Comments Locked

46 Comments

Back to Article

  • Owls - Tuesday, October 15, 2013 - link

    Reinterate? That's not a word.
  • Ryan Smith - Tuesday, October 15, 2013 - link

    D'oh. Indeed it's not.
  • nathanddrews - Tuesday, October 15, 2013 - link

    Of course it is! You know, like how you interate, but then do it again?
  • psuedonymous - Tuesday, October 15, 2013 - link

    It's a perfectly cromulent word!
  • Brutalizer - Tuesday, October 15, 2013 - link

    Yeah, I reinterate all day long! Almost every day. Who doesnt? In fact, sometimes I also reinterate in the night too, but dont tell anyone.
  • Flunk - Tuesday, October 15, 2013 - link

    Did you know that cromulent is now a real word? Check the dictionary, the Simpsons really added a word to the language.
  • makerofthegames - Tuesday, October 15, 2013 - link

    Interate (v) [jargon - computer science] - To cast a variable into an integer type. Eg.
    int foo;
    float bar = 0.0f;
    foo = (int) bar; // interate bar

    Thus:
    reinterate (v) [jargon - computer science] - to cast a variable into an integer type, when the variable is already of integer type. Eg.
    int foo;
    int bar;
    foo = (int) bar; // reinterate bar
  • inighthawki - Tuesday, October 15, 2013 - link

    I refuse to acknowledge this.
  • Ryan Smith - Tuesday, October 15, 2013 - link

    Haha. Okay, that's a good one. You sir win the cookie for the day.
  • tviceman - Tuesday, October 15, 2013 - link

    When AMD has a minority (by a significant margin) in both the dGPU and CPU market, their argument has very little weight.
  • 1Angelreloaded - Tuesday, October 15, 2013 - link

    I am a Nvidia fan, but not to acknowledge what AMD does for the price is just stupid and they are doing their best fighting a War on 2 fronts at 1 you have Intel and the other Nvidia and in case you haven't noticed they burn money like its going out of style. I give credit where is is due in the High end gaming market you are right, in mainstream it is mixed and in computational AMD goes the long mile. My choice of preference is based on the featuresets each GPU manufacturer puts out right now TressFX is in infancy and way 2 taxing on systems, but it has a place, as does Nvidias PhysX which is what made my choice for Nvidia, but that isn't saying AMD is dead in the water or isn't being watched, because I'm kinda routing for them to make a comeback so we can have a power GPU war again.
  • eanazag - Tuesday, October 15, 2013 - link

    When discussing the mobile market, then I would buy that statement without factual numbers. On desktop side in gaming machines, which is what would be important in this context, I am not so certain. I have two Nvidia cards in separate machines at home now for gaming. But am considering an AMD for one because I had an ATI 3870 that behaved better in Starcraft II, than my GTX 660 that I got over the summer. I am considering AMD because I had better stability in my typical setup (one CPU, one GPU, single 1080p monitor, and 5.1 speaker audio). I like what they are doing with the audio side of the equation and Mantle even if it doesn't pan out. The audio functionality seems like a more solid selling point.

    With developer buy-in Mantle could be a pretty compelling factor in purchase decisions. This could positively affect power usage in AMD dGPU environments from lower CPU utilization and more efficient utilization of graphics hardware along with better performance to boot. From a numbers standpoint - AMD CPUs with AMD GPUs could actually be competitive in gaming also (this is optimistic yet and I am thinking in single video card scenarios). I am weighing price, performance, and my desire to support a second vendor to Intel on CPU side when possible. No competition for Intel sucks for buyers. Intel server processor prices are ridiculous right now.
  • bill5 - Tuesday, October 15, 2013 - link

    Actually AMD has more gfx share than Nvidia and the gap is growing.

    I see you cleverly worded your post to exclude AMD APU's, though...

    But yeah, the consoles, also.
  • hfm - Tuesday, October 15, 2013 - link

    Source for this AMD market share lead data?
  • HisDivineOrder - Wednesday, October 16, 2013 - link

    Let's say you were hypothetically correct about AMD's having more GPU market share, which I haven't seen numbers on to be sure.

