BF4 CPU usage

[skipsneeky2]

Call me crazy but anyone with a core i5 should be running windows 8,i had a relentless stutter and after i installed windows 8 and noticed lower cpu usage as well as a stutter free experience.Most people in game who don't stutter keep posting they have i7 and 8 core chips with 7....while i see plenty of quad core i5 and amd folks with the stutter.

A member here figured since it was a fresh install of 8,that is why i had no stutter so i reinstalled windows 7 with the same exact drivers,programs in the same order and guess what?Stutter returns.

Who wants to bet that after i reinstall windows 8 that suddenly it disappears again?

 

[AtenRa]

I haven't downloaded the beta yet but for you guys that did how big is it?

I just upgraded BF3 to premium and that takes up like 34GB's.......

BF4 BETA is only 5GB

 

[JAG87]

Would it be more accurate to say this game supports X amount of cores and the amount of threads it has dosent matter?

In a rudimentary way yes, it's far more accurate than saying the game supports 4, 6, or 8 threads. The game spawns WAY more threads than that.

A certain forum member hits a whole new level of ignorance when stating for example that next gen consoles will use 6 threads for games and reserve 2 cores. This is gibberish and clearly there is lack of basic understanding of how a thread scheduler works in combination with instruction pipelines and out-of-order execution.

 

[PPB]

Just to put things in perspective,
Spawn point number C in Domination map is the Spawn Point number D in Conquest map.

Conquest Map

Domination Map

Yep, chances are that in release Domination FPS will be so high (just like what happened on Noshar TDM, for example) that most procs will hit +60fps to care if i5/i7's have a bigger lead or not.

What people cared in BF3 was not to be bottlenecked in those maps with lots of destruction in CQ 64 P (Seine Crossing, most maps on B2K, etc). For example, In seine I was hitting as low as 50 fps sometimes when everything was destroyed already, on the other hand, on Noshar TDM I was hitting +80 min fps, so it didnt really matter.

Hope reviewers this time don't stick with lame SP benches once the game is out, and have the balls to bench 64P CQ, which is what most people care in this game.

 

[Makaveli]

BF4 BETA is only 5GB

Thanks that isn't too bad.

I'm still undecided if I should bother with the beta.

 

[tential]

Yep, chances are that in release Domination FPS will be so high (just like what happened on Noshar TDM, for example) that most procs will hit +60fps to care if i5/i7's have a bigger lead or not.

What people cared in BF3 was not to be bottlenecked in those maps with lots of destruction in CQ 64 P (Seine Crossing, most maps on B2K, etc). For example, In seine I was hitting as low as 50 fps sometimes when everything was destroyed already, on the other hand, on Noshar TDM I was hitting +80 min fps, so it didnt really matter.

Hope reviewers this time don't stick with lame SP benches once the game is out, and have the balls to bench 64P CQ, which is what most people care in this game.

Yes, because reviewers dont' review 64P CQ online because they don't have balls.

It has NOTHING to do with being able to reproduce those results accurately over 10-20+ processors....
Since it's so easy, you're welcome to launch your own website and do it yourself you know.

 

[Erenhardt]

Yes, because reviewers dont' review 64P CQ online because they don't have balls.

It has NOTHING to do with being able to reproduce those results accurately over 10-20+ processors....
Since it's so easy, you're welcome to launch your own website and do it yourself you know.

Actually, the more player on the server, the more consistent results. Imagine BF4 was a 1v1 game mode. While testing one CPU the other player is shooting, flying jet and what not, and while testing the other CPU the opponent is AFK. That would put those CPUs on the opposite side of performance graph, even if the only difference between those is 100MHz clock frequency.

If you can't rule out variable part (single-player benchmark) then make sure to have as much samples as possible.

 

[PrincessFrosty]

Looks like it's using 6, not 8, and it looks like the 6300 is maxed out whereas the 8350 gets a pretty lofty clock speed advantage over it and two extra cores for background/os.

No doubt down to only 6 of the of the console cores available for actual game resources.

 

[Fjodor2001]

In a rudimentary way yes, it's far more accurate than saying the game supports 4, 6, or 8 threads. The game spawns WAY more threads than that.

A certain forum member hits a whole new level of ignorance when stating for example that next gen consoles will use 6 threads for games and reserve 2 cores. This is gibberish and clearly there is lack of basic understanding of how a thread scheduler works in combination with instruction pipelines and out-of-order execution.

Yeah, and it's really funny hearing people claim stuff like "one core is reserved for executing the OS".

I just wonder what those people think you would be doing with a one core CPU running Windows (or any other OS). Since the OS already would have the only CPU core exclusively allocated to it, you cannot do anything with your computer except "run the OS"... no core left for executing applications or games... :biggrin:

 

[PPB]

Yes, because reviewers dont' review 64P CQ online because they don't have balls.

