The AMD Mantle Thread

[mikk]

It's mostly better threading.

Do you have a source?

 

[krumme]

There is: pclab.pl

Excactly. There is nothing. No one can replicate what they have done prior either.
As one of our native Polish speaking members have said they just hate AMD and have gone completely crazy.

I knew 2 month ago, this stupid site would polute the entire discussion. And surely they do.

Take a kabini 2GHz and run it in dx under windows 8.1 on a 64 player map with a nv 660 gfx. Surely a wonderfull experience or what?

The improvements is not remotely comparable.

 

[jj109]

Excactly. There is nothing. No one can replicate what they have done prior either.

No one tries to reproduce what they found. All the big name review sites are terrified of rocking the boat and we end up with the same two benchmark types:

1. An overclocked i7 + varying GPU to show only GPU-bound behavior.
2. Varying CPU + one GPU to show only CPU-bound behavior.

Oh hey, let's look at what happens when reviewers do both at the same time:


http://www.hardwarepal.com/battlefield-4-benchmark-mp-cpu-gpu-w7-vs-w8-1/8/
http://techreport.com/review/25995/first-look-amd-mantle-cpu-performance-in-battlefield-4/2


This is where someone replies to me with the GPU-bound [H] review with a 4.8 GHz i7, right? Right?

Better yet, why don't someone here just try running this sequence with 64 players? Should be simple to debunk an obviously false claim.

 

[guskline]

Being a user of all Nvidia cards at present, based upon the tone of this thread I should be in "bash mode" about Mantle. I'm not. I'm excited for this new technology. It's pushing the envelope to improve gaming performance. That helps all of us.

 

[DrDoug]

Being a user of all Nvidia cards at present, based upon the tone of this thread I should be in "bash mode" about Mantle. I'm not. I'm excited for this new technology. It's pushing the envelope to improve gaming performance. That helps all of us.

I have the same attitude, whatever pushes gaming performance forward is good for everyone. I'm no brand loyalist/team player but then again I didn't watch the Superbowl. Currently I have Nvidia in my laptop and AMD in my desktop and in the past AMD in my laptop and Nvidia in my desktop. When one company makes a good move the other has to one up them, making us all winners in the end. I'm interested in performance moving forward, not brand loyalty.

Brand loyalty makes you blind to other possibilities. If Mantle succeeds then Nvidia will have to belly up to the bar and pour us a better drink.

Win-win.

 

[el etro]

High diffs. Seems to be Win 8.1 diff.





Win7/3930K@5Ghz test at Q4 2012, BF3 Multiplayer.

All review sites should substitute the BF4 Single Player tests with the Multiplayer tests.

 

[krumme]

All review sites should substitute the BF4 Single Player tests with the Multiplayer tests.

Ofcourse because that what people do. And they dont tend to play on empty servers. Its kind of boring.

And then the reviewers could just play the damn game. I know its doesnt give fancy graphs. But its pretty aparent actually gaming that Mantle improves the smoothness by a long stretch and also helps min fps. At least outside of 1000 usd heavily oc cpu on single gpu configurations - whereever they exist.

If i recall correctly i was the first in this thread after playing for an hour or two, who wrote about the smoothnes but also about the stuttering bug. It doesnt take a long education to game bf4 and trust your impression

H did come to excact the same conclusion because they take the time to actually play the games. Therefore they originally was also the first to notice and critizise CF total lack of smoothness.

Its not just about the added 50-80% extra performance when you are at critical places and situations on the big maps on my ib i5 4c 4t 4ghz. Thats not the important part. When i dump into a tank within a few seconds i can feel the extra smoothness. Thats what i care for. Then its just nice that i can confirm on the flatter graphs its also the behavior of mantle.

 

[Shivansps]

But what are the results with other cards? 280x, 270x, HD7950, HD7870, HD7850, HD7770, HD7750, etc? there are any reviews with that? 290x performance alone its highly insignificant.

 

[Cloudfire777]

But what are the results with other cards? 280x, 270x, HD7950, HD7870, HD7850, HD7770, HD7750, etc? there are any reviews with that? 290x performance alone its highly insignificant.

Putting out a review with most of those cards would put a quick damper on the whole Mantle reviews.

Unless you run those with a $50 CPU/APU.

 

[el etro]

Its simple to test Mantle effectiveness: always test Video cards with processors that cost half the GPUs.

 

[frozentundra123456]

It's mostly better threading. This is why Mantle pulls way ahead with Intel dual cores. No matter how well the renderer distributes workload there just isn't enough limited CPU resources to go around.



People like you are hilarious. It's either dismissal based on gut feeling or linking irrelevant GPU-limited benchmarks.

GPU limited benchmarks are hardly "irrelevant" since that is the situation for the vast majority of PC games. You do need to look at both cpu and gpu limited cases, but gpu limited cases are very relevant, because they show one of the limitations of mantle, that it shows minimal benefit except in cpu limited scenarios.

 

[96Firebird]

Its simple to test Mantle effectiveness: always test Video cards with processors that cost half the GPUs.

How about we stop testing for "Mantle effectiveness" and keep testing as normal?

 

[krumme]

GPU limited benchmarks are hardly "irrelevant" since that is the situation for the vast majority of PC games. You do need to look at both cpu and gpu limited cases, but gpu limited cases are very relevant, because they show one of the limitations of mantle, that it shows minimal benefit except in cpu limited scenarios.

Low fps sucks if its caused by cpu or gpu. The purely cpu and gpu limited situation is imho of minor importance at least for gameplay. What matters is actual gameplay with a typical consumer setup.

That certainly exclude eg weak amd apu with 290x at 720p and 1000usd heavily oc cpu. And ofcource sp gaming.

Its expensive to do this bm properly on a wide variety of setups. I think many will therefore have to rely on other peoples experience.

I think the problem we gave for that is the beta nature of the drivers. We have eg an 7850 1gb ram in house on a hw 4c machine and its a stuttering mess on mantle and needs to be fixed before its working. I think we will have to wait a driver change or two before we can start to do some proper evaluation. Right now gcn 1.0 is mostly not ready looking at the comments and 1.1 is still needed a overhaul.

It would be nice to know when they expect the next beta driver to arive?

 

[ASM-coder]

Again when Mantle finally comes out for BF4 and the posted results are minimal/marginal where real world in game improvements are in reality non existant, make sure you post back to this thread.

Obviously mileage varies, more benchmarking is needed, and more work needs to be done, but I have seen/heard some impressive results.

 

[ASM-coder]

GPU limited benchmarks are hardly "irrelevant" since that is the situation for the vast majority of PC games.

I would say...potentially, the vast majority of PC games can become GPU limited. I think Mantle is proving that many CPUs can't get the work to the GPU fast enough.

And I don't know about "vast majority." That's pretty nebulous, since it includes a lot of games, but it also includes a lot of crappy GPU's.

 

[Shivansps]

Putting out a review with most of those cards would put a quick damper on the whole Mantle reviews.

Unless you run those with a $50 CPU/APU.

AMD cant expect any devs to use unless it start giving out results, CF and 290x performance is good, but not important for the whole gaming industry, Mantle needs to start giving out results with every card and every cpu, and a results that worth wasting time working on another API, say more than 50%?....

If not, well, its gona have the same fate as Physx and remember that "EA givememoney" also adopted it...

So yeah, reviews with every card IS important, they need to be done.

 

[Arkadrel]

How about we stop testing for "Mantle effectiveness" and keep testing as normal?

The intire reason people read reviews is to see what kinda performance they could expect if they had X card.

The problem is most people dont have a 1000$ cpu, or a cpu thats overclocked to 5ghz, like most reviewers do.

And like it or not, the cpu effects the results alot.

Which means "most people" wont get the results the reviewers show when they buy X card.

For a avg gamer with a stock cpu, thats only 300$ mantle is great.
For a rich person with a 1000$ cpu overclocked to 5ghz not as much.

 

[frozentundra123456]

I would say...potentially, the vast majority of PC games can become GPU limited. I think Mantle is proving that many CPUs can't get the work to the GPU fast enough.

And I don't know about "vast majority." That's pretty nebulous, since it includes a lot of games, but it also includes a lot of crappy GPU's.

I would say that "potentially" what, optimistically 15 or 20 games within the next 2 years will use mantle? That is a drop in the bucket of the thousands of PC games on the market now and those coming. Mantle seems to show some nice improvements in cpu bound scenarios that support it, but because it is not a universal API, it is very limited in application.

 

[PPB]

Putting out a review with most of those cards would put a quick damper on the whole Mantle reviews.

Unless you run those with a $50 CPU/APU.

On the contraire, myself on a 7950 am seeing higher gains than the DICE 2# test. This game becomes horribly CPU limited when you enable mesh quality ultra and leave the eye candy off.

And considering a lot of people actually play the game that way (mesh quality ultra for higher view distance and being to spot enemies further, eye candy on low to see stuff better), those people would see similar gains, even on a not-so-optimized cards (for now).

 

[Stuka87]

I would say that "potentially" what, optimistically 15 or 20 games within the next 2 years will use mantle? That is a drop in the bucket of the thousands of PC games on the market now and those coming. Mantle seems to show some nice improvements in cpu bound scenarios that support it, but because it is not a universal API, it is very limited in application.