    We're not talking about AMD having greater market share. We're talking about AMD's GCN-based products having greater market share. Mantle doesn't benefit anything except GCN-GPU's, so that excludes...

    Every AMD discrete GPU pre-7xxx series and every APU, excluding Kabini, Temash, and the yet unreleased Kaveri.

    Mantle also will not support any nVidia or Intel GPU.

    The percentage of the market in the PC space for this PC-only API that Mantle will support is a tiny fraction of an already tiny fraction of overall gamers. Moreover, the advantage it brings is merely "better performance," which overall is not more important to developers than either "make porting cheaper" and "making porting simpler." Sure, Mantle by itself in a world without DX might have had a shot. But DX and OpenGL exist. If a publisher wants to make money selling PC games, they're going to support to DirectX/OpenGL.

    So that leaves adding on the extra cost to support Mantle as the only optional part. Mantle will always cost more to support because DirectX/OpenGL are a given, so Mantle's always an add-on. Extra cost. More performance for a tiny fraction of an already small group of users isn't going to be compelling to publishers too lazy to even add FOV sliders, mouse acceleration disable, or high resolution options to PC ports.

    Those publishers are going to stick with a single API that gives them as large an audience as possible with as little trouble as possible to patch, update, and QA. That's going to be either DirectX or maybe if we're very, very lucky in the future OpenGL.

    Short of being paid millions of dollars (to essentially buy the game plus pay for the Mantle support), a publisher just is not going to waste money and manhours supporting a SECOND API for no reason except just more performance for that small group of users.

    This is especially true now that the, "Code once, port to all platforms" nonsense has finally been put down. By Microsoft of all companies, too.
  • Despoiler - Wednesday, October 16, 2013 - link

    I'm sorry, but what part of powering 100% of next gen consoles has little weight? Console to PC porting is going to be that much faster due to being able to port over a larger chunk of code than normally would be possible. AMD is helping developers and also boosting performance for their PC customers. End of story.
  • Nenad - Thursday, October 17, 2013 - link

    I agree on this.
    What I fail to see is what is preventing Microsoft to add this "Low Level API" (that already exists on their DirectX for console) to general (PC) DirectX standard API ?

    Obviously, ATI would be able to support such API extension (already does in their XboxOne drivers). Question is if Microsoft would be able to make "low level" API in such way that Nvidia could support it with same level of efficiency. Another question is, if that is possible, why MS didn't already add it to DirectX APIs.

    But if its possible, that would prevent fragmentation on this low level stuff.
  • Gunbuster - Tuesday, October 15, 2013 - link

    Port your code from Xbox One or PS4 to Mantle for PC that only works on our brand new cards... Hmmm let me know how that works out.
  • MamiyaOtaru - Tuesday, October 15, 2013 - link

    will probably work relatively well for pushing new cards. Which is what they are interested in
  • inighthawki - Tuesday, October 15, 2013 - link

    I thought they mentioned it was for all GCN cards, which means it's compatible with a couple generations worth of cards...
  • chizow - Tuesday, October 15, 2013 - link

    One generation of cards (rebranded a few times), Tahiti and newer. Bonaire and Hawaii are updated versions of GCN, but still of the same generation.
  • Dentons - Tuesday, October 15, 2013 - link

    Effectively, both new consoles are just PC's, but they do have something few PC's have, a massive amount of memory available to the graphics subsystem. 8GB shared memory. Game developers are certainly going to make use of that RAM. Currently, PC users have few options to reproduce that capability.

    If video cards with 8 or more GB of RAM will be required to run cross-developed console titles at comparable quality to the consoles, AMD may be anticipating an unprecedented rush to upgrade PC video cards.

    What better way to sell consumers on AMD cards than by making their PC video card ecosystem as similar as possible to the new consoles for which most new game titles will be primarily designed?
  • bill5 - Tuesday, October 15, 2013 - link

    Nah, the actual game usable RAM on both PS4 and X1 is 5GB. They both reserve 3Gb for OS.

    Yeah it's a lot reserved but it's true.

