Mantle vS 780Ti in BF4

humbug · 7 Feb 2014 at 15:11

Gregster, can you post the link of that review please, i want to put a few points to Ryan.

SS-89 · 7 Feb 2014 at 15:16

*Sigh* I don't even see the point of this thread, it's just become a cesspit for fanboys to argue with each other. Who cares? The nvidia lot aren't going to believe in mantle gains and the AMD lot are going to talk up Mantle like the second coming of Christ.

Frankly it's all meaningless at the moment as 1. Mantle is in early infancy and 2. No consistent benchmark to showcase difference in performance.

I've had a 780 which was a great card! I have 290x's now and again great cards (albeit noisy) but whilst I have seen reasonable gains (I'm hardly pushing the boat at 1080p) the 14.1 driver was too unstable so have reverted back to 13.12.

Gregster · 7 Feb 2014 at 15:18

Rusty0611 said:
Sees article which do not like. Attempt to question credibility of the author without proof. Repeat.

And when Mantle shows big gains:

Sees article which do like. Bigs up person for posting said article.

Like I said, I am no better.

People were saying "medium settings testing is wrong" I get one that shows ultra testing but now that is also wrong, as it is using a £500 CPU/they are biased/nVidia paid off/tested the wrong part of the game/didn't show Mantle in a perfect situation....

There is people who want the highest frames possible btw and will turn down settings to keep minimums really high. Not how I like it but fair play to those that do.

humbug · 7 Feb 2014 at 15:25

Gregster said:
And when Mantle shows big gains:

Sees article which do like. Bigs up person for posting said article.

Like I said, I am no better.

People were saying "medium settings testing is wrong" I get one that shows ultra testing but now that is also wrong, as it is using a £500 CPU/they are biased/nVidia paid off/tested the wrong part of the game/didn't show Mantle in a perfect situation....

There is people who want the highest frames possible btw and will turn down settings to keep minimums really high. Not how I like it but fair play to those that do.

Those Graphs are GPU bound, i know you understand that. the CPU power that they are using for the setting / setup makes Mantle irrelevant as the CPU on its own in DX is powerful enough to drive their testing method.

That makes the whole thing a Fallacy, Mantle is designed to overcome the CPU bottleneck, if you design a test where there is no CPU bottleneck then there is no difference with Mantle or not.

This is why you see Mantle working and doing its job on a single GPU with £100 to £200 CPU's but not on a single GPU where a £500 CPU is used.

Silent_Scone · 7 Feb 2014 at 15:26

I think for me at least, questioning very early beta results and saying they're 'wrong' just because they're not the same as others speaks words for itself as well as ignorance. It's just not any fun talking about this new tech when people are so hung up on defending their purchases. Bored of it. Tech Reports setup would be CPU limited with those settings, so least prove one article wrong before moving on to another, Humbug.

I think it's fair to say the 14.1 drivers aren't a fair representation of any standard, especially being beta. So it's not even worth taking any of it at face value, before you contemplate messaging authors telling them how to do their jobs.

Rusty0611 · 7 Feb 2014 at 15:28

Right humbug, so for the majority of people here with decent overclocked 4c/8t processors or above, Mantle offers little. Is that what you're saying?

I know the benefits for multi-GPU and low end CPUs already before you reel those off.

Frenzy · 7 Feb 2014 at 15:28

uksoldierboy said:
Forgive me i don't have the time to crop and crap, really need my full breaky! lol

alt+print screen captures the active window

Gregster · 7 Feb 2014 at 15:31

humbug said:
Those Graphs are GPU bound, i know you understand that. the CPU power that they are using for the setting / setup makes Mantle irrelevant as the CPU on its own in DX is powerful enough to drive their testing method.

That makes the whole thing a Fallacy, Mantle is designed to overcome the CPU bottleneck, if you design a test where there is no CPU bottleneck then there is no difference with Mantle or not.

This is why you see Mantle working on a single GPU with £100 to £200 CPU's but not on a single GPU where a £500 CPU is used.

Of course I understand it and I said prior to Mantle release that it will alleviate the bottleneck that shows on older/lesser CPU's. Mantle was bigged up by AMD as a massive performance booster over DX. Claims that it will ridicule a Titan. Sure, stick it with a Pentium 4 and a 290X/Titan, I am sure the 290X will ridicule a Titan but big bold claims from the guys at AMD are what makes me chuckle.

As for other games that are CPU bottlenecked, is Mantle backward compatible? Will it help in anything other than Mantle games? (don't answer, it is rhetorical) and the answer is no. So sure, if BF4 is your only game, go for a cheap AMD CPU and an AMD card and enjoy (when it finally gets fixed).

TheRealDeal · 7 Feb 2014 at 15:35

Gregster said:
And when Mantle shows big gains:

Sees article which do like. Bigs up person for posting said article.

Like I said, I am no better.

People were saying "medium settings testing is wrong" I get one that shows ultra testing but now that is also wrong, as it is using a £500 CPU/they are biased/nVidia paid off/tested the wrong part of the game/didn't show Mantle in a perfect situation....

There is people who want the highest frames possible btw and will turn down settings to keep minimums really high. Not how I like it but fair play to those that do.

I just think they should have tested at a point that's more cpu bound rather than gpu bound. My reason is dice and amd have said that's where the benefits will be seen. Testing in a gpu bound situation is not gonna show us if there's any benefits to using mantle hence there results and me not being surprised. There is enough users in here reporting decent gains to show that there are gains when its working properly. It's far to early for me in a buggy game to conclude anything atm.

Calin Banc · 7 Feb 2014 at 15:35

Those differences between Mantle and DX are there, it REALLY depends on what part of the game are you testing. I'll repost my own boost under mantle:
http://imgur.com/idv20MP
http://imgur.com/5OszVSU
http://imgur.com/UmODhlT

Yeah, that 53% I'm having for a single 7950@1135/1600 and 2500k@4,5GHz. From then I've added SSAO to the mix and the card can run it at above 60fps for than 95% of the time and even when it dips under, it's just in the 50fps. I'm NOT talking about averages, but minimums, that matter MORE than averages (although, the median - which is a more better way of seeing things, should be better than in dx).

But also, there are more subtle changes that you don't see in fps numbers, like input lag, which is lower under Mantle (and you see that in the original review posted on this thread) and the perceived difference is perhaps even higher, so that 50fps it feels almost like a butter smooth 60fps.

Mantle works, there is no real question about it, it just depends in what situations are you testing it. For enthusiasts that spend a huge amount of cash on CPU, then they probably go multi gpu as well. Just test on that uber high end i7, 2-3-4 cards and see the gains, because in THAT hardware setup / situation is what matters. There is no point in dumping $1k on a cpu and put it alongside only 1 gpu.

On a side note, r290 custom is probably the best solution. With some OC on it would get very close to a r290x/780ti at much less money.

JediFragger · 7 Feb 2014 at 15:36

MjFrosty said:
Still lol'ing at the idea of foam graph fights.

humbug · 7 Feb 2014 at 15:36

Rusty0611 said:
Right humbug, so for the majority of people here with decent overclocked 4c/8t processors or above, Mantle offers little. Is that what you're saying?

I know the benefits for multi-GPU and low end CPUs already before you reel those off.

i agree, although an 8 thread i7 still gives you a gain, all be it a small one, if your less fortunate; then you can get similar performance on a CPU costing half as much, where you can't in DX.

Using a £500 CPU to drive one GPU is just showing you where it does not benefit, what use is that? its not even journalism, i don't know what it is but it detracts from what Mantle is for and suggests something that is a blatant fallacy.
Whats important is to show their readers where it will benefit them. That will show their readership what its for.

Sugarhell · 7 Feb 2014 at 15:42

So amd guys spam graphs that benefits amd and nvidia guys spam graphs that benefits nvidia when the real problem is that most of these graphs sucks.Without proper testing + video with the run i will never gonna believe TR claims. Because i know DC is a fail especially in good multithread engines like frostbite because they already do it on engine level.Anyone who knows about graphic engines he will understand it.We are here and we discuss about bf4 SP ( who plays single player) and why 780ti is faster at 1080p @ high.

Gregster · 7 Feb 2014 at 15:43

humbug said:
i agree, although an 8 thread i7 still gives you a gain, all be it a small one, if your less fortunate; then you can get similar performance on a CPU costing half as much, where you can't in DX.

Using a £500 CPU to drive one GPU is just showing you where it does not benefit, what use is that? its not even journalism, i don't know what it is but it detracts from what Mantle is for and suggests something that is a blatant fallacy.
Whats important is to show their readers where it will benefit them. That will show their readership what its for.

The original post was showing how a 780Ti is faster than a 290X using Mantle. They used a 4770K CPU, which is a very good chip. I didn't hunt for a 780Ti vS Mantle using a very old/slower chip, as that wasn't what this article was about. What would be the point either? I was showing how a 780Ti in the tech reports article was faster than a 290X. Why would the tech report change their testing methodology unless they were wanting to show Mantle in a favourable light? And in which case, would make TheTechReport look biased.

The original article is a very good read, as is PCper. These are 2 of many sites I read and find of interest. I posted a review of the 780Ti vS a 290X running Mantle to give the nVidia guys a smile. No trolling, no baiting intended. So many threads about AMD this, that and the other recently, I felt us nVidia guys could do with some good news as well.

tommybhoy · 7 Feb 2014 at 15:47

What happened to 'trusting' user reviews, not so long ago that was the only data accepted as credible due to accepted reviewers bias(either way), now said user Mantle performance data is generally positive, it isn't credible data any more.

You couldn't make it up with the **** that goes in here, it's embarrassing tbph-or it would be if I post quotes.

Rusty0611 said:
Sees article which do not like. Attempt to question credibility of the author without proof. Repeat.

Sees cherry picked BM which don't like, refute the performance gain with cherry picked BM reporting null gain.

Repeat.

What's the overall percentage of gain/null Mantle performance in all the reviews to date?

There in lies the answer.

humbug · 7 Feb 2014 at 15:49

Gregster said:
The original post was showing how a 780Ti is faster than a 290X using Mantle. They used a 4770K CPU, which is a very good chip. I didn't hunt for a 780Ti vS Mantle using a very old/slower chip, as that wasn't what this article was about. What would be the point either? I was showing how a 780Ti in the tech reports article was faster than a 290X. Why would the tech report change their testing methodology unless they were wanting to show Mantle in a favourable light? And in which case, would make TheTechReport look biased.

you need to go back a couple of pages where i posted this link to show BF4 does not use DX Thread Batching.

Not only does the BF4 creator clearly say that "what TechReport say is the reason for their result" is actually not good performance at all, certainly compared with Mantle; But they don't even use it. So how can that be a reason for what is a very dodgy Graph to start with?

chief barker · 7 Feb 2014 at 15:52

Mantle or no Mantle.

AMD or Nvidia.

I still suck at BF4 :mad:

LtMatt · 7 Feb 2014 at 15:55

For what value user data is worth these days (not much anymore it seems

) vs review sites i saw about a 10%+ performance advantage using Mantle Vs DX11.1 with an i7 2700k @ 4.8ghz and a 290 Pro at stock. That was on single player.

Gregster · 7 Feb 2014 at 16:09

humbug said:
you need to go back a couple of pages where i posted this link to show BF4 does not use DX Thread Batching.

Not only does the BF4 creator clearly say that "what TechReport say is the reason for their result" is actually not good performance at all, certainly compared with Mantle; But they don't even use it. So how can that be a reason for what is a very dodgy Graph to start with?

I don't need to do anything. I started the thread and know exactly what Thetechreport has to say on it. They are very articulate and know what they are talking about.

Suarez7 · 7 Feb 2014 at 16:12

I don't think any of the AMD guys are trying to justify their purchase, without mantle the 290x is just as fast if not faster in BF4 anyway and £100 cheaper than a TI. What do they need to justify?

Maybe close this thread till Mantle is fully working as for the moment it's going round in circles.