• Hey Guest. Check out your NeoGAF Wrapped 2025 results here!

AMD Launches Mantle in Beta – Apparently its for Low End CPUs not GPUs!

Oemenia said:
See, this is proof that console optimisation is a myth!
Click to expand...

Actually, it is just that. With the data we have so far it's clear that Mantle mainly benefits systems which are completely unbalanced in terms of CPU and GPU power, either at the low end (very weak CPU) or at the extreme high-end (multiple monster graphics cards).

So, for the use cases that are relevant to the vast majority of gamers, either on console or PC, Mantle-like optimizations provide performance increases in the single digits. That is telling.
 
alexandros said:
Actually, it is just that. With the data we have so far it's clear that Mantle mainly benefits systems which are completely unbalanced in terms of CPU and GPU power, either at the low end (very weak CPU) or at the extreme high-end (multiple monster graphics cards).

So, for the use cases that are relevant to the vast majority of gamers, either on console or PC, Mantle-like optimizations provide performance increases in the single digits. That is telling.
Click to expand...

Eh what?

The beta drivers haven't even released, there's only one supporting game released and they patched support in, how is anything clear?

What graphics API you know that is feature and performance peaked at beta/first release?

How is your post anything but a fallacy?
 
iamshadowlark said:
Eh what?

The beta drivers haven't even released, there's only one supporting game released and they patched support in, how is anything clear?

What graphics API you know that is feature and performance peaked at beta/first release?

How is your post anything but a fallacy?
Click to expand...

I was responding to a post that used what little data we have so far to declare (through sarcasm) that console optimizations have a big effect. I explained that, based on the same data that the post was based on, the poster's conclusion is wrong. Nothing more than that.
 
Even if this isn't as good as some people were hoping it would be, it's still a good thing.

Worst case scenario it does nothing for most people and never gains traction, best case scenario it reduces CPU load and lets CPU-limited games run better, and maybe even becomes a new standard.

Any development into graphics APIs seems like a good thing to me, even if perhaps their effort could be better spent in a different way or a different API.
 
spwolf said:
so is this turning into amd vs intel/nvidia?

since i cant see why else would people downplay the numbers.
Click to expand...

Why would anyone want to downplay it? It is more performance for free, so it's great news for everyone. Mantle should also put some pressure on Intel and Nvidia to drop their prices if it gains traction.
 
Reading the mantle discussions always crack me up.

People will mutilate their systems, spend hours and hours overclocking and messing with configuration files for a small 5-10% improvement.

This is offering at minimum at 5% with some gains up to 50% for free (to the consumer) and you still have people bitching.

Internet.
 
This should help people who are building new gaming rigs. They can move some of the money they would have spent on the CPU over to the GPU. That should allow for a bigger bang for the buck. This should also help people who plan to upgrade since dropping in a better GPU while keeping the CPU the same will give better results.
 
Foaloal said:
Even if this isn't as good as some people were hoping it would be, it's still a good thing.

Worst case scenario it does nothing for most people and never gains traction, best case scenario it reduces CPU load and lets CPU-limited games run better, and maybe even becomes a new standard.

Any development into graphics APIs seems like a good thing to me, even if perhaps their effort could be better spent in a different way or a different API.
Click to expand...
Your best and worst case scenarios are swapped.
Best case - it won't get any traction but will push both OpenGL and DirectX into evolving a new thin API layer faster.
Worst case - it will get a lot of traction further fragmenting PC software infrastructure and making game engine development 33% harder (meaning more bugs and less stability of PC games overall, for all APIs).
 
dr_rus said:
Your best and worst case scenarios are swapped.
Best case - it won't get any traction but will push both OpenGL and DirectX into evolving a new thin API layer faster.
Worst case - it will get a lot of traction further fragmenting PC software infrastructure and making game engine development 33% harder (meaning more bugs and less stability of PC games overall, for all APIs).
Click to expand...

I see your point, but I think if it gets little/no traction and doesn't provide many benefits for most users it won't be able to put much pressure on the developers of DirectX or OpenGL.

If it does get traction then yeah, it will fragment things, but it will either overtake directx/opengl (unlikely) or have enough traction that it will be a serious competitor and force the Dx/OpenGL devs to improve their APIs to stop the fragmentation.
 
dr_rus said:
Your best and worst case scenarios are swapped.
Best case - it won't get any traction but will push both OpenGL and DirectX into evolving a new thin API layer faster.
Worst case - it will get a lot of traction further fragmenting PC software infrastructure and making game engine development 33% harder (meaning more bugs and less stability of PC games overall, for all APIs).
Click to expand...
This is some first class concern trolling and FUD. Bravo.
 
