The AMD Mantle Thread

[krumme]

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.

144fps? Where did that number come from?

Besides from the made up number, there is just the problem with this argument that those that actually game dont agree. H for instance.

 

[ASM-coder]

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

An interesting thing that they did...
"One final word on how we recorded the data: for the standard DirectX results we've used a 60-second FRAPS recording as usual, and repeated the benchmark at least three times in each case to ensure accuracy. However, FRAPS doesn't recognise the Mantle API, so for those results we've had to record the data using a new in-game console command."
Why not use the in-game console command for both?

 

[krumme]

Goal post - add to the list:

Mantle is just hype - where is it?? ("2 month and still not here")
Windows 8 makes just a huge gains
The gains is not significant
Mantle is only comming for at most 20 games the next 2 years.
Its fragmenting

 

[el etro]

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

Why normal is to pair CPUs and GPUs that cost the same?

My i5 3570($200) plus GTX 670($350) 99% of times gave me the same framerates of a 3970x with GTX670 system. Then why would i have to pair a 3770k + HD 7870 system otherwise?

 

[Despoiler]

An interesting thing that they did...
"One final word on how we recorded the data: for the standard DirectX results we've used a 60-second FRAPS recording as usual, and repeated the benchmark at least three times in each case to ensure accuracy. However, FRAPS doesn't recognise the Mantle API, so for those results we've had to record the data using a new in-game console command."
Why not use the in-game console command for both?

Also, see known issue from the 14.1 drivers.

Mantle performance for the AMD Radeon™ HD 7000/HD 8000 Series GPUs and AMD Radeon™ R9 280X and R9 270X GPUs will be optimized for BattleField 4™ in future AMD Catalyst™ releases. These products will see limited gains in BattleField 4™ and AMD is currently investigating optimizations for them.

 

[krumme]

An interesting thing that they did...
"One final word on how we recorded the data: for the standard DirectX results we've used a 60-second FRAPS recording as usual, and repeated the benchmark at least three times in each case to ensure accuracy. However, FRAPS doesn't recognise the Mantle API, so for those results we've had to record the data using a new in-game console command."
Why not use the in-game console command for both?

It must be the exclution from "repeatable and accurate results takes precedence".

One can say for AMD its good they have EA moving mantle and is not reliant on a market used to please a gfx vendor.

 

[stahlhart]

All of you need to back off the confrontational tone of your posting and stop trying to provoke fights, or this thread is going to get locked.
-- stahlhart

 

[krumme]

Even when underclocking the CPU down to 2.8GHz

That is. Even when they underclocked their overclocked cpu.

 

[Gloomy]

144fps? Where did that number come from?

Besides from the made up number, there is just the problem with this argument that those that actually game dont agree. H for instance.

144FPS is what you need to "max" a 144Hz monitor. That's where that number came from. :|

 

[krumme]

144FPS is what you need to "max" a 144Hz monitor. That's where that number came from. :|

Yes. But you dont get there in the cpu heavy situations, even with mantle. There is plenty of situations and places at eg. siege of shanghai or golmud, in 64 player battles, where you get sub 80 frames with mantle. Its seldom but its there.

If you are having a 144Hz monitor - what are you gunning for on average and min?

I would say you are going for at - least -160 on average. Is that even doable on the big maps with mantle? - even with a heavily oc 5ghz 1000usd cpu.?

We can always use more cpu power. We all know how it sucks to be cpu limited. Mantle didnt change that. Look at the cpu advice people get in this forum. Its damn powerfull stuff, and thats for a reason.

 

[GaiaHunter]

One thing to remember is that GPU power is still growing at a faster rate than CPU power.

 

[3DVagabond]

One thing to remember is that GPU power is still growing at a faster rate than CPU power.

^This.

Whether people want to admit it or not, we all know that the top cards are CPU limited a lot of times. As time goes on and we continue to see GPU performance gains outstrip CPU performance gains, and by substantial amounts, it's going to get worse. From reading what the devs are saying they are finding it difficult to optimize CPU performance any further without hindering the artistic content.

I honestly believe once Mantle is refined enough AMD will make it open and anyone who wants to use it will be able to.

 

[turn_pike]

Thanks for the pcgameshardware.de link
It is not often that a tech site actually test high end GPUs with prev gen CPU like Phenom II. Far too often tech site and forum live in a bubble where everyone upgrade their rig every two years.

Thanks to Mantle I can actually play BF4 multiplayer with my good 'ol Phenom II 955.

 

[realibrad]

^This.

Whether people want to admit it or not, we all know that the top cards are CPU limited a lot of times. As time goes on and we continue to see GPU performance gains outstrip CPU performance gains, and by substantial amounts, it's going to get worse. From reading what the devs are saying they are finding it difficult to optimize CPU performance any further without hindering the artistic content.

I honestly believe once Mantle is refined enough AMD will make it open and anyone who wants to use it will be able to.

I would say its less about CPU power, and more about how the cpu is being used. The reason there are any gains with Mantle at all, has to do with how the cpu was being fed data. "Most" of the time top end cards are used with mid to top end cpus, and a mid-range cpu will not bottleneck a single top end card. An i5 2500k with an OC to 4Ghz is more than enough not to bottleneck a single GPU.

 

[sm625]

I think this chart makes clear what AMD was trying to do with mantle: bail out the poor performance of their own cpus. For intel cpus, there really isnt much gain to be had. It is still an epic fail because a 290x still cannot beat a 780.


Posting in this manner will just derail the discussion.

-Rvenger

 

[Despoiler]

Interesting interview just posted on PCper. It covers some of the "how hard will it be to port DX to Mantle and vice versa" type questions.

http://www.pcper.com/reviews/General-Tech/Core-Mantle-Questions-Interview-AMDs-Guennadi-Riguer

[Guennadi] With the current shader compilation strategy in Mantle, the developers can directly leverage DirectX shader optimization efforts in Mantle. They would use the same front-end HLSL compiler for DX and Mantle, and inside of the DX and Mantle drivers we share the shader compiler that generates the shader code our hardware understands.

 

[itsmydamnation]

I think this chart makes clear what AMD was trying to do with mantle: bail out the poor performance of their own cpus. For intel cpus, there really isnt much gain to be had. It is still an epic fail because a 290x still cannot beat a 780.

there is so much logic fail here i can only assume your a shill
first
you have a 36% improvement on a 4770k
you have a 50% improvement on a 3770k
you have a 51% improvement on a 8350
so thats only favoring weak amd cpus and is also not much gained??

second the agenda driven shills who completely ignored my last post http://forums.anandtech.com/showpost.php?p=36034633&postcount=6649 and have made no attempt to determine where the GPU's themselves actually sit compared to each other. just a bunch of hand waving that completely ignores what the damn game dev has said and also what the data says about the 780 performance scale.

third why is it only about beating a 780, guess what mantle and BF4 will still be around when the next set of GPU's hit.

there is either so much stupidity or dishonesty going on in this thread its nauseating......


Personal attacks and use of the word 'shill' will not be tolerated here. You will only derail the thread.

-Rvenger

 

[Skurge]

CPU limited, I don't see nVidia doing better than AMD here.

 

[96Firebird]

^Wish they would have tested on Win 8.1...

 

[jj109]

CPU limited, I don't see nVidia doing better than AMD here.

How is that in anyway CPU limited? The 7990 is running 30 fps faster than 290X or 780ti. It's obviously GPU-bound.

 

[ASM-coder]

How is that in anyway CPU limited? The 7990 is running 30 fps faster than 290X or 780ti. It's obviously GPU-bound.

The key is the 4xAA. That is the GPU hog.

 

[bystander36]

nm.

 

[itsmydamnation]

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.

but they can, this took me about 25 mins last night, two quick runs

Zavod 64 player
I7-3770@ 4.3ghz
16gb 1866 ramm
6970 ( 6950 bois flashed to 6970) @ 3840x1024 most settings @ medium no AA no AO, 75 FOV

getting good real world representative data out of BF4 is easy as, people should be trying to control the environment, just let the law of averages sort it out for you

 

[jj109]

The key is the 4xAA. That is the GPU hog.

So, GPU-bound like I said. It's hard to imagine how someone can look at the 7990 results and claim that the graph shows CPU bound behavior.

 

[ASM-coder]

So, GPU-bound like I said. It's hard to imagine how someone can look at the 7990 results and claim that the graph shows CPU bound behavior.

So what's the debate? It still doesn't show "nVidia doing better than AMD here."