AMD's Mantle is interesting, but it's no game changer – yet

AMD's Mantle is interesting, but it's no game changer – yet

Summary: AMD's new Mantle API is an interesting science experiment, but it's not yet ready to go up against Microsoft's DirectX API.

TOPICS: Hardware

Back in February of 2013, AMD announced that it was working on a new graphics API called Mantle, and that this was going to be an alternative to Microsoft's dominant DirectX API, and also the OpenGL API. On paper it seems good, but in reality things don't seem to be as impressive.

Benchmarks tests carried out by AnandTech, which are based on beta code as opposed to the final release final, show that Mantle provides big performance gains in certain situations, with small performance gains otherwise.

The benchmark results show that the best performance gains from Mantle are to be had when a high-end GPU is combined, which can result in as much as a 30 percent performance increase over DirectX. However, when the same high-end GPU is combined with a high-end processor, those gain fall to under 10 percent.

So, Mantle seems to be able to offer a good workaround for a slow CPU, but when both the CPU and GPU are high-end, it's not much better than the DirectX.

And since not many people are going to put a $500 GPU in with a cheap CPU, the appeal of mantle is going to be limited.

What we don't have is data on how well Mantle performs when a mid-range or budget GPU is combined with a budget GPU. If gains here are good then there might be a case for Mantle – but success hinges on a lot of factors.

These include:

  • Performance of the final code – This is beta code that's been tested here, so things might improve
  • Developer support – Crucial to any new API is going to take off, and while there are some 20 games in development using Mantle, it's too early to tell how well it will be adopted
  • Streamlined development – Anything that helps make a developer's life easier is going to attract attention
  • Cross-platform support – AMD has talked a lot about Mantle being "cross-platform" but also added that "the initial iteration of Mantle is intended specifically for PCs," which leaves it open for future versions to have a broader audience, which would be good
  • Console support – Word is that Mantle support won't be added to the Xbox One or PlayStation 4, but that still leaves an opening for Steam OS and Steam Box devices

Right now, I think that Mantle is a nice science project for AMD, and good things could come of it, but it's not a game changer and certainly not in a position to start taking on the likes of DirectX. 

Topic: Hardware

Kick off your day with ZDNet's daily email newsletter. It's the freshest tech news and opinion, served hot. Get it.


Log in or register to join the discussion
  • CPUs age

    You are wrong Mr Hughes and here is why....CPUs age. Today's high end becomes tomorrow's low end which means you'l be able to keep that CPU longer by using Mantle over DirectX.

    I'll use myself as an example, the Phenom II X6 was the high end of the AMD line when I bought it in 2009 and now that is a generation behind. I have a friend that is still happy with his Yorktown C2Q for gaming, how long ago did those come out?

    So while I agree about adoption by the game industry being key I can't agree that its only for low end CPUs because what is hot today will be long in the tooth tomorrow.
    PC builder
    • yep

      very close to what I wrote... I had huge gains on my Phenom II x6, which is actually better now as more games are using more cores the AMD chips are getting better with age instead of worse!!
      • Flawed benchmark test

        i looked at the benchmark they used to come to this conclusion. They use an Intel processor (not mantle compatible) with an AMD GPU so of course the test is only going to show you speed increases on only the GPU's.

        Maybe before proclaiming something as not a game changer they should benchmark it with all the hardware its supposed to be be a game changer for....
  • DirectX

    Is under a severe threat from the Khronos Group, it's only a matter of time before it goes the way of WMV, WMA, Silverlight etc.
    Alan Smithie
  • You are so wrong...

    Because most gamers are not running i7's and the performance of current games are all held back by Direct X, there are so many limitations and this is why people need such strong CPUs for gaming. When in reality CPU's shouldnt be a bottleneck for these new GPU's. Also there are alot more gains to be had. The tests done were very small and not varied, it seems that the world of review sites has gone more anti AMD in the recent months as they have announced more game changer stuff.
    Shows whos pockets the review sites are in. On my personal machines I had gains of 500% in certain scenarios and I have a 7870Ghz and a 1045T OC to 3.3Ghz... That alone is enough to show that it is a game changer as my hardware isn't at all expensive but Direct X is over taxing for no reason.
  • Not A Chance...


