bit-tech.net

Battlefield 4 Performance Analysis

Comments 1 to 25 of 35

Reply
Shirty 27th November 2013, 10:25 Quote
Nice article, reminiscent of old. Well done Matthew and well done Bit-tech ;)

Before all the fanbois arrive an ruin this comments thread, is it just me or is there almost no discernible difference between the high and ultra screenshots? I can only imagine that the differences are even less noticeable in-game.
Corky42 27th November 2013, 10:31 Quote
Quote:
The main pattern we've observed is that Battlefield 4 tends to favour AMD's hardware

While this is true on Windows 7, apparently the gap is narrowed when tested on Windows 8.1.
With R9 290X at around 3% faster on 8.1, and the GTX 780 at around 6% faster.
Mentai 27th November 2013, 10:40 Quote
This article came at a great time considering the new consoles have come out. It's very reassuring to see that I'll be getting better frame rates than the ps4 on well made multi platform games. This certainly wasn't the case when xbox360 came out unless you had really top tier gear. I think in the next few years there will be a big shift to higher than 1080p gaming as AMD/Nvidia continue to put out better gpu's and the demands of games remain relatively static. Until 4k TVs become affordable though I will be sticking with what I have. I'm looking forward to my current rig lasting me a long time.
jinq-sea 27th November 2013, 11:11 Quote
A most enjoyable (and informative) read - thanks!

Interesting results as a result of CPU core disabling.
Lenderz 27th November 2013, 11:42 Quote
Would really like to see the impact of Mantle on this review when the patch comes out in December supporting it.

Would help me decide what to do about my next card, still deciding if I'm due an upgrade and if I get one if I stick with Nvidia or go back to AMD.
saxovtsmike 27th November 2013, 11:54 Quote
awsome test, but sadly it´s lacking Vmem impact
As I can say my 680 2gb is at it´s maximum avaliable mem with 1950 @ 1440p
sotu1 27th November 2013, 11:54 Quote
I'm running an AMD 6900 gpu from the days of BF3 and i'm still clocking an average of 45 on high. surprisingly good for a 2 year old pc rig.
Shirty 27th November 2013, 11:56 Quote
Quote:
Originally Posted by saxovtsmike
awsome test, but sadly it´s lacking Vmem impact
As I can say my 680 2gb is at it´s maximum avaliable mem with 1950 @ 1440p

Looking at the results I'd be sweating if I had less than 3GB, although the top end Nvidia cards from last generation still seem to be competitive.
xaser04 27th November 2013, 12:03 Quote
Fantastic article guys, a real return to form for Bit.

Love seeing the brutal effect of ultra settings in Eyefinity / Surround. The 64 ROP backend of the 290 GPU's is really showing its muscle here (noticably higher minimums).

Good to see that my GTX780M will be able to manage a mixture of Medium and High settings whilst retaining a constant 60FPS (can probably get away with the majority on high judging by the results).

(GTX780M is a 1536SP "Full" GK104 clocked at 849/5000 although I run mine at +100/+250 for more demanding titles. It offers roughly GTX670-GTX680 performance depending on the title).
Pookie 27th November 2013, 12:12 Quote
Based on the cards tested @ 1920x1080 with ultra settings which is comparable to a 6970?
rollo 27th November 2013, 12:16 Quote
your core disabling tests were funny, So no difference at all between them hmm. Thought this was ment to support like 8 cores ( Andy seems to think it does anyway) your tests kinda suggest it does not.
xaser04 27th November 2013, 12:22 Quote
Quote:
Originally Posted by Pookie
Based on the cards tested @ 1920x1080 with ultra settings which is comparable to a 6970?

An HD6970 sits somewhere between a HD7850 and HD7870. IIRC it is a bit closer to the latter than the former.
Pookie 27th November 2013, 12:32 Quote
Quote:
Originally Posted by xaser04
An HD6970 sits somewhere between a HD7850 and HD7870. IIRC it is a bit closer to the latter than the former.

So i can do ultra @1080p and get playable results! Not bad :D Isnt the minimium FPS considered to 25 before things get too choppy?
Harlequin 27th November 2013, 12:34 Quote
would like to have seen this test on an AMD FX83xx rig as well
xaser04 27th November 2013, 13:23 Quote
Quote:
Originally Posted by Pookie
So ican do ultr @1080p and get playable results! Not bad :D Isnt the minimium FPS considered to 25 befor thing get too choppy?

