***THE BF4 BENCHMARKS THREAD***

[Kaapstad]

I am rubbish at this game, I keep getting killed without seeing anyone coming.

Spoiler

Spoiler

Spoiler

Spoiler

1600p max settings

2 x GTX 690s @stock

3960X @4.9

 

[Silent_Scone]

No offence but I don't think you know what you're talking about. As I said, unless you care to elaborate with a technical explantion?



Doesn't sound likely. It's not that complex. The game just isn't optimised for us nVidia bums . Kaap doesn't need to show it - I don't think he's lying. He already said that SLI wasn't working because he was in windowed mode. Switching to full screen fixed it.

Well clearly I've a better understanding than you, otherwise you'd realise. Write a stern email to HardOCP if it's incomprehensible lol. You might be right the drivers aren't optimised as it is a beta after all. But the down low is that a reputable site has found that there is considerable stutter on the 2GB card that isn't present on the 4GB. It's staring you in the face mate - I won't say anything more on it :/

 

[Rusty0611]

Well clearly I've a better understanding than you, otherwise you'd realise. Write a stern email to HardOCP if it's incomprehensible lol. You might be right the drivers aren't optimised as it is a beta after all. But the down low is that a reputable site has found that there is considerable stutter on the 2GB card. It's staring you in the face mate - I won't say anything more on it :/

You haven't - read your driver/VRAM thing above. It doesn't make any sense.

I'm having to repeat myself now but they guessed at it being the 2GB limit - they didn't offer any technical explanation. And the differences in frame times were negligible and the difference in frame rate was next to nothing.

Just FYI - you aren't allowed to refer to HardOCP as reputable on this forum .

 

[cjph]

cpu matters, I would like to see,
4ghz 4 core without HT
4ghz 6 core without HT
4ghz 4core with HT
4ghz 6 core with HT
Or such on 64 map Multiplayer.
Both amd and Intel.
Mantle might make a huge difference down the line for amd.

Beta runs ok, better than Bf3 did for me
more fun gameplay makes a huge difference.

+1

I was not in the BF3 beta - was there a significant change in CPU/GPU performance between them, ie is there likely optimisation going to happen before release which might mean lowly i5 owners like me get a boost ?

There is an article somewhere in the early days of BF3 which benched various Intel CPUs with HT on and off, with the conclusion that HT was not necessarily a good thing at higher frequencies. Maybe the combination latest HT CPUs and W8 thread handling have reversed their conclusion.

 

[LtMatt]

Doesn't sound likely. It's not that complex. The game just isn't optimised for us nVidia bums . Kaap doesn't need to show it - I don't think he's lying. He already said that SLI wasn't working because he was in windowed mode. Switching to full screen fixed it.

But we already have reports from other users than optimisation is very good. Setter confirmed as much with his dual 670's, scotty confirmed that tri sli is working great for his titan. Think its pretty well optimised already. The only thing that can't be optimised is a 2gb frame buffer.

If kaap wants to run 1600p, Ultra +4AA with FOV 90 and video/bench it with a single 690 on 64 man Conquest large im all ears. I'm betting he'll be producing the same fps as swe clockers and other 690 users on tech forums and in battlelog.

How do you explain the 7950 boost and 760 results then? The 760 has higher minimum and average framerates : /

The problem here is neither cards have the grunt. The 7950 is also throttling no doubt as its a boost version.

The only results that we can learn from here is the Mult GPU results. Those numbers cannot be ignored. Single cards of titan and below just don't have the grunt to run this high res.

 

[Jono8]

The problem here is neither cards have the grunt. The 7950 is also throttling no doubt as its a boost version.

The only results that we can learn from here is the Mult GPU results. Those numbers cannot be ignore. Single cards titan and below just dont have the grunt to run this high res.

? But surely if vram was an issue, at that resolution, the 760 would have much worse minimum framerates than the 7950?

 

[Locky]

This is what I get for a whole game of conquest.

All the settings where on ultra with 4x msaa

7970's where at stock and cpu at 4.6

Frames Time (ms) Min Max Avg
56289 938641 57 62 59.969

 

[Rusty0611]

But we already have reports from other users than optimisation is very good. Setter confirmed as much with his dual 670's, scotty confirmed that tri sli is working great for his titan. Think its pretty well optimised already. The only thing that can't be optimised is a 2gb frame buffer.

If kaap wants to run 1600p, Ultra +4AA with FOV 90 and video/bench it with a single 690 on 64 man Conquest large im all ears. I'm betting he'll be producing the same fps as swe clockers and other 690 users on tech forums and in battlelog.

I'm not saying you're wrong - I'm saying there is too many doubts to make a definite conclusion. In principle I agree with your sentiment but I'm more neutral than you .

 

[Kaapstad]

But we already have reports from other users than optimisation is very good. Setter confirmed as much with his dual 670's, scotty confirmed that tri sli is working great for his titan. Think its pretty well optimised already. The only thing that can't be optimised is a 2gb frame buffer.

If kaap wants to run 1600p, Ultra +4AA with FOV 90 and video/bench it with a single 690 on 64 man Conquest large im all ears. I'm betting he'll be producing the same fps as swe clockers and other 690 users on tech forums and in battlelog.



The problem here is neither cards have the grunt. The 7950 is also throttling no doubt as its a boost version.

The only results that we can learn from here is the Mult GPU results. Those numbers cannot be ignored. Single cards of titan and below just don't have the grunt to run this high res.

I think my GTX 690s did ok check the fps in my post above.

 

[pgi947]

I will fire over an email now Paul. Sorry for my part in ruining this thread.

Thread isn't ruined, its just near impossible for someone to drop in and pick up some useful information

@Matt, correct, I had an avg drop of around 4 frames and min drop of around 10 frames going from smaller 32 to larger 64 player, vram usage only increased by around 80mb however

 

[shankly1985]

Tonester, i believe you owe me a copy of Battlefield 4. Whilst one card can not provide playable fps and exceed the 2gb limit at 1440p/1600p, two cards certainly can.






Expect the parcel force storm map to use even more vram and be more demanding than Shanghai imo as well.

http://www.sweclockers.com/artikel/17679-grafikprestanda-i-battlefield-4-beta/4#pagehead

Will you look at that bang on what I said hwinfo64 reported 2.7GB

 

[Silent_Scone]

This is making me want to crack out the 680 and borrow my little ones Classified lol.

Recommended specs are recommended.

 

[pgi947]

This is what I get for a whole game of conquest.

All the settings where on ultra with 4x msaa

7970's where at stock and cpu at 4.6

Frames Time (ms) Min Max Avg
56289 938641 57 62 59.969

Looks like you had vsync enabled locky?

 

[LtMatt]

? But surely if vram was an issue, at that resolution, the 760 would have much worse minimum framerates than the 7950?

If they don't have grunt required it won't matter how much vram they have.

Look at the 770 SLI vs 7970 crossfire results for an indication.

I'm not saying you're wrong - I'm saying there is too many doubts to make a definite conclusion. In principle I agree with your sentiment but I'm more neutral than you .

I'm saying it presents itself in more ways than one. I'm not saying you or anyone else is wrong. I guess this is as close to a settlement as we'll get.

I think my GTX 690s did ok check the fps in my post above.

I'd rather see a fraps bench for a whole round on 64 man conquest large as i posted above. Use one 690 as well like swe clockers and copy the settings used.

Your screenshots are nice But you're either looking at the sky or standing in an empty street. Really doesn't prove anything. I have high fps at those times too with 2x7950's. The problem arises as per swe clockers the more you play when things get hectic and the fps tanks on 2gb cards but does not happen on other cards. Its no use showing up with 2x690's standing in an empty street really im afraid.

 

[flopper]

+1

I was not in the BF3 beta - was there a significant change in CPU/GPU performance between them, ie is there likely optimisation going to happen before release which might mean lowly i5 owners like me get a boost ?

There is an article somewhere in the early days of BF3 which benched various Intel CPUs with HT on and off, with the conclusion that HT was not necessarily a good thing at higher frequencies. Maybe the combination latest HT CPUs and W8 thread handling have reversed their conclusion.

I found the BF3 beta and also the game of BF3, choppy, not clear distinct and this has to do with netcode.
as far, the BF4 beta has improved the netcode immensly this makes the delay much better, still is some but way better now.
The map is well designed and offers plenty of tactics, defend a flag, attack behind enemy lines, go heli, boat or take down C tower?

so this allows better gameplay even if I still dont like how the avatar is moving, BF2 had it good, Bc2 worse and BF3 hopeless, beta BF4 in between Bc2 and Bf3, so ok.

 

[Rusty0611]

Don't forget memory bandwidth as well - especially with AA applied at 2560 resolution.

 

[LtMatt]

Don't forget memory bandwidth as well - especially with AA applied at 2560 resolution.

Indeed Mr Galloway.

EDIT

Damn ive missed these vram debates.

 

[Jono8]

If they don't have grunt required it won't matter how much vram they have.

Look at the 770 SLI vs 7970 crossfire results for an indication.

That doesn't make any sense. It would still be expected to have the same amount of vram available so if 2gb wasn't enough I would fully expect to see lower framerates ( at least lower minimums) on the 760 compared with the 7950 boost.

All that graph could point to is a problem with SLI. The 690 has a worse minimum framerate than the 760 so something isn't right somewhere!

 

[Kaapstad]

I'd rather see a fraps bench for a whole round on 64 man conquest large as i posted above. Use one 690 as well like swe clockers and copy the settings used.

Your screenshots are nice But you're either looking at the sky or standing in an empty street.

Unfortunately I can not play the game without getting killed in seconds so someone else will have to do the video.

The point is the game runs very well on my GTX 690s but I did need to overclock the CPU to get the best out of them.

I also still have the option to overclock my GTX 690s if needed but the fps were more than enough at stock.

2gb of vram works fine.

 

[pgi947]

I am reinstalling the game now and will give it another go.

Make sure to double check video settings in game, it defaulted to windowed borderless for me, had to manually change it to full screen