It has NOTHING to do with being able to reproduce those results accurately over 10-20+ processors....
Since it's so easy, you're welcome to launch your own website and do it yourself you know.

1°: Didnt say it was easy. Neither I'm saying it's particularly hard, for the record. It just involves taking a longer sample of gameplay to reduce the impacts of individual anomalities that might show up in such gameplay. You know, the longer/bigger the sample, less the dispersion perceived.

2°:You dont need to bench 10-20 procs to show the general trend in a certain benchmark, at least not in current times.

3°: That last sad attempt at a fallacy doesnt deserve an answer :\

EDIT: About the post above, I remember those old times with single Core CPUs when you couldnt run anything at all, because you know, the SO had to be run on that single thread. Oh, wait :hmm:

 

[AtenRa]

Ehm, gameGPU.ru updated their CPU usage graph with FX9370 and FX9590

 

[ShintaiDK]

Ehm, gameGPU.ru updated their CPU usage graph with FX9370 and FX9590

65FPS to 74FPS with +300Mhz? Even you know thats wrong.

And any reason why they dont use 3rd and 4th generation Core CPUs? All the Intel CPUs in that chart is replaced. Some twice.

 

[inf64]

Can't be right, AMD is on top!

 

[Durvelle27]

Core usage on my FX 8350 @4.8GHz. Pegs at 64%

 

[frozentundra123456]

65FPS to 74FPS with +300Mhz? Even you know thats wrong.

And any reason why they dont use 3rd and 4th generation Core CPUs? All the Intel CPUs in that chart is replaced. Some twice.

Well, to be fair, the scaling from 4.0 to 5.0 ghz is almost linear, 25% increase in clockspeed for 23% increase in framerate, so the numbers are not totally impossible. But the 4.7 ghz value shows much poorer scaling, so one of them has to be off.

In any case, the fly in the ointment for AMD is that they are comparing basically a max overclocked FX8350 to stock intel hex cores, which could probably be overclocked by at least 20%.

 

[Erenhardt]

Well, to be fair, the scaling from 4.0 to 5.0 ghz is almost linear, 25% increase in clockspeed for 23% increase in framerate, so the numbers are not totally impossible. But the 4.7 ghz value shows much poorer scaling, so one of them has to be off.

In any case, the fly in the ointment for AMD is that they are comparing basically a max overclocked FX8350 to stock intel hex cores, which could probably be overclocked by at least 20%.

There is a note that FX9590 was simulated on FX9370 by increasing multiplier. So they overclocked to 5GHz. If the turbo is disabled it is constantly working on 5Ghz whereas FX9370 is 4,7 GHz with turbo which may decrease to lower frequency if needed.

 

[mindbomb]

the beta doesn't have an in game bench, does it?
A game like bf4 is pretty hard to bench otherwise.

So I buy the cpu usage descriptions, but I'm skeptical of the fps results.

 

[frozentundra123456]

There is a note that FX9590 was simulated on FX9370 by increasing multiplier. So they overclocked to 5GHz. If the turbo is disabled it is constantly working on 5Ghz whereas FX9370 is 4,7 GHz with turbo which may decrease to lower frequency if needed.

I wondered what the * meant. Somehow the page didnt translate for me.

 

[Arkaign]

There is a note that FX9590 was simulated on FX9370 by increasing multiplier. So they overclocked to 5GHz. If the turbo is disabled it is constantly working on 5Ghz whereas FX9370 is 4,7 GHz with turbo which may decrease to lower frequency if needed.

lol the pro review site didn't even want to waste their $$ on that idiotic 9590

As they felt like OC'ing the 9370 to beyond stock 9590 speeds (ie : no turbo downclock), wonder why they didn't OC the Intel procs that are basically OC maniacs? Ah, that would have results they don't want I suppose.

 

[Erenhardt]

lol the pro review site didn't even want to waste their $$ on that idiotic 9590

Whelp... they didn't bother with whole hasfail...

 

[Tempered81]

Ehm, gameGPU.ru updated their CPU usage graph with FX9370 and FX9590

More results to compare from pclab.pl

 

[monstercameron]

so even if we could accept those results and assuming they were accurate, then what would the effects of mantle be for an fx-8/9 gpu without the dx11 overhead? I know there isnt an answer to this question yet but it looks quite promising...

 

[soccerballtux]

so even if we could accept those results and assuming they were accurate, then what would the effects of mantle be for an fx-8/9 gpu without the dx11 overhead? I know there isnt an answer to this question yet but it looks quite promising...

mantle probably all marketing sadly IMHO

 

[Shakabutt]

More results to compare from pclab.pl

The polish benches are from Domination mode, infantry only 32players, Gamegpu.ru are conquest large.

Can't seem to understand the polish dudes, why bench a gamemode where the map , player count and game features are cut in half.

 

[BallaTheFeared]

I can't even play BF3 because the new battlelog doesn't install properly, woe is me.