BF3 CPU bottleneck? Really?

[BTRY B 529th FA BN]

I feel you, I really do.
But without a methodology for proof, it is what it is.
If you can think up a way to demonstrate, say on a frame pr. sec basis. one setup to another, that would be groundbreaking.

Ah, i don't believe it

 

[moonbogg]

Me too. People keep laying around on maps instead of actually playing so the numbers in this thread are useless.

It provides a comparative performance increase that carries over into mins and averages during active gameplay.

 

[BD231]

Me too. People keep laying around on maps instead of actually playing so the numbers in this thread are useless.

It provides a comparative performance increase that carries over into mins and averages during active gameplay.

Not really, 60/70 fps avg seems to be upper limit when playing even in sli. Sitting around doing nothing wont explain that.

 

[Termie]

how are you that cpu limited in Batman AC? I can get better results running just a single core of my 2500K.

I'd like to see that. I've run the Batman AC benchmark probably 50 times while testing. No way you're gonna run it on one core.

Guys, hyperthreading works. Moonbogg has done very thorough testing. BF3 is really difficult to benchmark so his technique makes sense.

 

[Rvenger]

how are you that cpu limited in Batman AC? I can get better results running just a single core of my 2500K.

I'd like to see the application launch one 1 core.

 

[moonbogg]

Not really, 60/70 fps avg seems to be upper limit when playing even in sli. Sitting around doing nothing wont explain that.

Sitting around doing nothing in the corner of the map overlooking everything is the most CPU intense the game gets. I have tested during active gameplay with averages. They are posted on this thread a few pages back. The increases carry over. If the increase is from 55 to 60fps, it won't be obvious during gameplay unless you bench and take averages, which I have taken the time to do. Sounds like you are making guesses and assertions.

 

[toyota]

I'd like to see that. I've

well I had to look through your posts to find the exact settings you used. you had physx on which I did not. my 2 cores at 4.4 got 45 fps at your settings but that is also with a weaker gpu than you have. 1 core could only get 23 fps though using the bench at the settings you were using. btw you cant really uses DX11 high settings. if its in DX11 then its very high or extreme. if you select high it will then disable DX11 features. now you can leave the setting on high and then select the DX11 features but technically it will be very high as that is just a glitch in options menu.

 

[toyota]

I'd like to see the application launch one 1 core.

it will play fine with one core in DX9 especially without physx. on medium DX9 I can average 125 fps with just a single core of my 2500k. I had tested that because some guy was thinking it was his i3 making the game slow even on DX9 medium which of course was false.

 

[Termie]

well I had to look through your posts to find the exact settings you used. you had physx on which I did not. my 2 cores at 4.4 got 45 fps at your settings but that is also with a weaker gpu than you have. 1 core could only get 23 fps though using the bench at the settings you were using. btw you cant really uses DX11 high settings. if its in DX11 then its very high or extreme. if you select high it will then disable DX11 features. now you can leave the setting on high and then select the DX11 features but technically it will be very high as that is just a glitch in options menu.

Ok, that makes sense. Your two cores (with HT?) at 4.4 are certainly enough to run this game. Did you try with and without HT?

Didn't know about the "high" glitch. I actually set it to "very high" after I got done benching, and it only dropped the fps by 2fps, so I'd be very surprised if DX11 had been completely off before.

BTW, just looked up the specs on that 560 SE you have. I was surprised to see that it really does spec a notch below the 460-768, despite selling for more than the 460 did a year ago. Too bad you sold that 570, huh?

 

[toyota]

Ok, that makes sense. Your two cores (with HT?) at 4.4 are certainly enough to run this game. Did you try with and without HT?

Didn't know about the "high" glitch. I actually set it to "very high" after I got done benching, and it only dropped the fps by 2fps, so I'd be very surprised if DX11 had been completely off before.

BTW, just looked up the specs on that 560 SE you have. I was surprised to see that it really does spec a notch below the 460-768, despite selling for more than the 460 did a year ago. Too bad you sold that 570, huh?

no HT on the 2500k. now for DX11 on max settings with physx it is helpful to have to have all cores of the 2500k or at least 3 of them. really the game is horribly optimized in DX11 and will drop framerates randomly even when under relatively little gpu load. heck even in DX9 it will just drop at times when flying around.

yeah I wish I still had the 570. the 560se is basically dead even with the gtx460 768mb though so its not too bad in games. plus it cost me less than 100 bucks and makes for a good back up card as my only other card is an 8600gt which is abysmal.

 

[Termie]

no HT on the 2500k. now for DX11 on max settings with physx it is helpful to have to have all cores of the 2500k or at least 3 of them. really the game is horribly optimized in DX11 and will drop framerates randomly even when under relatively little gpu load. heck even in DX9 it will just drop at times when flying around.

yeah I wish I still had the 570. the 560se is basically dead even with the gtx460 768mb though so its not too bad in games. plus it cost me less than 100 bucks and makes for a good back up card as my only other card is an 8600gt which is abysmal.

Oh yeah, duh, no HT. Sorry about that. Still, two of your cores at 4.4 are pretty serious firepower for the game, and I'm not too surprised they could keep up with four i7-860 cores @3.0.

How does the 560SE overclock? It has fewer cores but a higher default clock, so at stock it's probably even with a 460-768 (has the same memory bus). I got lucky with the OC on my 460, so it's stuck around a lot longer than I thought it would...

 

[toyota]

Oh yeah, duh, no HT. Sorry about that. Still, two of your cores at 4.4 are pretty serious firepower for the game, and I'm not too surprised they could keep up with four i7-860 cores @3.0.

How does the 560SE overclock? It has fewer cores but a higher default clock, so at stock it's probably even with a 460-768 (has the same memory bus). I got lucky with the OC on my 460, so it's stuck around a lot longer than I thought it would...

well mine comes at 776 but its not listed as being overclocked yet the official clocks are 746 from what I understand. it will almost do 900 on the stock 1.000 voltage. I am not sure how high it will really go but 900 at 1.013 would be the highest I would ever want to run it long term though. personally I keep it at 858 and stock voltage on a daily basis though.

as for the memory, it will do 4592 perfectly fine which is the max that the slider goes to in precision. I wish the slider went further so I could check the actual max memory oc it could handle. I do run it at 4400 though as I don't like to push things too hard.

 

[hrrnghhh]

FYI moonbogg, no drops on TV Station as well in 64 player server

 

[Black Mesa]

I suggest with that level of hardware you NEED a 2560x1440 monitor then as an added bonus on top of the resolution boost it will move the bottle neck onto your GPUs.

 

[Termie]

well mine comes at 776 but its not listed as being overclocked yet the official clocks are 746 from what I understand. it will almost do 900 on the stock 1.000 voltage. I am not sure how high it will really go but 900 at 1.013 would be the highest I would ever want to run it long term though. personally I keep it at 858 and stock voltage on a daily basis though.

as for the memory, it will do 4592 perfectly fine which is the max that the slider goes to in precision. I wish the slider went further so I could check the actual max memory oc it could handle. I do run it at 4400 though as I don't like to push things too hard.

Wow - the memory controller on the 500 series must be improved, because I've never seen a memory OC like that on a 400 card. I used to run mine at 4000 but backed off to 3800 because I wasn't sure 4000 was helping.

Anyway, back to HT...I think it would be great to have a new thread (if someone wanted to start it) to get people to benchmark whatever games they have with HT on and off. As usual, this is a pretty relevant topic that is left completely untouched by professional reviewers.

 

[Akantus]

Wow - the memory controller on the 500 series must be improved, because I've never seen a memory OC like that on a 400 card. I used to run mine at 4000 but backed off to 3800 because I wasn't sure 4000 was helping.

Well, I run my 460 at 4200MHz on memory, could probably go higher, but I don't want to limit my core OC, and without more core OC higher memory won't help me.
Sorry for OT

Yea, it would be nice to see how modern games behave with HT on/off.

 

[moonbogg]

FYI moonbogg, no drops on TV Station as well in 64 player server

Must be nice! I need to stop complaining though, really. Thats good that you can PWN without any distracting slowdown.

 

[poohbear]

Hey, i'm a pretty avid BF3 player and i used to run FRAPS alot to see the FPS variance just before i upgraded.

I went from a PhenomII X6 1100T @ 3.8ghz to a 2500k @ 4.4ghz (all other settings the same), with Ultra settings @ 1080 I DID NOT NOTICE A SINGLE FPS DIFFERENCE IN 64 PERSON MP with my 670 GTX. It was really disappointing to say the least, but definitely a learning experience.

People were talking about how i'd notice a jump in minumum FPS, that's complete BS! I didn't notice any difference at all with Ultra settings (still dipped to ~30ish FPS when alot of crazy stuff was happening, but avg is 60fps and it was the same with my X6).

in OLD RTS games(Dawn of War 2) i noticed a jump in performance for sure. But even modern RTS games like Shogun 2 fully maxed out i hardly noticed any difference at all, real shame but it just goes to show alot of people on these forums that advocate how one CPU is better in higher resolutions with completely GPU dependant features turned on (like AA & ultra quality) don't have a clue what they're talking about.

 

[moonbogg]

Hey, i'm a pretty avid BF3 player and i used to run FRAPS alot to see the FPS variance just before i upgraded.

I went from a PhenomII X6 1100T @ 3.8ghz to a 2500k @ 4.4ghz (all other settings the same), with Ultra settings @ 1080 I DID NOT NOTICE A SINGLE FPS DIFFERENCE IN 64 PERSON MP with my 670 GTX. It was really disappointing to say the least, but definitely a learning experience.

People were talking about how i'd notice a jump in minumum FPS, that's complete BS! I didn't notice any difference at all with Ultra settings (still dipped to ~30ish FPS when alot of crazy stuff was happening, but avg is 60fps and it was the same with my X6).

in OLD RTS games(Dawn of War 2) i noticed a jump in performance for sure. But even modern RTS games like Shogun 2 fully maxed out i hardly noticed any difference at all, real shame but it just goes to show alot of people on these forums that advocate how one CPU is better in higher resolutions with completely GPU dependant features turned on (like AA & ultra quality) don't have a clue what they're talking about.

Sorry to hear that. A single 670 can barely handle BF3 as you have seen. Get a second one and your CPU will shine more. 30FPS dips are very low and it doesn't take much of a CPU to pull that off.

 

[Blastman]

8 Threads is where its at for this game.

1 core = 1fps
2 cores = 20fps
3 cores = 35fps
4 cores = 50fps
5 cores = 53fps
6 cores = 56fps

4 threads = 50fps
6 threads = 57fps
8 threads = 63fps
10 threads = 64fps
12 threads = 64fps

Your numbers are quite similar to what this guy got in his testing of BF3 on this youtube video. Scroll to 16:00 minutes in the video to see his CPU performance chart.

2C/4T was faster than 3C/3T, so HT really made a big difference for 2 cores. And 4C/8T was faster than 4C/4T although the performance improvement past 4C/4T was not huge, more than 4 threads still helped.

 

[Akantus]

Hey, i'm a pretty avid BF3 player and i used to run FRAPS alot to see the FPS variance just before i upgraded.

I went from a PhenomII X6 1100T @ 3.8ghz to a 2500k @ 4.4ghz (all other settings the same), with Ultra settings @ 1080 I DID NOT NOTICE A SINGLE FPS DIFFERENCE IN 64 PERSON MP with my 670 GTX. It was really disappointing to say the least, but definitely a learning experience.

People were talking about how i'd notice a jump in minumum FPS, that's complete BS! I didn't notice any difference at all with Ultra settings (still dipped to ~30ish FPS when alot of crazy stuff was happening, but avg is 60fps and it was the same with my X6).

in OLD RTS games(Dawn of War 2) i noticed a jump in performance for sure. But even modern RTS games like Shogun 2 fully maxed out i hardly noticed any difference at all, real shame but it just goes to show alot of people on these forums that advocate how one CPU is better in higher resolutions with completely GPU dependant features turned on (like AA & ultra quality) don't have a clue what they're talking about.

Well, it should come as no surprise that when you are GPU bound, better CPU won't help. Try lowering some graphical settings (MSAA?) and see if it help. I would say 30fps is too low for i5 at 4.4GHz.

But you had X6 which is very good CPU for BF3 and 2550k is side/marginal upgrade at best for this game. (Unless you would OC to 5.0GHz)

 

[poohbear]

Sorry to hear that. A single 670 can barely handle BF3 as you have seen. Get a second one and your CPU will shine more. 30FPS dips are very low and it doesn't take much of a CPU to pull that off.

huh? no it can handles it great, i avg 60+ fps with Ultra settings. where'd you get it can "barely handle BF3"? I'm gaming @ 1080p here. It's just when all kinds of crazy stuff happens, i mean a thousand different things happening on a large 64 player map, the lowest it'll dip is to 30fps with my 2500k @ 4.4ghz (and even then just for a second or 2). In Close Quarters i hardly ever see it drop below 50fps. My point was NOT my GTX 670 which is a beast, its that a 2500k is NOT better than a PhenomII X6. I think a GTX 670 is the best card u can get for 1080+!

 

[2is]

Yeah I'm to sure where that claim of a 670 can barely handle bf3 is coming from either. Arguably the 2nd fastest single GPU card out right now will handle bf3 just fine @ 1080p

 

[moonbogg]

It's just when all kinds of crazy stuff happens, i mean a thousand different things happening on a large 64 player map, the lowest it'll dip is to 30fps with my 2500k @ 4.4ghz (and even then just for a second or 2).......its that a 2500k is NOT better than a PhenomII X6.

Thats when a card's power is supposed to shine, when all the crazy stuff happens on a full map. If it hits 30fps when things get busy, then that counts as barely handling the game in my opinion. Your opinion is obviously very different.
Regarding the 2500K not being better than an X6, I have no comment as no comment would suffice for you.

 

[2is]

Thats when a card's power is supposed to shine, when all the crazy stuff happens on a full map. If it hits 30fps when things get busy, then that counts as barely handling the game in my opinion. Your opinion is obviously very different.
Regarding the 2500K not being better than an X6, I have no comment as no comment would suffice for you.

When all kinds of "stuff" happnes is usually when the CPU kicks in, not the GPU. The main exception being smoke.