at the same speed i'm more than willing to do so but the performance difference will still be very large.
 
ok by popular demand

ran the same benchmark with 2 cores but running the 4.2 ghz OC

Frames
24607
Time (ms)
300000
Min
51
Max
153
Avg
82.023

vs with 4 cores everything else the same

Frames
40855
Time (ms)
300000
Min
70
Max
201
Avg
136.183

here are screen shots of the overclocked dual core performance graph

bf32011-11-2410-13-35-68.jpg


notice large performance hits when outdoors.

bf32011-11-2410-19-27-79.jpg

bf32011-11-2410-19-31-44.jpg


performance indoors was much better even under heavy action but still poor compared to with all 4 cores. i would conclude that an i3 sandy bridge CPU can provide ok performance if you could overclock one but the performance outdoors really suffers bad so better stick to metro if you want higher frame rates :D
 
Last edited:
be interested in seeing 4 cores at 3.4ghz no turbo

can do.

4 cores 3.4 Ghz.
Frames
34828
Time (ms)
300000
Min
72
Max
174
Avg
116.093

vs 4 cores 4.2 Ghz.
Frames
40855
Time (ms)
300000
Min
70
Max
201
Avg
136.183

some performance graph shots
bf32011-11-2410-52-24-04.jpg

once again notice outdoors is where CPU performance is most important apparently.
bf32011-11-2410-53-03-00.jpg

bf32011-11-2410-53-10-10.jpg

indoors performance is close to the same as with my OC, game does not feel different, in light of the outdoor performance hits that are clearly apparent i think next i should do some benchmarking in a map like caspian border because clearly its the large outdoor enviroments that are the source of CPU induced performance hits not rockets and bullets flying all over the place in the metro subway :)
 
Last edited:
One problem: i3 sandy bridge cannot be overclocked, so that point is moot.
 
One problem: i3 sandy bridge cannot be overclocked, so that point is moot.

indeed and even if you could overclock it your still better off with a locked i5 according to my benchmarks.
 
Very good info Frito, thanks for taking the time to do this

not a problem, people have been asking for proof the game requires CPU performance and nobody seems to want to provide it even though its clearly apparent to people like me who had a core 2 quad when the game came out and had terrible performance due to it and upgraded to fix that. its worth noting that even though i get higher average frame rates with my CPU overclocked CPU use is not maxed out with or without the OC when im running 4 cores, the only way i saw pegged CPU use was when running only 2 cores. the game seems to just want to use spare CPU power if it can once your over a certain frame rate/performance threshold. if your under that threshold it runs it to the max.
 
Very nice comparison, thanks for the effort and sharing it here.

If I may ask, what's the general performance difference between a Core i7 9xx compared to Sandy Bridge? I have a Core i7 930 @ 4GHz. I'm planning on upgrading my graphic card, if my CPU isn't going to severely hold back the performance that is :(
 
Very nice comparison, thanks for the effort and sharing it here.

If I may ask, what's the general performance difference between a Core i7 9xx compared to Sandy Bridge? I have a Core i7 930 @ 4GHz. I'm planning on upgrading my graphic card, if my CPU isn't going to severely hold back the performance that is :(

i think you will be fine with an i7 9xx thats overclocked, the sandy bridge chips do have a bit of a FPU performance advantage clock for clock but its not night and day.

http://www.anandtech.com/bench/Product/142?vs=287

the fallout 3 benchmark at the bottom is probably a good indicator of the performance difference clock for clock (well close to clock for clock, could not find an exact mhz match) 86.1 fps vs 96.9fps thats about 12% faster is all in that game and most of the other non synthetic FPU heavy benchmarks appear to show the same gain on sandy bridge or less
 
ah finally a benchmark site ran multiplayer benchmarks on various CPU's, i assume these are all stock clocked.

i70m.png


arg the image background is transparent :(

from top to bottom those are
i7 2600K
i7 975
i5 2500K
FX-8150
Phenom II X6 1100T
Phenom II X4 980
Phenom II X2 560

if you want to see their article i hope you can read Swedish or get out google translate :D

http://www.sweclockers.com/artikel/14650-prestandaanalys-battlefield-3/5#pagehead
 
Last edited:
Just copying this here since it's relevant:

before and after, I did the fraps benchmarks basically during entire multiplayer rounds on the same server full 64 players. I started and stopped the bench during the game to avoid menus etc.

before, i5-750 @ 3.8, P55-GD80, 8gb ram, 6990+6970, 11.10 + 11.9 cap4, 1080p @ 120hz bf3 on all ultra -EXCEPT MSAAx2 and Post = mid

cpu @ 99%, gpu's (x3) mid 80's 2 fraps examples below.

Frames, Time (ms), Min, Max, Avg
89237, 915304, 49, 195, 97.494

Frames, Time (ms), Min, Max, Avg
68049, 704266, 45, 202, 96.624

---------------------------------------------------------------------------------
After: i7-2600k @ 4.9, P8z68-v PRO, 8gb ram, 6990+6970, 11.10 + 11.9 cap4, 1080p @ 120hz bf3 on all ultra -EXCEPT MSAAx2 and Post = mid

cpu @ 65%, gpu's (x3) mid to high 90's 1 fraps examples below.

Frames, Time (ms), Min, Max, Avg
195245, 1611101, 63, 201, 121.187
 
Excellent benchmarks guys... and yeah, TechSpot has a lot of CPU benchmarks in recent games which consistently show a Core 2 generation CPU is way way behind a same-#-of-cores i5/i7 2nd-gen, even at the same clock speeds (let alone the overclockability and stock speed differences). It's often the CPU that's the issue, not the graphics card, or at least more of an issue, and people are being guided incorrectly due to this misconception that video cards are always the bottleneck.
 
Excellent benchmarks guys... and yeah, TechSpot has a lot of CPU benchmarks in recent games which consistently show a Core 2 generation CPU is way way behind a same-#-of-cores i5/i7 2nd-gen, even at the same clock speeds (let alone the overclockability and stock speed differences). It's often the CPU that's the issue, not the graphics card, or at least more of an issue, and people are being guided incorrectly due to this misconception that video cards are always the bottleneck.

yeah well what is not helping is lots of review sites are overlooking CPU performance in BF3 because they are just testing it in single player and single player is like night and day vs multiplayer when it comes to CPU use. i can point out a bunch of main stream reviewers that have stated that the game is not very CPU dependent and just say broadly that a quad core is enough and that is not the case with older quad's in multiplayer. techspot didn't do a good job just like many others and just tested single player CPU use, nor did they test any core 2 cpu's http://www.techspot.com/review/458-battlefield-3-performance/page7.html
 
Toms hardware I think also had a review that showed no difference whatsoever between a vast range of cpu's which is clearly not the case for multiplayer. A game like bf3 should ONLY be reviewed/benchmarked from a multiplayer viewpoint because that's what everyone does with them (even if you play single you will be done in a few hours)
 
Toms hardware I think also had a review that showed no difference whatsoever between a vast range of cpu's which is clearly not the case for multiplayer. A game like bf3 should ONLY be reviewed/benchmarked from a multiplayer viewpoint because that's what everyone does with them (even if you play single you will be done in a few hours)

yeah you are correct tom's hardware did a short CPU use section on the game and once again failed to test multiplayer and concluded CPU power is not important to the game.
 
Nice work, Frito, and quite relevant to me as I am putting together a system now that will certainly be running BF3.

I'm curious how the frame rates scale with degree of overclock. Your sig indicates you get up to 4.7GHZ with your i5-2500k yet you show results for only 4.2 GHz.

Could you please try a run with a higher OC? I'm curious if buying a better overclocking board would be good here.

And how reproducable are your numbers? Are they consistent over a few runs?

Thanks.
 
Nice work, Frito, and quite relevant to me as I am putting together a system now that will certainly be running BF3.

I'm curious how the frame rates scale with degree of overclock. Your sig indicates you get up to 4.7GHZ with your i5-2500k yet you show results for only 4.2 GHz.

Could you please try a run with a higher OC? I'm curious if buying a better overclocking board would be good here.

And how reproducable are your numbers? Are they consistent over a few runs?

Thanks.

i could do a test run at 4.7 ghz but my cpu is not 100% stable at that speed, it will play BF3 for hours on end at it though it just fails prime95 tests. i've been running at 4.2 and a lower voltage to keep the heat down on my CPU and i don't like taking it past 1.4v the performance i get at 4.2 i'm happy with so i've kept it that way for awhile just to be on the safe side but my minimum frame rates are higher at 4.7 at least on low settings for sure.

as for consistancy of the benchmark, the reason i did 5 min long benchmarks on full 64 player metro servers was for consistency, its about as controlled of an environment you can get in multiplayer due to how the map has major choke points at the middle flag and most of the time on 64 player servers both teams are fighting back and forth at 2 escalator entrances and 1 hallway/stairway entrance. there will naturally be a variance on average frame rates but i would guess the margin of error to be around 5% perhaps later today when i'm in some metro servers ill run the bench mark a few more times at 4.2 ghz just to see how much it might vary from my first run.

oh another thing, the best indicator if a higher overclock will yield more frame rates is watching GPU use while in game, because i have 2 monitors i can do this and earlier i was playing on caspian border (a very large and open map) and both my GPU's hovered from 80-90% use so i definately still have a CPU bottleneck on larger maps. metro is very small map in comparison and will go GPU limited more often, this is why i was thinking of redoing these tests on a bigger map but the problem and reason i didn't use them before is getting the consistency between runs will be much harder unless i just drive a jeep around the map over and over or something hehe
 
Last edited:
Yeah, don't want you to max stress the system just for a test, but should you ever try a run at a higher OC (say 4.4 or 4.5) please post the results.

thanks again for reporting your interesting results.
 
I found it hilarious that review sites were stating 'BF3 is not CPU dependent,' while testing only MP. For those of us who played BF:BC2 in MP, we know that CPU usage jumps vastly. Considering that BF3 doubled the player count with larger maps, you could only conclude that CPU was going increase more!

That's the single reason that I upgraded from a Core 2 Quad at 3.4GHz to a 2500k at 4.8GHz. Didn't want CPU usage slowing me down ever.
 
good work m8.

I wonder how my c2duo @ 4GHz stacks up to a sandybridge!

I know I should upgrade but can't stomach ditching my whole rig bar drives and gfx card :(

Are you using fraps to bench this? Might give it a go.
 
good work m8.

I wonder how my c2duo @ 4GHz stacks up to a sandybridge!

I know I should upgrade but can't stomach ditching my whole rig bar drives and gfx card :(

Are you using fraps to bench this? Might give it a go.

yeah i used fraps to do the benchmarks, if your on a C2D still as long as you have a decent graphics card (560 or higher) i can guarantee you will see massive frame rate increases in multiplayer. i upgraded from a C2Q Q8200 @ 2.8 ghz to what i have now because it wasn't fast enough to keep up with 1 560 Ti let alone 2 in SLI in this game.
 
yeah well what is not helping is lots of review sites are overlooking CPU performance in BF3 because they are just testing it in single player and single player is like night and day vs multiplayer when it comes to CPU use. i can point out a bunch of main stream reviewers that have stated that the game is not very CPU dependent and just say broadly that a quad core is enough and that is not the case with older quad's in multiplayer. techspot didn't do a good job just like many others and just tested single player CPU use, nor did they test any core 2 cpu's http://www.techspot.com/review/458-battlefield-3-performance/page7.html

Wow that is interesting, they did so for Crysis 2 as an example: http://www.techspot.com/review/379-crysis-2-performance/page8.html and most of the others I looked through there, kinda odd they didn't on BF3.
 
Yeah, don't want you to max stress the system just for a test, but should you ever try a run at a higher OC (say 4.4 or 4.5) please post the results.

thanks again for reporting your interesting results.

ok just did 2 benchmark runs at 4.5 ghz

Frames
40442
Time (ms)
300000
Min
83
Max
185
Avg
134.807


Frames
40363
Time (ms)
300000
Min
82
Max
188
Avg
134.543

pretty consistent results

here was my first test i did yesterday at 4.2 ghz

Frames
40855
Time (ms)
300000
Min
70
Max
201
Avg
136.183

my minimum frame rate went up and that is what i expected because unless you run maxed out video settings generally your CPU is what determines the minium frame rate in my experience. for whatever reason however my max went down, this may be because when i did the first tests i was playing assault class and using a F2000 with the IRNV scope (night vision/thermal) so zooming into aim view with that may have caused higher top frame rates, while this time i was playing support and using an M249 with a normal scope.

either way on metro at least the additional overclock had very little impact, in a bigger map i'm sure it will have more of an impact because like i said earlier my GPU's use runs near maxed out on metro (95-100%), in bigger maps it floats around 80-90% indicating a CPU bottleneck.
 
Thank you frito, this was just along the lines of what I wanted to see when I posted my performance thread.

Now, I just need to start the fund for the quad core upgrade, as it will make a significant difference :D
 
Well, the first post convinced me to upgrade later then. I mean I actually bought a motherboard to upgrade so I can have a last computer with WinXP. But, if or when I decide I want to try BF3, then I think it will be with a Win8 machine running the next generation CPU/GPU.
 
thank god someone posted this, i was searching all over for benchmarks to show me if cpu overclocks were even going to matter! All i will care about is multiplayer framerate.

It will be interesting to see what settings I have to use to get 60+ fps average on a GTX 460 @~950-1000mhz with my i3-530 clarkdale OC'd to 4.5ghz (and what diff the OC makes).

(btw, i say all in future tense cause I'm still waiting for parts to arrive)
 
I wouldn't mind seeing if people can play offline with 1680x1050 at medium settings with cheaper hardware like a Core 2 Duo 8400 and GTX 550 Ti.
 
Stupid question but, how exactly are you forcing your CPU to use a certain amount of cores like that? Is this something you have to specific in the BIOS or something? Is my 2500k not actually running on 4 cores by default?
 
Hey Frito thanks for posting these. Just a silly question. What program are you using that monitors in graphs for your GPU and CPU in that screenshot. Also I am viewing this at 2560x1600 and it's difficult to read the text on the graph on your screenshot. I could just download the screenshot and magnify it :p
 
I wouldn't mind seeing if people can play offline with 1680x1050 at medium settings with cheaper hardware like a Core 2 Duo 8400 and GTX 550 Ti.

single player is less demanding on the system as a whole, you will beable to play the game and it should have playable frame rates at medium settings.

Stupid question but, how exactly are you forcing your CPU to use a certain amount of cores like that? Is this something you have to specific in the BIOS or something? Is my 2500k not actually running on 4 cores by default?

at least in my motherboards BIOS i can disable as many cores as i want until i'm running just one. (mobo is in my sig)

Hey Frito thanks for posting these. Just a silly question. What program are you using that monitors in graphs for your GPU and CPU in that screenshot. Also I am viewing this at 2560x1600 and it's difficult to read the text on the graph on your screenshot. I could just download the screenshot and magnify it :p

those graphs are produced by the game's engine

to display it in game open the console (~ key)

then put in
render.perfoverlayvisible 1
 
This just makes me want a 2600 and HT even more, gee thanks a lot Frito:rolleyes:
but seriously, thanks a lot man, this puts my questions about a q9550 being enough to hold me over for 38xx or ivy to rest. I am fairly certain going from an e6750 @ 3.38 to whatever I can get from a q9550 will not be enough to justify buying it. Do you think the 2600 and HT is worth the extra $ over the 2500? It seems so for bf3 and future HT games.
 
This just makes me want a 2600 and HT even more, gee thanks a lot Frito:rolleyes:
but seriously, thanks a lot man, this puts my questions about a q9550 being enough to hold me over for 38xx or ivy to rest. I am fairly certain going from an e6750 @ 3.38 to whatever I can get from a q9550 will not be enough to justify buying it. Do you think the 2600 and HT is worth the extra $ over the 2500? It seems so for bf3 and future HT games.

its not tbh. many people with i7's are reporting a studdering problem with HT enabled and are having to disable HT in their BIOS to stop it from happening. apparently DICE did not code the engine with HT in mind just multi core cpu's and having hyperthreading on causes strange performance problems. i imagine they will probably address this in a future patch but at the end of the day HT has never had any positive impact in any games so unless you want it for some non-gaming reason i would say get a 2500k
 
Back
Top