    Then on top of that the consoles have to run the CPU out of the same unified 5GB RAM pool as the GPU. So how much VRAM is the GPU getting? I think we can assume often 3-4GB or less.

    So yeah, with higher end GPU's already often being 3, 4, and even 6GB, I think the VRAM gap to consoles is already slim to none on PC, and will be even less in a little time. Plus those PC GPU's are backed with up to 16GB of DDR in system memory, unlike the consoles.
  • hfm - Tuesday, October 15, 2013 - link

    Not to mention my gpu barely uses 1G most of the time at 1080p.
  • MADDER1 - Tuesday, October 15, 2013 - link

    It's the usage model in consoles (PS4/XboxOne) that will make a difference. If the developers can get the usage right, you would no longer pay the performance penalty to copy the textures from system memory to graphics memory (VRAM) in current PC systems.
  • testbug00 - Wednesday, October 16, 2013 - link

    actually, the PS4 currently only gaurentees 4.5GB of RAM (has 1GB that is "flexible" last I checked) and reserves only 2.5GB for the OS.

    Provided sony does what they did no the PS3, the PS4 will eventually have 1.5-2GB of memory reserved for Os (6-6.5GB for games).

    provided they don't, they will end up with 5.5GB for games.
  • andrewaggb - Tuesday, October 15, 2013 - link

    well... it's still interesting. As I see it what this likely means is that games will have directx support with some custom code for the then current generation of video cards. It's still probably a win.
  • yowanvista - Tuesday, October 15, 2013 - link

    This somewhat reminds of 3dfx Glide.
  • 1Angelreloaded - Tuesday, October 15, 2013 - link

    So basically, apples and oranges to an extent M$ is optimizing and streamlining for consoles while Mantle does the same for PC but because they are easier and normalized it will make it easier to port things over..................................so Whats the big deal? good for them maybe we will get better ports as M$ exclusive doesn't include the PC as a platform. Now Nvidia whats your deal on all this, and what will you brig to the table?
  • chizow - Tuesday, October 15, 2013 - link

    I just don't understand how AMD thinks Mantle will somehow streamline the PC development process. Source code is going to be written with one of the predominant HLSL (DX/OGL) regardless and then ported to whatever console API as needed. Mantle just adds one more layer of stratification to the mix for a PC platform that will already need to support one of those two APIs.

    It's obvious that while one Dev may have asked for this (EA/DICE), many more are not so keen on it. Ubi has called Mantle a "double-edged sword" specifically citing longer development times and more work, and Activision seemed less than thrilled by saying they may be forced to support it, jokingly asking if someone could just provide them with a lightweight OGL/DX Mantle wrapper.
  • klagermkii - Tuesday, October 15, 2013 - link

    From what we learnt at the Mantle reveal it's DX HLSL compatible, so at least they won't have to rewrite all the shader code.
  • andrewaggb - Tuesday, October 15, 2013 - link

    If the performance is actually better and mantle works in linux then it could be a success. But It's also my understanding that amd is exposing more stuff through opengl extensions as well, and as ms stated, 11.2 adds some of the mantle features as well, so it's kinda hard to say how this will all play out.

    At least I have a 7000 series gpu so I can try it first hand :-)
  • chizow - Thursday, October 17, 2013 - link

    They'll have to recompile the DX or OGL HLSL to Mantle, then do any of the low level optimizations they want to from there. It's more work for Devs, plain and simple. We will soon see if it is worth it, from both a Dev and end-user standpoint.
  • MADDER1 - Tuesday, October 15, 2013 - link

    Arrgh! I just want more performance! If developers don't want to maximize hardware, sounds like they are wasting electricity when they can do better. Oh wait, it all comes down to $$$ , not performance. So sad.
  • chizow - Thursday, October 17, 2013 - link

    Yep exactly, it is always going to come down to $$$ and not just performance, and adding another code path to support is going to further eat into that time/money budget that Devs always face.
  • twtech - Tuesday, October 15, 2013 - link

    I don't believe "similar" is good enough, unless it also fully supports nVidia GPUs. If Mantle were on the consoles, you could make a case for providing it as an option on PC for cross-platform games.

    But the PC part of the market just isn't big enough to justify supporting both cross-platform Direct3D or OpenGL as required, and then Mantle just for Radeons. Game developers will make the Direct3D version because they have to, and then that will be the end of it.
  • Gigaplex - Tuesday, October 15, 2013 - link

    "What Mantle creates for the PC is a development environment that's *similar* to the consoles, which already offer low-level APIs, close-to-metal programming, easier development and more (vs. the complicated PC environment)."

    Um... The PC environment isn't more complicated to write code for. The high level abstractions which rob performance are there specifically to make it EASIER to write code for.
  • ddriver - Wednesday, October 16, 2013 - link

    AMD is just digging itself deeper. They should focus on making good OpenGL drivers and improve performance there, AMD won't win the hearths and mind of developers by forcing yet another vendor limited API down their throats. There is no place for vendor-locked APIs in the future, not even in the multiplatform present IMO. MS has done enough damage with its DX...
  • ddriver - Wednesday, October 16, 2013 - link

    "Mantle: improves time to market; reduces development costs"

    Really? How does it improve any of those considering that if you want your game to run on non AMD hardware you have to do everything all over again?

    Surely, there is nothing more beneficial to development cost and time than doing the same thing twice... or why not even 3 times? Mantle for AMD, DX for MS and OGL for Android.
  • chizow - Thursday, October 17, 2013 - link

    Yep, totally agree, I've raised that question as well. It's really just marketing jargon by AMD to try and encourage uptake, same as their misleading comments about Mantle being Open source, Open standard, multi-platform. Clearly, it's none of this, it's AMD proprietary API for GCN-only, PC-only.

    If anything AMD said about improving time to market or streamlining the development process were true, we would at the very least see Mantle launch alongside the Windows DX version of BF4, but of course, it's coming a full month+ later in December (maybe).
  • Wolfpup - Wednesday, October 16, 2013 - link

    If Mantle isn't pretty much a straight shot from the consoles to PC, then it's pointless.

    "Mantle: improves time to market; reduces development costs>>>

    No, it INCREASES development costs, as now you've got to support OpenGL and/or Direct3D, AND Mantle. And you're doing it to support some current AMD hardware, and who knows for how long they'll even support it? Eh?

    <<<and allows for considerably more efficient rendering, improving performance for gamers.>>>

    If it actually does that, it means there's something wrong with OpenGL and/or Direct3D, and they need to help fix them.
  • erple2 - Thursday, October 17, 2013 - link

    Not having actually used Mantle, or developed a game in it, I don't think that you can necessarily make that claim. I also don't know how it makes things easier and our cheaper, but I've not used it. I can say that developing for directx is surprisingly complex. A lower level programming language could be easier and or cheaper. Even programming the hlsl is highly dependent on whether you want to optimize performance for nv or amd hardware. Perhaps with Mantle there is less tweaking you need to do for amd hardware?
  • Th-z - Thursday, October 17, 2013 - link

    AMD did say it's not for everyone, DirectX will continue to be supported by all cards. On consoles both high and low level API exist, people can choose which one to use depending on type of their game, cost, and time. Same thing here, for game developers and game engine makers who want to push the envelope, who don't settle for good enough ports, or to showcase what they can achieve, there is an option on PC side. When game engines and middle wares support variety of APIs including high performance API like Mantle, developers who use these tools can inherent these supports for their games.
  • Dman23 - Wednesday, October 16, 2013 - link

    I have to say that I'm looking forward to what Mantle has to offer. It looks like it's going to be very good
  • Shiggy Piggy - Monday, October 21, 2013 - link

    So, AMD has made an Xbox One emulator for windows?
  • polaco - Monday, October 21, 2013 - link

    maybe what AMD means is that if you use Mantle then you can compile take advantage in some kind of compiler at the time the code is compiled for certain platform. If Mantle API is similar to Xbox One DX 11.X API or PS4 apis, then maybe the compiler can translate a mantle instruction to a console instruction. Maybe native support for Mantle is not needed. Then to pc each hardware vendor migth be able to provide their "translation" functions or set of functions from a Mantle api call to a specific hardware call or something like that.

Log in

Don't have an account? Sign up now