Multiplayer with 64 Players and 5 runs for average numbers for each result:
Test-Setting:
Ultra Details (Core i7-3770K, 16 GB DDR3-1333, Radeon R9 290X @ 1.000/2.500 MHz, Windows 8.1 x64, Catalyst 14.1 Beta)

01-battlefield-4-siege-of-shanghai-chart.png


http://www.golem.de/news/amds-mantle-api-im-test-der-prozessor-katalysator-1402-104261-3.html
 
That graph simultaneously also proves how shit the MSAA coverage is in BF4 for its performance.

I would probably inject SMAA and use their resolution scaling option. (SMAA injection impossible under mantle though)
 
Good results so far. One thing I'm unclear on is if you get a bigger boost using an AMD CPU + GPU combo or is an Intel CPU + AMD GPU going to give you just as big a boost?
 
Dictator93 said:
That graph simultaneously also proves how shit the MSAA coverage is in BF4 for its performance.

I would probably inject SMAA and use their resolution scaling option. (SMAA injection impossible under mantle though)
Click to expand...

Ive dropped using AA altogether, up the res and it looks quite clean.
 
GHG said:
Good results so far. One thing I'm unclear on is if you get a bigger boost using an AMD CPU + GPU combo or is an Intel CPU + AMD GPU going to give you just as big a boost?
Click to expand...

More cores=more performance apparently. I want to know how it does on midrange specs, like i5s and r270s
 
GHG said:
Good results so far. One thing I'm unclear on is if you get a bigger boost using an AMD CPU + GPU combo or is an Intel CPU + AMD GPU going to give you just as big a boost?
Click to expand...
You will get a bigger performance improvement with a slower CPU. As AMD GPUs are generally slower than Intel CPUs you'll generally get a smaller improvement with an Intel CPU. However, I imagine you could get a larger boost with an old Intel Q6600 or something like that than with the latest 5 Ghz 8 core FX processor.

symphonask said:
So, smaller gains at higher resolutions?
Click to expand...
Smaller gains in less CPU limited scenarios. So, smaller gains at higher resolution (or higher IQ settings) and smaller gains with faster CPUs.

Which, by the way, is exactly what I predicted shortly after the announcement.
 
symphonask said:
Wow, exact same specs as my PC aside from Win8.

So, smaller gains at higher resolutions? I mainly play at 1440p, but this is still really impressive.
Click to expand...

You also can have high gains in theory at high resolutions... problem is, normally with high resolutions you are gpu-bounded, not cpu-bounded. But if you get to have spare gpu power somehow, you also would have good gains.
 
GHG said:
Good results so far. One thing I'm unclear on is if you get a bigger boost using an AMD CPU + GPU combo or is an Intel CPU + AMD GPU going to give you just as big a boost?
Click to expand...

In a way, the worse the cpu (be Amd or Intel), the better the gains, as the cpu workload is the highest difference.

That's overall. Additionally, Mantle seems to scale better in all the cpu cores better than D3D/OGL, so if you have something like a 8-core cpu, in theory there should be an extra % to obtain in comparison to another cpu.

And last, Mantle is suppose to use new Crossfire hybrid configurations, like amd APU + amd GPU, though I'm not sure if that's is already working on this first version of Mantle or it will be something for the future. It was in a slide.
 
Locuza said:
Multiplayer with 64 Players and 5 runs for average numbers for each result:
Test-Setting:
Ultra Details (Core i7-3770K, 16 GB DDR3-1333, Radeon R9 290X @ 1.000/2.500 MHz, Windows 8.1 x64, Catalyst 14.1 Beta)

01-battlefield-4-siege-of-shanghai-chart.png


http://www.golem.de/news/amds-mantle-api-im-test-der-prozessor-katalysator-1402-104261-3.html
Click to expand...

Dictator93 said:
http://www.computerbase.de/news/2014-02/battlefield-4-mit-mantle-erster-eigener-benchmark/

Single player gains in a secenario with an overclocked i7 and a 290X.

So... definitely has some stretching room.
Click to expand...
Soooo underwhelming. DICE shouldn't even have bothered
 
As time goes on I see this as becoming a pretty big deal for those who can't afford upgrading often. i.e. not having to upgrade CPU as often, only GPU.
 
iamshadowlark said:
Soooo underwhelming. DICE shouldn't even have bothered
Click to expand...

I think the underwhelming part is more of a cognitive thing for me. I really enjoy the fact that CPU bottlenecks are trying to be dealth with on a level... but I dislike how it is limited to a certain group of hardware, using a new API, with an application to only "some games."

I wish AMD tried to do something similar with a reformation of Open GL or something. It would seem more practical for the rest of the gaming population.