Completely subjective. The consistancy of the framerate (timing of the individual frames) is far more important than absolute FPS number.
dhughes 27th November 2013, 13:37 Quote
This article is great, and interesting to see that the HBAO doesn't have as much impact on performance as it did in 3. Personally I run texture quality and filtering on Ultra and the rest on High but may try all on ultra without MSAA after this. The article below made me feel better about my 2Gb 670. If it's reflective of other 2Gb vs 4Gb setups in BF4 then I'm not going to bother upgrading just yet and it is very interesting but I'm willing to accept it could just be the cards reviewed.

http://www.legitreviews.com/gigabyte-geforce-gtx-760-4gb-video-card-review-2gb-4gb_129062/4
Spreadie 27th November 2013, 13:49 Quote
GTX 670 with high detail produces playable rates @ 2560x1600...

...it almost makes me want to buy the game. Oh, no... wait... yeah, that was a lie.

[edit] 6,000 posts... that's quite of lot of BS right there!
Fishlock 27th November 2013, 14:25 Quote
Brilliant article, thank you.
Shirty 27th November 2013, 14:27 Quote
Quote:
Originally Posted by Spreadie
GTX 670 with high detail produces playable rates @ 2560x1600...

...it almost makes me want to buy the game. Oh, no... wait... yeah, that was a lie.

[edit] 6,000 posts... that's quite of lot of BS right there!

If it makes you feel any better (it won't), I won't be touching the game either. I might acquire it at a later date if I wish to benchmark something, but the game itself does nothing for me.

And for the record I've enjoyed a small number of your BS posts.
Hustler 27th November 2013, 14:32 Quote
Nicely done but kind of pointless when you consider the only frame rate that matters with this game is multi player.

The single player campaign is an after thought, which many won't even bother with, I know I haven't even played the campaign in Bad Company 2 never mind BF 3 or 4.

People need to know what hardware is needed to play @1080p med/high/ultra detail with 64 players and get 50-60fps...anything else is irrelevant with a game like this.
Shirty 27th November 2013, 14:43 Quote
I think the point of this is that BF4 will be used as a GPU benchmark for the next year or two, and it's impossible to replicate multiplayer results accurately, so think of this as an introduction to the BF4 benchmark.
Hustler 27th November 2013, 14:51 Quote
Quote:
Originally Posted by Shirty
and it's impossible to replicate multiplayer results accurately

Which is why I think it's a pointless benchmark to use, when the results won't be applicable 99% of the time to the use it's being put to.
Asouter 27th November 2013, 14:57 Quote
Not the typical yadda, yadda, yadda GPU test ... thanks for posting those screen shots for comparison. very useful ....
Shirty 27th November 2013, 14:57 Quote
Quote:
Originally Posted by Hustler
Which is why I think it's a pointless benchmark to use, when the results won't be applicable 99% of the time to the use it's being put to.

You're applying the wrong logic. It isn't possible to benchmark the multiplayer part of the game reliably and consistently, because there are too many variables.

Therefore the only way to produce consistent benchmarks is to use the single player part of the game. I appreciate that the results of these tests will not strictly ape the multiplayer results, but they actually provide better guidance than multiplayer because they are repeatable.

This feature is as much (if not more) about the cards as it is about the game. If a card performs well in SP, then the chances are higher that it will perform well in MP too.
GeorgeStorm 27th November 2013, 15:06 Quote
Quote:
Originally Posted by Shirty
You're applying the wrong logic. It isn't possible to benchmark the multiplayer part of the game reliably and consistently, because there are too many variables.

Therefore the only way to produce consistent benchmarks is to use the single player part of the game. I appreciate that the results of these tests will not strictly ape the multiplayer results, but they actually provide better guidance than multiplayer because they are repeatable.

This feature is as much (if not more) about the cards as it is about the game. If a card performs well in SP, then the chances are higher that it will perform well in MP too.

+1

I wouldn't take the raw numbers, as they aren't going to reflect multiplayer gaming very well.
But it should be a good comparison between cards, so you can see how cards compare to each other, even if the raw data doesn't apply, it will be good for relative performance.
Log in

You are not logged in, please login with your forum account below. If you don't already have an account please register to start contributing.



Discuss in the forums