    Until Mantle is endorsed by a MAJOR party, chances that it can dislodge DirectX or OpenGL are almost none. Inertia and investments made in these techs are key factors here, whatever the performance gains.
  • better

    the one word I see throughout this article is 'better';
    "not much better'', ''a little better'', ''10% better'', ''30% better"...
    it seems the whole concept here is better, even if only to a degree.
    and while directx will surely rise to meet the challenge of this competition, this is only beta, right?
    I have no interest high end gaming or map rendering, plenty of people will do anything for even the 10% you mention...
  • Wow you don't get it - "not many people...$500 GPU with cheap CPU"

    "And since not many people are going to put a $500 GPU in with a cheap CPU..."

    Well why aren't they??? Maybe it's because currently they NEED the high end CPU to be able to see any benefit from their very high end GPU in games. I'm surprised you don't see how naive a comment you made there. If Mantle can remove the bottlenecking, and people can get the fantastic performance out of a higher end GPU they should be getting, whether or not they got the best CPU to go with, then why should they spend the extra money?

    Here I'll put it yet another way. Say someone is building a new computer and they have a budget of $x. They want to play games and want a high end GPU. They check out their possible specs and see that they will get the same game performance with a $180 decent CPU as they would with a $300 high end CPU, then guess what, they can choose to go with the decent CPU have an extra $120 for a bigger SSD!

    Yet another scenario, this one happened to me - a gamer has an aging computer, and wants to upgrade. They upgrade the GPU easily (this does not require much work, pop it in and install drivers). However, they might see now their old CPU is really holding them back. Say they have a Core 2 Duo or something. They buy a faster model Core 2 Duo on ebay to speed it up a little, they overclock, still isn't not quite as fast as they want. In the end they need to bite the bullet and do the big upgrade - Motherboard, CPU, memory - including all the hassle of reinstalling the OS and programs, because they were stuck with a dead end 775 socket, and DDR2 memory. Now if Mantle had been in the equation, the GPU upgrade could have been all they needed for maybe another year or two! This is real dollars saved.

    I suggest you go back and read the Anandtech article and other info about Mantle.
  • The author is missing a demographic point.

    Mantle was explicitly designed to eliminate CPU overhead on draw calls. Many published benchmarks on Mantle seem intent on avoiding testing this aspect of the API as much as possible.

    "I know a great way to test something that reduces CPU overhead! Let's run BF4 on ultra at 1440p with 4x MSAA and 150% resolution scaling. That's definitely CPU limited!"

    Yeah, that's the ticket.

    My R9-290x and Phenom II x6 @ 3.7GHz went from 40fps minimum on all low due to being CPU limited on DX11 to 70fps minimum on an ultra/high mixture just from using Mantle.

    The broader impact of Mantle is actually huge, since there are a massive number of active PC gamers who do not have top end CPU's that could easily be CPU bottlenecked with a graphics upgrade, which essentially includes anyone using any AMD CPU or APU, and also Nehalem/Core2 users who have no upgrade path without replacing their motherboard. Anyone without an X'Bridge or Haswell i5/i7 has something clear to gain here, and that is probably over a third of the desktop gaming market.

    These people might otherwise fall out of the market because of the high cost to upgrade an entire system. A $500 GPU is a much better deal than building a new $1000 machine.

    And let's not forget Devon's Paradox. With the increase in CPU efficiency that Mantle brings, developers will have a lot more CPU headroom to work with. And what will they do? They will fill it up with better AI, better physics, better interpolation in the netcode, etc.

    The consoles have the Mantle API in all but name already. The impetus for Mantle in the first place was the creation of such an API for the PS4 and XBone. Adapting the internal expertise of building an x86 GCN graphics API for two consoles to also be used on the broader set of x86 chips and GCN cards was clearly not an insurmountable task, since it is literally working right now.

    Mantle is an open ended API that is going to be around at least through the end of this console generation, so nVidia could certainly leverage something out of it too, though it won't be as strictly to-the-metal as it is with GCN. But that is something that they can certainly work-around on the driver side.

    Mantle is all good news, just not equally good news for everybody all right now.
  • AMD Mantle on Xbox Cloud Servers

    No one is talking about the connection this MIGHT have with AMD's Mantle. If AMD Mantle is ONLY for PC hardware and Microsoft has confirmed that the XBOX only uses the DirectX 11.2, then maybe Mantle is going to be running on the servers in the cloud. So imagine, Mantle on the cloud (server hardware) and Direct X on your actual Xbox.

    Someone wrote about it here: