[H] Battlefield 4 Video Card Performance and IQ Review

[wand3r3r]

Goalpost shift detected...

As mentioned, AMD subsidized the game. NV may bring the drivers closer, however it's pretty much the reverse of BF3 which AMD eventually got a lot closer to.

 

[BallaTheFeared]

You're digging yourself a deeper hole...

How is that possible when I'm not contesting the 290x in uber has a performance advantage over the Titan?

What do I have to lose?

I understand the conflict, MP is what matters, but MP results are too dynamic to measure reasonably unless they're so far out of whack that a clear problem seems to emerge.

There are obviously people unable to track what is being said.

I'm not saying the 290x in uber mode isn't faster than the Titan. Nor am I saying the results would play out any differently in SP. What I'm saying is testing GPU performance in a dynamic environment goes against the consistent concept which GPU benchmarking was founded upon.

There are no imaginary goal posts shifting, only a question of merit in the method. Not once have the results stating 290x in uber is faster than Titan been drawn into question. That includes SP or MP, it's funny getting attacked by the AMD supporters because they think I'm trying to say the 290x in uber isn't faster than Titan when that isn't even what is being discussed.

 

[Carfax83]

Not really surprised. As has been mentioned, AMD has been optimizing for this game longer than NVidia has, and the engine itself may also inherently favor GCN architecture.

BF4 is probably always going to be faster on AMD hardware, due to Mantle anyway.

I do agree somewhat with Balla on the subject of GPU testing in MP though. There is too much variation involved, especially with the game being in the state that it is, to have any kind of solid testing methodology that can accurately (or even semi accurately) reflect performance.

This is why the benchmarks seem to be all over the place, with NVidia leading in some, and AMD in others. I have no doubt though that the game does run faster on AMD hardware for aforementioned reasons.

Driver updates and patches will definitely alter performance down the road, as it did in BF3.

 

[Carfax83]

Goalpost shift detected...

As mentioned, AMD subsidized the game. NV may bring the drivers closer, however it's pretty much the reverse of BF3 which AMD eventually got a lot closer to.

If I recall, BF3 was an AMD evolved title as well, and AMD had the lead in that game early on. It wasn't until later than NVidia caught up after a few driver and patch updates.

 

[blastingcap]

If I recall, BF3 was an AMD evolved title as well, and AMD had the lead in that game early on. It wasn't until later than NVidia caught up after a few driver and patch updates.

IIRC NV had the lead from the get-go in BF3 and that it was actually AMD who had to catch up via driver updates etc.

 

[Silverforce11]

MP is too inconsistent, and the engine is exactly the same as SP.

Not sure why you'd want them to test graphics performance in such a variable situation of cpu necking and randomness.

Not that I think it will matter much, but evidently the 760 is faster than the 770 according to H.

Because only in MP will you ever find situations in which so many explosions and levolution occuring at the same time, on the entire 64MP map, that would cause major bottlenecks. Note the terrible min FPS and how [H] stated their in-game experience was poor due to many drops in performance. They suggest its an NV driver issue.

SP testing has NEVER reflected user results in MP in any BF game. All it did was artificially inflated people's expectations, which comes crashing down as they play 64 Conquest maps.

This is why [H] spent so much time playing the game, rather than doing canned benches. They come to a conclusion only after extensive gameplay sessions, which is why the 780ti did not make it and only why they are NOW releasing the MP test results, after many other sites already did their BF4 analysis (mostly with SP.. irrelevant).

[H] was the first site to claim AMD's CF was horrible, before all the FCAT FRAPS bandwagon, because they play games with their tests, not start a bench and come back later. If you consider [H] biased towards one brand over another, you are wrong. They are biased however, towards gaming enthusiasts.

 

[Randum]

Interesting...I just appreciate the screen captures and differences - helps cut down the troubleshooting when setting up your video card. I must admit, my 290 does kill this game very well...looks fantastic.

 

[BallaTheFeared]

The problem is that it could only occur on one card and not the other.

^ Proof of problematic testing method within the same review.

 

[Silverforce11]

The problem is that it could only occur on one card and not the other.

^ Proof of problematic testing method within the same review.

You missed the remarks relating to that. The 760 test ran through very light conditions on the level they played, whereas the 770 experienced more combat. It is the nature of MP benches, and it needs to come with editorial remarks relating to it and a conclusion after MANY hours spent PLAYING the game.

Its also why so few sites do it. Takes too much time and effort!

 

[BallaTheFeared]

You missed the remarks relating to that. The 760 test ran through very light conditions on the level they played, whereas the 770 experienced more combat. It is the nature of MP benches, and it needs to come with editorial remarks relating to it and a conclusion after MANY hours spent PLAYING the game.

Its also why so few sites do it.

I didn't miss it, it's proof of point.

Variation in runs can easily create an illusion of performance that doesn't hold up in more practical methods.

The swing for the 760 -> 770 is simply massive and a similar swing could have easily allowed Titan to outpace the R290x in uber... What do I gain from this type of randomness as a reader?

How can I reproduce his results? I can't, nobody can.


The 1600 Ram was sad to see as well.

 

[krumme]

With mantle bf4 driver amd cgn gpu will walk all over nv where it really hurts in mp battles. Because cpu bottlenecks for normal i5 starts to come into play at the same time as the serious gpu bottle necks.

Best way to avoid the situation: Play singleplayer on a nv gpu.

No hear no see. No problem.

 

[BallaTheFeared]

I was getting 86 or something on average with my 7950s on my i5.

I love the trolling with Mantle though, as if Nvidia doesn't have a counter

 

[AnandThenMan]

I love the trolling with Mantle though, as if Nvidia doesn't have a counter

And what is it?

