TR Forums

michael_d wrote:

I am waiting for Carmack (ID Tech 5), Sweeney (Unreal 4), Crytek and the dudes behind Metro 2003/Last Light 4A Engine to weigh in on this. I was not impressed with Frostbite 3 at all but at least we will be able to compare DX11 and Mantle when BF4 is out.

What's wrong with Frostbite... at this point in time, despite there's not a single game that uses all of the engine, not even BF4, I would think it's probably one of the nicest engines out there. And I expect it to continue to improve since they basically have a whole minor development house on the engine alone.

I have yet to see as good games from most of the other engines. Although I would say that all the engines you mentioned have great strengths, but as for the games out there... it's a bit lacking, although that cant really be attributed to the engine alone.

Fighterpilot wrote:

JohnC wrote:

:lol: Warsam, people can read official PR releases, watch official video presentations and see official slides on AMD's own website and other places, this thread was not intended for PR propaganda or pricing lists...

Umm did you forget that you aren't a mod?
Show a bit of class when a manufacturer's rep bothers to post.

1)I never assumed that I was a mod. Please stop trying to force your personal assumptions onto others, it's not a very positive personal trait.
2)I never asked for him to post this obvious "brand awareness" propaganda, which I personally dislike (regardless from which brand it comes - AMD, Intel or Nvidia). And I have simply expressed my personal subjective dislike for such information in my reply. Without attacking that person's sexual/religious preferences, intelligence level or nationality/ethnicity. Or am I not allowed to do that anymore?
3)If you have nothing to say about the topic - please don't say anything at all or at least keep it to PM's; focusing on forum participant's personality or behavior is not very helpful for any other participants (and just because others do it it doesn't mean that you should do the same). Thank you.

Disclaimer: I like the Frostbite engine. Having said that...

Frostbite isn't the most efficient engine out there. A good engine must have a balance of good visuals, speed, and available developer tools. The latter aside, the only recent engine that fills those requirements, IMO, is the iD Tech 5 engine. Despite a... patchy.... launch (zing!), I was dumbstruck at how good it looks while maintaining excellent frame rates.

Now, when you put developer tools into the equation, Unreal Engine 4 will probably wipe everyone off the map, I'm afraid.

morphine wrote:

Now, when you put developer tools into the equation, Unreal Engine 4 will probably wipe everyone off the map, I'm afraid.

Perhaps it will... Large companies like EA and Activision are highly unlikely to ever use it, though.

Aphasia wrote:

michael_d wrote:

I am waiting for Carmack (ID Tech 5), Sweeney (Unreal 4), Crytek and the dudes behind Metro 2003/Last Light 4A Engine to weigh in on this. I was not impressed with Frostbite 3 at all but at least we will be able to compare DX11 and Mantle when BF4 is out.

What's wrong with Frostbite... at this point in time, despite there's not a single game that uses all of the engine, not even BF4, I would think it's probably one of the nicest engines out there. And I expect it to continue to improve since they basically have a whole minor development house on the engine alone.

I have yet to see as good games from most of the other engines. Although I would say that all the engines you mentioned have great strengths, but as for the games out there... it's a bit lacking, although that cant really be attributed to the engine alone.

Frostbite graphics look "synthetic", world seems to be made of "plastic" and physical objects seem to have no "weight". Lighting and particles look subpar especially in comparison to CryEngine and 4A Engine. Perhaps this is due to poor art work of BF3.

The lighting does look extremely overdone in BF3 but I think it's just the way the game was intended to look like, not a fault of the engine itself... I'd say people should wait for next Dragon Age, it's being done by different team so it might look significantly different even though it uses same engine.

jihadjoe wrote:

michael_d wrote:

If Mantle is low level API it will probably require more coding and development time. Direct access to the hardware will probably require more responsible coding to prevent crashing/BSODs.

Actually I had exactly that in mind, and was thinking Mantle probably wouldn't gain much traction because of the extra work involved and those potential pitfalls. Then I realized that developers are probably ALREADY coding for Mantle because of the PS4 and XBone. Does anyone really believe that XBone is going to be using DX? And PS4 will be on OGL? Neither Sony nor Microsoft will want to leave any performance on the table, so the consoles are very likely being coded to a lower level API which I'll bet a beer is Mantle.

So now all AMD has to do is expose Mantle in Windows, and suddenly every game that's also on console is faster when run on their GCN hardware. It's a brilliant move by them, although potentially bad for us in the long run when GCN is out of date. What's low-level now is potentially baggage later on, but I doubt AMD is thinking that far. Right now they just want that performance lead and $$$.

If console developers are using Mantle then it will take almost no afford to port to PCs. Hopefully developers will spend extra spare time on enhancing PC experience both physics and visuals. But some lazy bums might just leave PC version no different than console.

Console developers are not using Mantle. XBone has its proprietary DX-whatever and PS4 has its proprietary GNM API's. I don't believe these have anything to do with Mantle (other than the fact that they are also proprietary close-to-metal API's). I might be wrong, though, but so far I hasn't seen anything that says otherwise.

jihadjoe wrote:

michael_d wrote:

If Mantle is low level API it will probably require more coding and development time. Direct access to the hardware will probably require more responsible coding to prevent crashing/BSODs.

Actually I had exactly that in mind, and was thinking Mantle probably wouldn't gain much traction because of the extra work involved and those potential pitfalls. Then I realized that developers are probably ALREADY coding for Mantle because of the PS4 and XBone. Does anyone really believe that XBone is going to be using DX? And PS4 will be on OGL? Neither Sony nor Microsoft will want to leave any performance on the table, so the consoles are very likely being coded to a lower level API which I'll bet a beer is Mantle.

So now all AMD has to do is expose Mantle in Windows, and suddenly every game that's also on console is faster when run on their GCN hardware. It's a brilliant move by them, although potentially bad for us in the long run when GCN is out of date. What's low-level now is potentially baggage later on, but I doubt AMD is thinking that far. Right now they just want that performance lead and $$$.

i dont think it will be as bad for future hardware as you think - mantle is just replacing DX and will still have its own drivers so theoretically newer graphics architectures could just have their own drivers written for mantle in the same way newer architectures have their own drivers written for DX

i just see mantle as a part time replacement for DX - DX11 will still be around and used but some devs that want that little bit more performance out of their engines can target mantle (and hopefully an nvidia equivalent) for extra performance and target DX11 at the same time for compatability (notice BF4 has a DX as well as a mantle path)

some people think its the "end of the world" and we are moving "back to the stone age" with mantle but in my view its the best way forward as microsoft dont want people playing games on windows anymore they want us to play on their Xbone so advancements in DX have stalled and imo this is the best way forward
i just hope nvidia follow suit and makes their own low level api for their hardware to target then they both have control over their DX replacements and can advance them with their hardware while also giving developers much needed lower level access to the hardware

f0d wrote:

i just hope nvidia follow suit and makes their own low level api for their hardware to target

I hope you're not serious... That will most likely increase the developer's costs and time, which undoubtedly will be passed onto the consumers. Nvidia might still do it out of spite (because AMD just did it) but I hope they won't.

Why not instead work together to continue improving OpenGL, adding in new features and reducing overhead, etc.?

The only reason why this might work is because it's almost certainly already being used for the PS4 and Xbone. In which case this would actually make porting easier.

JohnC wrote:

f0d wrote:

i just hope nvidia follow suit and makes their own low level api for their hardware to target

I hope you're not serious... That will most likely increase the developer's costs and time, which undoubtedly will be passed onto the consumers. Nvidia might still do it out of spite (because AMD just did it) but I hope they won't.

Why not instead work together to continue improving OpenGL, adding in new features and reducing overhead, etc.?

because it takes too long - openGL (as well as DX) updates are few and far between
you will never get all the manufacturers (intel/amd/nvidia) to agree on anything within a reasonable timeframe

look at mantle - right out of the gate it efficiently supports their hardware without having to worry about DX or OGL supporting it

i agree an open graphics library would be for the best but why not HAVE BOTH? its not like they will just stop advancing OGL or DX because of mantle and mantle is optional anyways so its not like mantle is taking over anything
its entirely OPTIONAL

Options are fine as long as hardware vendors won't start throwing more of their money at game developers and "encourage" them to use a particular (their own) API... Which has a higher chance of happening if both Nvidia and AMD will have their own API's. And if both AMD and Nvidia will encourage a same game developer to use both of their API's it will also increase the development time, which is still not a positive thing.

JohnC wrote:

Options are fine as long as hardware vendors won't start throwing more of their money at game developers and "encourage" them to use a particular (their own) API... Which has a higher chance of happening if both Nvidia and AMD will have their own API's.

again it will still be optional
sure amd might start throwing money at companys (like ea/dice) to use mantle - but they will still target DX also (like bf4 uses both)
so we wont be any worse off than we are now with just DX as the new "low level api" games will still target DX also

what i think will happen is just a few highly technical/graphical games (like bf4) will target these new low level api's (both nvidia's and amd's) but most games will still target just DX for most games/engines that DX isnt much of a limitation for

so there is room for improvement in DX still as there will still be plenty of use for it and in a way i agree with you but i hope that microsoft/khronos can see the threat of these api's and get off their ass and start doing something FAST instead of taking years and years like they have been

JohnC wrote:

And if both AMD and Nvidia will encourage a same game developer to use both of their API's it will also increase the development time, which is still not a positive thing.

dice were able to spit out a new battlefield and support mantle fairly fast - i dont think development time will be affected all that much imo

Yes, but here is a more precise example for you. Let's say I am currently satisfied with performance of all my other games and my current Nvidia card BUT I also want the BEST possible performance out of BF4 when playing it. Since the BF4 will be using Mantle and upcoming AMD's top (or even mid-range) GPU will most likely perform much better than even Nvidia's Titan (which is forced to use DX11) when using Mantle I am basically artificially "encouraged" to buy AMD's hardware, just for this single game. Even though the game will still technically run on my Titan. And if Nvidia will go the same way - same thing will happen to AMD users. Meaning that if you want to get the BEST performance out of ALL your favorite games you will be artificially encouraged to want to have cards by both of these vendors. Which is wrong, regardless of which particular brand someone might worship (even though I can afford having both of AMD's and Nvidia's most expensive cards I know that plenty of people cannot).
I know that similar situation exists even today with all the nonsense "The way it meant to be played" campaigns, even with games using DX/OGL, but I am afraid that the situation (the performance difference) will get worse with proprietary API's...

JohnC wrote:

I know that similar situation exists even today with all the nonsense "The way it meant to be played" campaigns, even with games using DX/OGL, but I am afraid that the situation will get worse with proprietary API's...

And it started with 3dfx/Glide vs. OpenGL.

Yes, it did, but there were other API's as well. And I am glad that Glide and S3's MeTaL has died off - I definitely didn't like the fact that I had to buy a card with a GPU from a specific vendor (S3) to make Unreal and Unreal Tournament to look its best even though that card didn't really perform well with other games. And hopefully I won't be forced to do the same in future.

JohnC wrote:

Yes, it did, but there were other API's as well. And I am glad that Glide and S3's MeTaL has died off - I definitely didn't like the fact that I had to buy a card with a GPU from a specific vendor (S3) to make Unreal and Unreal Tournament to look its best even though that card didn't really perform well with other games. And hopefully I won't be forced to do the same in future.

All depends on how well the hardware abstraction layer is written. Can't blame AMD for trying to leverage its console hardware wins into the PC space.

Captain Ned wrote:

Can't blame AMD for trying to leverage its console hardware wins into the PC space.

Of course, it's natural for them to try such thing, I just hope it won't turn into the very extreme situations I've explained above and hopefully Nvidia won't try to do the same... I'd rather prefer AMD to focus more on fixing the current issues with DX/OGL drivers and do it in a more rapid way.

JohnC wrote:

Of course, it's natural for them to try such thing, I just hope it won't turn into the very extreme situations I've explained above and hopefully Nvidia won't try to do the same... I'd rather prefer AMD to focus more on fixing the current issues with DX/OGL drivers and do it in a more rapid way.

What if, and this is completely speculative, Mantle does do it better?

Mantle will be a big deal if and only if they get nVidia on board to use it too. Given the games Microsoft has played with DirectX over the years nobody sees it as a reliable long-term platform anymore but no sane developer is going to alienate half their customers because they have the wrong card.

JohnC wrote:

Console developers are not using Mantle. XBone has its proprietary DX-whatever and PS4 has its proprietary GNM API's. I don't believe these have anything to do with Mantle (other than the fact that they are also proprietary close-to-metal API's). I might be wrong, though, but so far I hasn't seen anything that says otherwise.

Anand thinks otherwise:

What’s not being said, but what becomes increasingly hinted at as we read through AMD’s material, is not just that Mantle is a low level API, but rather Mantle is the low level API. As in it’s either a direct copy or a very close derivative of the Xbox One’s low level graphics API. All of the pieces are there; AMD will tell you from the start that Mantle is designed to leverage the optimization work done for games on the next generation consoles, and furthermore Mantle can even use the Direct3D High Level Shader Language (HLSL), the high level shader language Xbox One shaders will be coded against in the first place.

As I mentioned in my previous post, I don't think Mantle would gain traction if it was just another new API. AMD knows that, and the reason they're so confident is because Mantle is already THE low level API in use on the consoles.

Interesting. I still would like to see more detailed info on how it directly relates to console development, such as how you can port the Mantle-based games to proprietary PS4-specific APIs. Hopefully AMD will give more details in November...

We pretty much have 5 game engine from 5 companies

Epic, Crytek, ID, Dice and Unity

We already know that Dice will be releasing a Mantle power version of their engine.

Speculation: What we might actually see if this is a success for them, is that Dice will incorporate Mantle as their own abstraction layer.
So Dice would use Mantle as the abstraction for the Xbox360/XboxOne/Ps3/Ps4/AMD PC & Non AMD PC platform.

Currently they have a proprietary abstraction layer that take into account all HW & available API, but that maps to nothing directly.
So Mantle bring a lot of benefit for cross platform engine development and why Dice demanded Mantle to be a cross platform API.

So this is not an added layer, it would replace what they currently have in place.
Benefit, you bypass DirectX11 or OpenGL entirely on the PC, removing a lot of latency, bloat, inefficiency and limitations.
Not only on AMD HW, but any HW vendor that decide to offer a Mantle API on their core driver...
If they dont, its:
Mantle -> DirectX/OpenGL/... - >Core driver

Today its (No diference as a game using Mantle on non supported HW, like older AMD cards. that are not GCN)
Proprietary Abstraction Layer -> DirectX/OpenGL/... - >Core driver

With Mantle
Mantle - >Core driver

So for platforms that decide that doing a native Mantle layer is not worth it, nothing changed.. you go through legacy APIs (OpenGL / DirectX) to access the GPU

To recap this is what all those 5 engine do today:
Proprietery Abstraction Layer -> DirectX/OpenGL - >Core driver

This is what Dice is doing
Mantle -> DirectX/OpenGL[optional, only used on platform without Mantle on their core driver] - >Core driver


The end result, game that run on a platform with a Mantle will be able to lower CPU requirements... and possibly also GPU.

Their is absolutely no negative to all this.

NovusBogus wrote:

Mantle will be a big deal if and only if they get nVidia on board to use it too.

The Mantle is a very close-to-metal API and very hardware-specific, no? As such I do not see how can nVidia get "on board". And if they could and it is a truly "open" API for everyone - why hasn't Nvidia (or any other hardware vendor) announced support for it?

JohnC wrote:

The Mantle is a very close-to-metal API and very hardware-specific, no? As such I do not see how can nVidia get "on board". And if they could and it is a truly "open" API for everyone - why hasn't Nvidia (or any other hardware vendor) announced support for it?

^ Pretty much that.
AMD themselves said that Mantle is an API for working with GCN.
The API may be Open, but it doesn't mean it'll work on anybody else's hardware.
I doubt it will even work on HD6000 series and older Radeons.

.

sschaem wrote:

Benefit, you bypass DirectX11 or OpenGL entirely on the PC, removing a lot of latency, bloat, inefficiency and limitations.

Of course.

sschaem wrote:

but any HW vendor that decide to offer a Mantle API on their core driver...

Can they? How much will it cost to them? Will that still allow the applications/games to take advantage of vendor-specific optimizations/features just as well as for AMD's own GPUs? This is what I'd like to see answered. So far I hasn't seen any of these answers in AMD's official info.

.

clone wrote:

AMD will have put in place a challenge Nvidia may not be able to ignore

Or Intel, I heard they were getting serious about graphics, but...for real this time (gentlemen - contain your laughter).

I really didn't think AMD had it in them to pull something like this off. Anandtech lays it out very convincingly though, there are so many little clues I'd be more surprised at this point if Mantle wasn't running on consoles.

I'm really excited to see how these new cards perform when first released compared to when the first Mantle based implementations drop(~December).