BF4 Retail CPU scaling measured

[setter]

Impressive.

Can you do the same using v syn to see what your cpu and gpu useage is.

Ill try it later.

For those wondering about the difference between an i5 and an i7 with sli. Below are a few benches i ran with ht on/off. Cpu was at 4.5ghz.

Ht off.
Frames, Time (ms), Min, Max, Avg
10988, 180000, 6, 105, 61.044

Ht on.
Frames, Time (ms), Min, Max, Avg
10039, 120000, 57, 115, 83.658

With ht off the minimums and average take a bit of a hefty drop and the game is very stuttery. Cpu use hits 99%, gpu usage doesnt go over 80%.

 

[Finners]

Thats brilliant!, Could you disable SLI and do the same?

Mines pretty smooth on an i5 with a 680 but would like to know if an i7 would make a significant difference.

Is that 64 player online?

 

[setter]

For this game, youd only need an i7 if you decide to go sli at some point. As mentioned above, with a single card an i5 is fine. Prior to the cpu and mobo in sig i had an ivy bridge setup. Played bf3 fine with a 3570k and a single gtx 670. It was only when adding a second card that the limitation of the i5 became present. Moved to a 3770k and things were much better.

 

[mmj_uk]

Those results look flawed, no game should be hitting 6fps (unless it runs out of memory or something) and that has obviously dropped the average down a lot.

 

[setter]

That may be from the beta mmj, (ill need to check).

Screenshot of cpu, gpu usage with vsync enabled. Game settings at ultra, x4msaa, hbao on, blur off at 1920x1200.

Spoiler

 

[teppic]

It's perfectly possible for multithreaded games to hit regular lows in single digits because of stutters from high latencies where there aren't enough cores/HT to keep up perfectly. It's not the same as actual 6fps, just a fractional drop, stutter, then back up again.

 

[dmferrett]

That may be from the beta mmj, (ill need to check).

Screenshot of cpu, gpu usage with vsync enabled. Game settings at ultra, x4msaa, hbao on, blur off at 1920x1200.

Spoiler

Impressive again,
I suppose if you disable HT and apply vsyn that would simulate my findings Setter.

Is that single player as the cpu utilisation is 19%, which is incredibly lower. If you tried that on a 64 man server your utilisation should be getting hammered i.e should be much higher.
My bro has a 3930k @4.6Ghz and hes pulling 70% across 6 cores and 12 threads for example.

What ever the case that utilisation is impressive to say the least.

So we have proven vsyn drops cpu usage and gpu usage. I have to figure out how to post my results for 2x7950's and i5 3750k running in BF4 MP.

According to your results an i7 is 60% quicker / more powerful than an i5 ivy in BF4............Its tough to accept that but if thats the case its pretty damn amazing.

Look at this below also.

http://www.pcgameshardware.de/screenshots/original/2013/10/Battlefield-4-Beta-CPU-Benchmarks-pcgh.png

 

[setter]

The above was taken on a 64 player dawn breaker server, i was in cover under a flyover when i alt tabbed out to take the screen shot. With ht enabled and vsync off cpu use is at 65% ono, gpu's maxing at 99%. Though they do still drop a bit but ive found this with bf3 too. Crysis 3 is one game were theyre maxed out an awful lot more.

 

[alex_123_fra]

My bro has a 3930k @4.6Ghz and hes pulling 70% across 6 cores and 12 threads for example.

What GPU setup is he running? My 3930k with overclocked 780s is nowhere near 70% utilisation across 12 threads. Around 1-2 threads show 60-70% utilisation, the rest are very low. Overall, around 40% utilisation even at 4.5GHz.

 

[passey89]

Just ran a few BF4 tests no I have my cpu to 4.0ghz.

High default settings
FPS
70 ave 95 high 65low
1.5GB Vram usage
80% GPU usgae

Ultra
60Ave 70high 50low
1.8GB Vram usage
98% GPu usage

probably going to play on high as I don't like those lows going below 60fps.

Will try a better overclock on the cpu tomorrow once my proper cooler arrives

 

[Subliminal Aura]

What GPU setup is he running? My 3930k with overclocked 780s is nowhere near 70% utilisation across 12 threads. Around 1-2 threads show 60-70% utilisation, the rest are very low. Overall, around 40% utilisation even at 4.5GHz.

I'm with this.... I decided to install this piece of crock last night and noted the cpu utilisation.

From what I observed I saw 4 to 6 threads being used which is nice to see but dashes hopes of seeing the first game to scale to 8 threads and beyond

 

[silversurfer]

Hmm... Judging by that chart then it looks like I could get a rather sizable fps increase going from my i7 920 to an i7 4770k.

Just set the i7 to 4ghz.

 

[illuz]

You guys tried unparking your cores? Google it.

I'm getting 50-60% load with 100% thread utilisation. Game runs damn well if you ask me...

 

[Subliminal Aura]

You guys tried unparking your cores? Google it.

I'm getting 50-60% load with 100% thread utilisation. Game runs damn well if you ask me...

Switching core state should not affect your load average as it's not a common event. Performance gain is going to be almost negligible unless you can prove otherwise.

 

[wildman]

Very much so.
Might make Intel actually release something/drop some prices.

i doubt that. Just because AMD get top marks in a single game does not a superior CPU make...............

 

[Rofflay]

Try turning off HT. In BF3 HT hurt performance above four cores. Once it was disabled it rose frame rates by about 25%. It is probably the same case with BF4 in that only 4C 4T chips would want HT. Everything else should disable HT.

Please report back

 

[Martini1991]

i doubt that. Just because AMD get top marks in a single game does not a superior CPU make...............

I'm more on about Steamroller in general.
As it all points to it being a decent CPU, it should eat into i5 sales.

 

[alex_123_fra]

Switching core state should not affect your load average as it's not a common event. Performance gain is going to be almost negligible unless you can prove otherwise.

Totally agree. My testing (on page 11 of this thread) shows that core unparking makes no difference to frame rates or core utilisation when the CPU is being loaded.

Try turning off HT. In BF3 HT hurt performance above four cores. Once it was disabled it rose frame rates by about 25%. It is probably the same case with BF4 in that only 4C 4T chips would want HT. Everything else should disable HT.

Please report back

This is interesting. Haven't tried it with HT off and in theory there should be higher utilisation of 6 cores vs 6 cores + 6 threads. Certainly in my AMD rig with an FX8320, all 8 threads are being utilised almost 80%.

 

[dmferrett]

What GPU setup is he running? My 3930k with overclocked 780s is nowhere near 70% utilisation across 12 threads. Around 1-2 threads show 60-70% utilisation, the rest are very low. Overall, around 40% utilisation even at 4.5GHz.

Hes using 3x7950's at 5800x1080 and apparently hes declocked to 4.4Ghz for better stability (random crashes, which in retrospect are probably BF4's Fault). He playes in custom ultra with 4xMSAA. At this stage I dont how game settings, cpu clocks and eyeinfininty all tie together in terms of cpu usage...

I still find it incredible that a haswell run at a stock only uses 26% in sli. (2x670's)

Thats just jaw dropping...

Are the 670's a major restriction in cpu utilisation.

 

[dmferrett]

I'm more on about Steamroller in general.
As it all points to it being a decent CPU, it should eat into i5 sales.

Well if the i7 uses 26% in sli and I use 80-90% with an i5, it will be an i7 next for me..50-60% less cpu utilisation according to people on here, I just find that hard to grasp!!!!