AMD Mantle API [updating]

As an ex AAA Game Dev I only want to see Mantle before I applaud, OpenGL is appaling and D3D is lame, we need something better that let us leverage the true potential of the hardware.
If you're unhappy with the way things turn, then blame MS and the ARB/Khronos for providing so poor API !

 

@Rodéric: D3D is awesome and WDDM rocks!

// ducks

 

The situation right now though is that nobody gets the best experience in any game. If in future half of games were best on Nvidia's API and half were best on Mantle, we'd still be a lot better off than we are at this moment.

 

No plans on it, we simply didn't have time to fully evaluate it to see if it would be a benefit for BF specifically.

 

i know you likely cant comment yet, but is mantle really that far away from the console API's? logic to me would dictate that if your close to the hardware there are only so many good ways to construct a language/API to interact with it.

 

That's neurotic, at best. No offense.
If I develop a game which somehow utilizes a 290X at 100% vs. Titan 66% without Mantle, that argument still applies, but it is utter nonsense. Games are not developed with bringing justice to the world, they are developed because of the motivations of the individuals in a team (art and tech), and the common motivation of the team. The decision to use Mantle is not for the consumer to make or to dictate. A developer will use Mantle when it sees fit. And if a developer has 20 years experience with Nvidia architectures, and will only support NvAPI, then that's his decision alone, and no amount of crying about injustice will change that.
This decision has no negative impact on consumers or licensees of tech, period.

 

3D sound positioning (ie including elevation) would have been very welcome though (trueaudio or not)

 

Fair enough, thanks for the answer. I understand BF3 is already doing some pretty advanced stuff with audio anyway (presumably on CPU) so I'd assume BF4 will at least match that if not go beyond. Unfortunately I didn't see much press coverage on the audio side of BF3, that's just what I've heard on here.

May have to go and replay the SP again to see what I can spot. It seems that since all this talk of advanced audio has cropped up my ears have become much more attuned to flaws in game sound. I found the new Tombraider to be pretty weak on the sound quality (positional audio, reverb etc..) front despite looking incredible. To the point of being distracting at times.

EDIT: and now back to Mantle, sorry for the OT.

 

Be aware its not possible to finish the BF3 single player game if you rebind your keyboard

ps: BF2 was one of the few (if only) games to support the x-fi and its x-ram was that support removed from bf3 ? (too lazy to fire it up and check)

 

Let say, in TR, there's a lot of useless sound who are there without any reason, each time you approach a zone with enemy, you will hear them speaking, its the alarm you need take care..Even if the guy is alone you will hear him speak on the radio, call another guard etc..

But they have put many other sounds who make you believe someone or an animal is walking near you, on the corner etc, but there's nobody. ( im not sure this was the effect they wanted, maybe put an atmosphere around you.. ) .

On some specific part, the sound is incredible ( wind, rain, snow, tempest ), seriously, but on most openzone it is just a bit under.. This said, it was an extremely fun game to play and finish. Not perfect in all aspect, maybe some peoples was waiting something more like the old TR. But a good game.

 

My understanding is that NVIDIA does support Driver Command Lists in D3D, but that feature is not very useful at all for performance reasons. I guess it never panned out as Microsoft had planned.

 

I cannot imagine that an IHV-specific graphics API such as Mantle that is specifically tuned for lower level access to GCN-equipped hardware will ever become an industry standard API. This was attempted in the past with 3dfx's Glide API, and that API faded away as DirectX and OpenGL evolved and improved. If that is your ultimate goal, then fine, but there will be short-term pain for any IHV that is not able to take advantage of a vendor-specific render path in your game engine.

PC graphics hardware innovation is as strong as ever. In the near future, the CPU and GPU will share a unified virual memory space. DirectX and OpenGL and various OS's will evolve and improve to accomodate these newer architectures. In the ultra mobile space, there are a variety of different graphics vendors including NVIDIA, Intel, Qualcomm, ImgTech, ARM, Vivante, and the graphics performance is increasing exponentially in this space in the short-term. For these reasons, it makes little sense in my opinion to pursue or promote IHV-specific graphics API's over the longer term.

 

Somehow you still don't get it -- every modern engine has vendor-specific render paths, often more than one per vendor even. Why? Because D3D isn't 'just" D3D on every platform; each IHV (and many of the individual hardware generations within that IHV's portfolio) require their own unique tricks to get optimal performance and/or image quality -- or to choose the proper trade off.

That work is already happening, today, and it's not going away.

 

You're approaching this too much as a developer. You don't think AMD fans were negatively impacted about msaa being locked to nvidia on batman? Proprietary APIs hurt consumers. That's not to say all proprietary APIs are bad or that Mantle is bad, but you're naive in saying they have no negative impact.

And btw, last time I checked customers are paying for your product; they can dictate whatever they want.

 

Well yes, but that is all done within the framework of industry standard graphics API's such as DirectX and OpenGL, not through IHV-specific graphics API's.

 

If D3D worked the way you hypothesize it should, then it shouldn't be necessary to do vendor-specific render paths...

With the myriad extensions available per-IHV and per-hardware-class, this isn't so easy as you make it to be.

The reality is this: Mantle was designed not to entirely replace D3D, as has been quoted too many times to count, but to augment it. I will always and entirely trust a AAA developer (repi) to tell me what's up with that, instead of some no-name n00b (you).

 

Locking a code path that uses a standard API and would run on any card with enough performance to specific vendor IDs is not comparable to a game simply running faster because a card supports a faster way to run the game. People complain when features that should otherwise run acceptably are artificially restricted to certain vendors. In that case a certain group of users are being harmed specifically in a product they paid for.

If Mantle adoption just makes games run faster on AMD hardware while the nVidia performance remains the same as it would have been in any case, no one is getting hurt and people can make informed purchasing decisions with that knowledge just as they do today when comparing game performance between different cards.

 

That is ridiculous. Why would the industry need or want IHV-specific graphics API's to "augment" industry standard API's? 3dfx already tried that with Glide, and it failed. And even Johan admits that it would be extremely bad if other IHV's tried to pursue their own IHV-specific graphics API. And FWIW, there are some big name AAA developers such as Tim Sweeney (and others in the industry such as Carmack) who believe that Mantle is the wrong step for the industry, which is certainly more trustworthy than some no-name (you).

 

Howabout this? For some games, AMD is going to get faster, and NVIDIA isn't.

Don't like it? Too bad.

 

Well it is not just NVIDIA, it is Intel, Qualcomm, ImgTech, ARM, Vivante, etc. But who's counting