 

[Silverforce11]

What you gain at [H], Balla, is what you've always gained if you read their editorial remarks and conclusion.

Kyle had this to add: "Conclusions are not drawn from the small amount of data that we see on the graphs." http://hardforum.com/showpost.php?p=1040388967&postcount=5

This is why after extensive gaming sessions, they reached a conclusion, AMD is performing better in raw performance as well as smoothness. There appears to be a NV driver issue that isn't resolved since beta, stuttery gameplay in MP.

It's the complete opposite of their BF3 review, where NV had better raw performance as well as smoothness, with the radeons suffering large drops with lots of explosions nearby. It took AMD a long time to fix their BF3 issues.

 

[Gloomy]

And what is it?

edit: nvm, don't want to get banned

 

[BallaTheFeared]

To much trolling for me.

 

[blackened23]

If I recall, BF3 was an AMD evolved title as well, and AMD had the lead in that game early on. It wasn't until later than NVidia caught up after a few driver and patch updates.

Nope. The GTX 580 decimated the 6970 in BF3 at release and it was not AMD gaming evolved. The 680 decimated the 7970 as well and no future AMD driver releases changed that - BF3 was better on kepler.

NV had the lead from day one.

Tangent time. (not addressed to you, carfax) On a side note, I find it absolutely hilarious how H is being accused of bias when just 6 months ago everyone was calling them nvidia biased. Hilarious. If you see a review you don't like apparently that website is...biased. H goes from being nvidia to AMD biased back and forth on a regular basis. HardOCP was the favorite review source to cite after the 680 launched, because their graphs favored Kepler IIRC. The AMD guys called H "nvidia biased". Christ. Having seen the history of H go back and forth from NV to AMD biased - It's so stupid it blows my mind.

Some games favor certain GPU architectures. Square enix favors AMD. Blizzard favors nvidia. Warner games usually favor nvidia. Apparently, BF4 favors AMD. Why people are getting their panties in a bunch over this? I have no idea. This may change with future driver revisions or it may not. It goes without saying that just because a card wins one game, that does not make it a better card. Would anyone here if offered a free 780ti or a free 290X go with the latter? Uh. Well I know I wouldn't. The 780 series is still a better card (MY OPINION) even if AMD wins BF4. Who cares, seriously. I'll still enjoy BF4 on my card (and I *do* enjoy it) even if it is 3 frames slower - and a future driver can change things. Or maybe not. But who cares.

 

[Teizo]

You missed the remarks relating to that. The 760 test ran through very light conditions on the level they played, whereas the 770 experienced more combat. It is the nature of MP benches, and it needs to come with editorial remarks relating to it and a conclusion after MANY hours spent PLAYING the game.

Its also why so few sites do it. Takes too much time and effort!

That is actually a good point as to why such benchmarks should not be done imho. They are unreliable. And if so, what is the point to even doing them?

 

[Gloomy]

To much trolling for me.

If the counter you allude to existed, then there would be nothing to 'troll' about. But unless you have some insider info, it's just delusion on your part-- which isn't wholly unexpected considering the two-faced approach to interpreting results you have displayed here.

In any case, feel free to enlighten us.

 

[AnandThenMan]

To much trolling for me.

If you're going to make a statement, then refuse to back it up with anything, some would say that is a form of trolling. So I'll ask again, what does Nvidia have in the works to counter Mantle? You made the claim remember.

On BF4, the game was made for MP, single player benches are pretty much useless IMO.

 

[Lepton87]

Their cards project a delusion-fueled force field that keeps reality at bay.

Mine doesn't Can I RMA it?

One way to make MP testing more repetitive and reliable would be to get a bunch of people replaying the same scenario, but that would be really tedious and probably expensive.

 

[BallaTheFeared]

Nah I'm good, this conversation is going nowhere fast.

 

[blackened23]

Hard|OCP is only biased if your favorite vendor loses.

Pretty good summary there. I know i'm not the only one that saw HardOCP being called "nvidia biased" nonstop after the 680 was launched. That context makes it all the more hilarious when they are now called AMD biased. I seriously don't get it.

Honestly, I really don't care if BF4 plays to AMD's so called "strengths" in their GPU architecture. To me playing BF4 a few frames faster doesn't make it the better card, because BF4 is only one game in a field of many games. Admittedly, I love BF4 but i'm not going to purchase a card based on one game, and i'm not going to think that the 290s are better than the 780s (I think the 780 is better) just because AMD wins BF4. If someone disagrees with this, hey, that's cool - If someone thinks this makes the 290 better, that's cool. I know some people have put 800+ hours into BF3 so this is important to them. But to me it is one game and it does not make the 290X a better card (MY OPINION) - there are so many more considerations in a GPU purchase.

 

[Slomo4shO]

On BF4, the game was made for MP, single player benches are pretty much useless IMO.

That is why [H] has a MP review coming up...

We have spent a lot of time with BF4 simply learning how to properly test for our readers. We will continue to test BF4 in a multi-player only environment because we understand that is how the huge majority of our readers will be using it. BF4 is being included in our regular testing suite so you will see a lot of coverage going forward. Surely you will see the GTX 780 Ti and BF4 testing here on [H]ard|OCP soon, both at multi-display resolutions and in multi-GPU configurations. We also have some specialty articles focused around BF4 that we will be publishing as well. So stay tuned on that front.

 

[BallaTheFeared]

Pretty good summary there. I know i'm not the only one that saw HardOCP being called "nvidia biased" nonstop after the 680 was launched. That context makes it all the more hilarious when they are now called AMD biased. I seriously don't get it.

They were also called AMD bias during the 40nm generation as well. :thumbsup: