Reviews - AMD Mantle: Battlefield 4 performance ruminations

[SkarKrow]

A 4670K is not really comparable to the A10... There's a reasonable price gulf and performance gap between the the two, an i3 is more comparable. I'd also have liked some numbers froma 6300 or 8350...

[nchester]

Clearly you have tested a GPU bound scenario, not a CPU bound scenario as claimed. If the settings were reduced from ultra to high you will achieve a CPU bound scenario which will properly demonstrate the potential of Mantle. With settings reduced to high an A10-7850K + 290x using DX will give you around 70fps. Switch to Mantle and you'll get around 130fps.

[Irien]

Whilst A10+290x does (to a point) remove the GPU bottleneck from the equation, it seems a bit odd to use a GPU which costs twice as much as the CPU+mobo combined. I think most gamers on some semblence of a budget would tend to spend roughly equiv of CPU+mobo on gfx hardware. For example, if the CPU+mobo came to about 150-180 quid, they might buy, what, a 270X/280X? Looking into it, I guess these aren't supported at the moment, so it is 290X or 260X which really are two extremes. Frustrating then - and really shows AMD up for repackaging old cards in the 270/280 range.

[Blackfudge]

used mantle on my 7970 on bf4 and it stuttered like hell so for now i'll wait until it gets fixed and then i'll post my opinion of it.

[keithwalton]

I'd like to ask why the 780TI score is even shown here?, for an article thats about Mantle vs DX11, having the 780Ti score doesnt seem to help show the difference atall, it afterall is completely different hardware and drivers.

To show that no matter how much you polish a T, at the end of the day you still have just a T.

On a more serious note AMD need to knuckle down and produce some halo products and not some 'my first pc' products as to be frank the majority of people buying a pc already have one and they want to upgrade it and the current line of AMD prodcuts are only marginally better than those from a few years ago (mainly cpu's)

Mantle does now make sense when all of the significant gains are with the weak AMD apu's

[HalloweenJack]

used mantle on my 7970 on bf4 and it stuttered like hell so for now i'll wait until it gets fixed and then i'll post my opinion of it.

and if you actually READ the AMD and Dice release - your 7970 is NOT SUPPORTED at this time.


so yeah - good game . muppet.

[HalloweenJack]

To show that no matter how much you polish a T, at the end of the day you still have just a T.

Mantle does now make sense when all of the significant gains are with the weak AMD apu's

you ever used one? thought not


http://www.3dmark.com/3dm11/7921043

http://www.3dmark.com/fs/1656103

and my kids pc before they got a new gfx card:

http://www.3dmark.com/fs/240666 (GDDR5 on the card)

and that's the older 5800K - the newer 7850K is faster


http://www.overclockers.com/amd-a107...ri-apu-review/ *scroll down*


P2413 for 3DM11 and 1512 for FireStrike...

[hiyayhi]

I would love to see mantle stay and also be ported across to linux, I hate windows compared to some linux distros but I have to run it for my most favoured games.

[Namix]

I'd like to ask why the 780TI score is even shown here?, for an article thats about Mantle vs DX11, having the 780Ti score doesnt seem to help show the difference atall, it afterall is completely different hardware and drivers.

To show that no matter how much you polish a T, at the end of the day you still have just a T.

On a more serious note AMD need to knuckle down and produce some halo products and not some 'my first pc' products as to be frank the majority of people buying a pc already have one and they want to upgrade it and the current line of AMD prodcuts are only marginally better than those from a few years ago (mainly cpu's)

Mantle does now make sense when all of the significant gains are with the weak AMD apu's

I think my point is that this whole article is about Mantle vs DX11 and not about 780Ti vs 290x.
We know the 780 performs better so why show its results when it doesnt even use Mantle.

Now if a 780Ti could use Mantle that would be fine as we could see if Mantle is better than Directx, making its score relevant. If this is how testing is to be done its just silly, why test a piece of hardware that doesnt even use the software your supposed to be evaluating it with?

I suppose you could argue its to see if the 290 gets such a boost it equals a 780 but again how is that fair when everything Mantle is in beta and the article makes no mention of it being so.

[Irien]

I suspect the reason for 780ti results was exactly *because* it performs better than 290X in reviews.

Right now, 780ti is "top dog" (and also "top price") in the single-gpu stakes. A very valid question, therefore, is "does 290X + Mantle" exceed the performance of "780ti + DX11". Given the hype about Mantle, it was entirely possible that it might unlock performance levels in Battlefield 4 that allow 290X to overtake 780ti. Or, as the article suggests, not. Mantle appears to be a nice boost for existing ATI owners, but not the "game changer" that AMD might hope.

Don't get me wrong, AMDs offerings represent much better value for money, but in terms of pure fps, it looks like the 780ti is still "top dog" for now, if you can afford it.

[WrATHZer0]

This benchmark is rather biased towards Intel. In normal benchmarks an A10 getting near to an i5 would be unthinkable and noticed the FX are not on the benchmark because they have 6-8 cores and would slaughter the Intel CPUs because of Mantles full use of cores.

Shameful benchmark really, I think people need to stop with the Intel love, Mantel shows that AMD 6-8 core CPUs have a lot more value for money because they're actually being used, DirectX doesn't use them, hence why Intel per core performance wins on 4 cores and Hyperthreading is useless in Mantle because they're not real cores, only to make the physical cores perform better.

[Namix]

I suspect the reason for 780ti results was exactly *because* it performs better than 290X in reviews.

Right now, 780ti is "top dog" (and also "top price") in the single-gpu stakes. A very valid question, therefore, is "does 290X + Mantle" exceed the performance of "780ti + DX11". Given the hype about Mantle, it was entirely possible that it might unlock performance levels in Battlefield 4 that allow 290X to overtake 780ti. Or, as the article suggests, not. Mantle appears to be a nice boost for existing ATI owners, but not the "game changer" that AMD might hope.

Don't get me wrong, AMDs offerings represent much better value for money, but in terms of pure fps, it looks like the 780ti is still "top dog" for now, if you can afford it.

I wouldnt say its a game changer yet, but considering its a 5-10 FPS boost with pretty good hardware in its first beta versions and game engine integration its getting there. With refinement and updates it certainly could be a game changer. Not to mention that with lower spec rigs the boost increases. I would also like to see some tests with true 6/8 core AMD CPU's to see how its utilizing the extra cores when compared to DX.

[Iain162]

I wouldnt say its a game changer yet, but considering its a 5-10 FPS boost with pretty good hardware in its first beta versions and game engine integration its getting there. With refinement and updates it certainly could be a game changer. Not to mention that with lower spec rigs the boost increases. I would also like to see some tests with true 6/8 core AMD CPU's to see how its utilizing the extra cores when compared to DX.

What most people seem to be misunderstanding is that mantle is a low level process designed to make better use of the CPU not the GPU, its not an AMD only feature since they announced that it will be open to any manufacture willing to implement it after development ends but knowing NVidia they will flat out refuse to support it in their future drivers after mantle advances past its early buggy stages unless forced to do so by developers showing interest..

The reason for higher spec hardware seeing lesser gains in mantle is because its not struggling as much with the bulky workload of DirectX, the main benefit to mantle becoming widely used in the development of games would be that were no longer forced to upgrade to a newer version of windows just to gain the support of newer versions of DirectX.

I'd be extremely happy if a system for games maintained by a 3rd party was widely used as it would hit Micro****e pretty hard, I know for a fact I'd never upgrade from windows 7 if this were the case.

[r4ck0]

We could say that they pumped the numbers and that we didn't get >60 FPS at ultra but wait a second: we usually spend extra 100$ on hardware to get 5-10 FPS more ?
Now we get this with an API (I have problems too and my card isn't yet supported) and we whine cause our 300$ card isn't better than an 780TI cause I want it to be? How dare you Mantle?!?!
Let's get real - we got 5-10 fps more for free. On same hardware.
Take that 100$ and celebrate.

[crossy]

Mantle does now make sense when all of the significant gains are with the weak AMD apu's

you ever used one? thought not

KW just seems to be trotting out the same line that I've seen elsewhere, namely that Mantle only really makes sense where you've got a computation-power-challenged cpu paired with a "good" GPU. I don't think he was necessarily suggesting that overall APU's are poor - I think it's well accepted that while the Intel parts are better for crunching numbers, an AMD APU will streak into the lead when you're shunting the pixels around.

and if you actually READ the AMD and Dice release - your 7970 is NOT SUPPORTED at this time.

Care to explain how the R9 280X IS supported when it's (reportedly) the same core as the 7970? At the risk of being rude, I think you meant that the old 7970 isn't officially supported. So if it doesn't work well then don't be surprised. In fact if you go to Mantle renderer now available in Battlefield 4 you'll see

Test case 2: Standard 64-player multiplayer
CPU: AMD FX-8350, 8 cores @ 4 GHz
GPU: AMD Radeon 7970 3 GB
Settings: 1080p ULTRA 1x MSAA
OS: Windows 8 64-bit
Level: Siege of Shanghai
Level was tested with 64 “pseudo players” that we have for our own internal testing that simulates heavy game workload that we have in multiplayer in order to get more deterministic results compared to full real multiplayer. 64 players on the large Battlefield levels is really demanding of the CPU so this test case is primarily CPU-bound.
Result: 18.87 ms/f -> 15.08 ms/f = 25.1% faster

so yeah - good game . muppet.

uncalled for?

[ChinesePate]

since the whole mantle came out, it's the first time I cannot play BF4. seriously, up until now I didn't have absolutely any issues with the game. since the (I guess) the new driver, it's all broken. 9/10 games exits to desktop with "GPU Driver hung" message, no matter if I use DX11/Mantle

4670k+R9 290 Tri-X