Here is the real question though. How many of those games that do not have Mantle actually need it? The vast majority of the games out there do not require high end hardware to run. You can play them are pretty low end hardware.

If those 15-20 games make up a big chunk of the AAA games that typically require more hardware to run, than that can easily make it worthwhile. Its not like you will be unable to play the lower end games because they do not have mantle.

 

[Cloudfire777]

Bit-tech found very small gains with Mantle over DX11.
Even when underclocking the CPU down to 2.8GHz

And with R9 280x and down, they sometime got negative results when using Mantle over DX11.

http://www.bit-tech.net/hardware/graphics/2014/02/05/amd-mantle-bf4-performance/

 

[Gloomy]

On the contraire, myself on a 7950 am seeing higher gains than the DICE 2# test. This game becomes horribly CPU limited when you enable mesh quality ultra and leave the eye candy off.

And considering a lot of people actually play the game that way (mesh quality ultra for higher view distance and being to spot enemies further, eye candy on low to see stuff better), those people would see similar gains, even on a not-so-optimized cards (for now).

I guess if you need to hit 144FPS for your monitor then Mantle is significant. Otherwise, I'm just not seeing it in the high end.

You're pretty much fine if you have an 3xxxK or 4xxxK past 4GHz in this game, even without Mantle.

Well, there's the Multi-GPU factor, but that's not really working properly right now so I'll refrain from commenting.

 

[Nintendesert]

Here is the real question though. How many of those games that do not have Mantle actually need it? The vast majority of the games out there do not require high end hardware to run. You can play them are pretty low end hardware.

If those 15-20 games make up a big chunk of the AAA games that typically require more hardware to run, than that can easily make it worthwhile. Its not like you will be unable to play the lower end games because they do not have mantle.

That's the real reason EA is wanting to do Mantle, if Mantle primarily improves performance on the low end with APUs and slower processors and graphics cards (unfortunate it only works with GCN) that greatly increases the potential customer base for their casual games.

We all may have top of the line or near top of the line hardware on this forum, but Steam surveys and in general people don't have that great of hardware in their laptops or computers. Improving performance on the low end has much more financial gain for EA than catering to us.

That's why I think EA got on board with this, Intel was supposed to be the deliverer of a greater installed PC master race but has changed it's focus to the mobile world and the gains from their integrated graphics aren't materializing. If they do, EA really isn't out too much so it's worth getting this working with Dice for the moment.

 

[krumme]

Bit-tech found very small gains with Mantle over DX11.
Even when underclocking the CPU down to 2.8GHz

And with R9 280x and down, they sometime got negative results when using Mantle over DX11.

http://www.bit-tech.net/hardware/graphics/2014/02/05/amd-mantle-bf4-performance/

On one hand they say:

"Instead of testing every CPU and GPU combination under the sun, we've focussed our initial Mantle testing on hardware combinations that we think will be most relevant and interesting to our readers"

And then (quelle surprice)

"We've stuck to our standard single player benchmark for testing. Multiplayer is too varied and non-repeatable to produce consistent and reliable results, which are important when trying to tease out performance differences that are potentially only marginal. This is unfortunate as it's in multiplayer scenarios that the game is likely to be more CPU-limited and more relevant to the majority of users, but having repeatable and accurate results takes precedence. "

Lol. Pathetic.
"This is unfortunate"

We are so sorry. Lol.

Well. Playing in BF is a multiplayer sport. And no, its not difficult to tell the difference.

All this trolling in this thread is sad.

What about next time to say its a SP test?

You forgot that?

Warning issued for inflammatory language.
-- stahlhart

 

[PPB]

I guess if you need to hit 144FPS for your monitor then Mantle is significant. Otherwise, I'm just not seeing it in the high end.

You're pretty much fine if you have an 3xxxK or 4xxxK past 4GHz in this game, even without Mantle.

Well, there's the Multi-GPU factor, but that's not really working properly right now so I'll refrain from commenting.

Your logic has a problem, why would you need to OC when mantle already can give you better performance?

Walking the patch of inefficiency was a last resort to get your desired performance level in a game. With mantle, you can even UV/UC and still get there.

I know that with so many goal post moved by the same people over and over you can get lost, but I can recall quite a few people that arguments against mantle are the same argumenting about lousy FX efficiency compared to the Core line from Intel, now crapping over efficiency to reach 5ghz is all fine as long as is used to dismiss mantle gains? :awe:

This only shows DX just cant cut it sometimes in modern games. LoL and such non-demanding games are just fine on DX. Well, I could think Howling Abyss could get mantle treatment, or at least DX11 treatment for god's sake.