The-Bean said:
As time goes on I see this as becoming a pretty big deal for those who can't afford upgrading often. i.e. not having to upgrade CPU as often, only GPU.
Click to expand...

Depends on the games you play unfortunately.
 
Star Swarm is a very dynamic benchmarc and has very strange results and CPU usage.
I don't care much about these results, but BF4 is a real Game.
 
Locuza said:
Star Swarm is a very dynamic benchmarc and has very strange results and CPU usage.
I don't care much about these results, but BF4 is a real Game.
Click to expand...

The creators wrote a FAQ, it seems Follow is supposed to be a bit more stable and less wildly variable than Attract.
 
Have they said anything about a public API release yet? I don't doubt AMD did some work here, but I'm always skeptical of benchmarks like this on closed source software that we have no idea what's going on behind the scenes. Having a real test case like Battlefield 4 is a good indicator tho since their D3D renderer is probably very efficient, but I would be way more doubtful of that Star Swarm thing.

No better advertisement than every developer getting access to it IMO. :P
 
Looking at the prelim reviews, mantle seems to be really good.

The question now becomes how easy is mantle for devs to incorporate. Will be very interesting to see what non-partners have to say about that in the coming months.
 
I just realized, in the specific case of BF4, Windows 8.1 with DX11.1 performs faster than Windows 7 with DX11.0 right? If so, doesn't Mantle theoretically eradicate the performance difference between Win7 and Win8 by a good amount?

That would be good news to me as I don't really want to upgrade to Win8 just for a bit more FPS :P
 
Big drop at 4K, not surprising though since pretty much nothing on the market runs 4K impressively at the moment.

Seems like low end hardware receives the biggest boost, but high end hardware has its ceiling raised pretty significantly as well. Cool stuff.
 
I'd love to see some 1440p numbers. Even a 10-15fps gain in BF4 is really like upgrading the class of gpu for free.
 
Probably a very silly question, but could mantle be useful in emulation contexts? Right now you need really powerful CPUs to emulate properly a wii, glitches aside.
 
Locuza said:
Multiplayer with 64 Players and 5 runs for average numbers for each result:
Test-Setting:
Ultra Details (Core i7-3770K, 16 GB DDR3-1333, Radeon R9 290X @ 1.000/2.500 MHz, Windows 8.1 x64, Catalyst 14.1 Beta)

01-battlefield-4-siege-of-shanghai-chart.png


http://www.golem.de/news/amds-mantle-api-im-test-der-prozessor-katalysator-1402-104261-3.html
Click to expand...

These numbers are a bit weird, iirc the performance drop you get from adding MSAA on BF 4 is far bigger than 6 fps.

Edit: I mean the DirectX numbers, you can see that the Mantle numbers go down significantly once they've added MSAA.
 
OminoMichelin said:
Probably a very silly question, but could mantle be useful in emulation contexts? Right now you need really powerful CPUs to emulate properly a wii, glitches aside.
Click to expand...

There's been interest in the tech from numerous emulator developers recently, like this. There was an interesting article that went into depth about how it could be extremely useful for Gen 7 emulation (360/PS3), but that of course hinges on other aspects as well (documentation/homebrew etc).
 
Dictator93 said:
I think the underwhelming part is more of a cognitive thing for me. I really enjoy the fact that CPU bottlenecks are trying to be dealth with on a level... but I dislike how it is limited to a certain group of hardware, using a new API, with an application to only "some games."

I wish AMD tried to do something similar with a reformation of Open GL or something. It would seem more practical for the rest of the gaming population.



Depends on the games you play unfortunately.
Click to expand...

alexandros said:
It's probably sarcasm, but in any case these results are great for AMD. If they hold up in scrutiny, it means that for many popular games AMD will have a significant advantage. Well played Team Red!
Click to expand...

It's definitely sarcasm. Efficiency is a beautiful, beautiful thing.
 
You must log in or register to reply here.

Similar threads

PS3 emulator RPCS3 makes Cell CPU 'breakthrough' that improves performance in all games - 'All CPUs can benefit from this, from low-end to high-end'
News Game Dev  2
66
813
Radical_3d replied
NVIDIA announces supply difficulties for its GPUs
Opinion Business Hardware  2
51
1K
Dorfdad replied
Optiscaler updated to support FSR 4 on older AMD RX 6000 GPUs without the need for driver mods
Hardware Indie 
0
307
IbizaPocholo replied
[PC Gamer] Former head of FSR development at AMD posts 'big trouble' meme when asked about INT8 version of FSR 4 for older GPUs
31
249
spons replied
Intel is offering a bunch of games and software with their CPUs and GPUs
News 
7
643
xrnzaaas replied
Top Bottom