Mantle API presentation by AMD, DICE and Oxide - AMD Summit 2013

[phanbuey]

So that means mantle is not for you, why are you participating in this discussion then and painting mantle as not being successful because it doesn't do what you want ?

It's not designed to do what you want, that makes everything you said invalid towards mantle it self.

Yes your opinion is fine, but it's not mantle's fault, it's not a downside to mantle, it doesn't affect mantle, you and everyone else can have such opinions about mantle, but they are all irrelevant in the big picture.

Using populairty and self-benefit to compare different types is all invalid, you guess I probably dismiss any sort of FPS vs RTS comparison, in stupid gaming sites where they have GOTY, it's all a joke, there should be GOTY for all genres separately, there is no one game better then all the rest.

actually it is for me - i game and i love graphics and i loved glide (Voodoo banshee was my first real 3d card), I want better 3d graphics and mantle is designed for that.... soo i dont really see your point? I want mantle to succeed, i really do - i just dont think it will.

 

[the54thvoid]

I can't believe how absurd this thread has become. The main educated arguments ALL agree on the technical merits of the API. I don't think there is anyone that wants Mantle to fail (from the gaming community). The only downside to Mantle the same as other 'pre' viewers, like Anandtech, said is what is being said here and that is adoption. The cost to code has to be covered and some of us are interested in the financial implications which in all fairness are what will prove or kill it. Mantle can't be considered on the basis of its technical merit alone. That is naive. Like all tech it requires very astute business acumen to succeed and even more to 'change' the industry. Comparisons to Physx are poor. G sync is more of a comparison but i think that has too much cost complication for end consumer.
If i am to understand this thread it is my belief the supporters believe that Mantle will become the defacto API for coding in the future? If that's not the thought then a lot of the arguments are for nothing. If people believe Mantle will 'be there' as an accessory to other API's for the foreseeable future then yes, I can well see that being the case.

I can't wait see the effect it has on BF4 and hopefully we'll get a good selection of reviews to gauge it on. I think BF4 will be make or break it from a business investment perspective. I really hope EA let Dice push out Mantle this month and dont hold it back.

 

[Crap Daddy]

I can't believe how absurd this thread has become

All tech forums are full of this sort of mantle threads. I pop in from time to time to see if something is really happening. Nope, nothing.

 

[happita]

Very boring indeed. You hear the same regurgitated nonsense between so many different people arguing that it won't/will succeed, is inferior/superior to DX/OpenGL, won't/will change the industry, etc. Shit is getting old quick in here, I think I'll take a page from Mailman's book and come back to this thread in about a month or so.

 

[Frick]

Very boring indeed. You hear the same regurgitated nonsense between so many different people arguing that it won't/will succeed, is inferior/superior to DX/OpenGL, won't/will change the industry, etc. Shit is getting old quick in here, I think I'll take a page from Mailman's book and come back to this thread in about a month or so.

He said that a month ago, but the trolling opportunities is bringing him back.

 

[RoutedScripter]

Update


Someone says he knows Oxide demo will be out this month:

http://forums.elementalgame.com/451041/page/1/#3430480

 

[HD64G]

At last. Now all will have the experience of what Mantle would provide if they got a CGN GPU. Even from videos on tube. A new thread is going to explode by that time.

 

[the54thvoid]

Update


Someone says he knows Oxide demo will be out this month:

http://forums.elementalgame.com/451041/page/1/#3430480

From that link:

The biggest advantage of Mantle is that unlike DirectX, Mantle is truly multicore aware. With DirectX 11, more cores don't buy you nearly as much. That said, Nitrous, even on DirectX, is still two orders of magnitude faster than say a typical DirectX 9 engine.

A Mantle optimized game can show a massive performance gain depending on how many cores the user has on their CPU. Contrary to what I read on some forums, most games remain CPU or video driver bound (i.e. the GPU is waiting to be fed). Mantle lets you get a lot more stuff onto the GPU.

I come back to the question which seems answered above in bold and counters what people have replied to me - Mantle feeds much more info to the CPU, if it is multi core, which lets the GPU not get bogged down by CPU coded limitations.

Also, the Oxide benchmark is not intrinsically a Mantle benchmark so we'll all be able to run it, so we can get good comparisons of GCN benefits versus non GCN cards.

Finally, if a game is coded for Mantle, does that imply the GCN cards will natively run that API or will the Catalyst drivers need to be tweaked?

 

[RoutedScripter]

The game uses mantle for it's rendering pipeline and the shape and way it self of the pipeline is different, emulating D3D is not the way to go, so that's what Johan meant the engine code looks similar to PS4 one because they have a good low-level API compared to xone API.

No mantle doesn't feed anything.

You need a new driver for mantle, it has nothing to do with directX drivers.

Whether or not both drivers could be installed at the same time is still unknown.

The API code could be in DLLs shipped with the game (version specific), or with the drivers them selfs, we don't know.

 

[15th Warlock]

Finally, if a game is coded for Mantle, does that imply the GCN cards will natively run that API or will the Catalyst drivers need to be tweaked?

It's my understanding that Mantle completely bypasses the graphics driver, speaking directly to the hardware in the video card.

 

[RoutedScripter]

It's my understanding that Mantle completely bypasses the graphics driver, speaking directly to the hardware in the video card.

A wrong understanding.

 

[15th Warlock]

A wrong understanding.

So you're saying Mantle still uses the Catalyst driver? because that was the54thvoid's question, and from what I've read the intention of using Mantle is to completely bypass said driver, working as an API that communicates directly to the hardware, thereby, bypassing all the Catalyst driver layers...

Now you're confusing me...

 

[erocker]

It's my understanding that Mantle completely bypasses the graphics driver, speaking directly to the hardware in the video card.

I think that video cards will still "need" drivers to work with Mantle. It's Direct X (mostly) that is getting bypassed/replaced.

 

[15th Warlock]

I think that video cards will still "need" drivers to work with Mantle. It's Direct X (mostly) that is getting bypassed/replaced.

I see, thanks for clearing that up!

 

[HD64G]

Some dll's in catalyst and some profiles for the games using it, will enable Mantle for CGN GPUs. No need to install 2 set of drivers.

 

[Frag_Maniac]

Yeah, a graphics API should just be a thin abstraction layer between the graphics driver and graphics program (game). There won't be a need for an additional driver, though that isn't to say AMD won't be doing a bit of driver reworking to fine tune performance.

For those whom may not know, abstraction in this context just means pulling out bits of data, but the API will also go further and sort and assign draw calls. Mantle is also capable of very intricately finding errors through it's wealth of debugging coding.

Debugging is where it's going to implement never before used features in a graphics API. The debugging feature set alone should make development go much smoother, while allowing devs to confidently push the envelope on what they can do in games.

 

[RoutedScripter]

Some dll's in catalyst and some profiles for the games using it, will enable Mantle for CGN GPUs. No need to install 2 set of drivers.

There is no any app-specific driver code any of those mantle games will use.

Most of the validation is on developer's hands, the API and driver has minimal to no validation ("code security") compared to what currently exists in DX and DX drivers.

 

[Mathragh]

Quote from Frogboy, Oxide/Stardock developer: "BTW, Star Swarm (the benchmark) is going to go up on Steam this month so all this admitted hyping I'm doing is something you will be able to verify first hand in a few weeks."

Star swarm is the mantle demo showed by Oxide, so this means that we'll see any mantle improvements on our own PC by the end of this month.

Edit: another quote from later in the thread:
"
Hmm, is the version of Star Swarm being released on Steam going to have any sort of modding support? Cause I can't help but drool a little over the potential this offers.


//


Yes. It's very moddable. It would be interesting to see someone put together the Revenge of the Sith opening battle with it
"

 

[Steevo]

Just in time for me to replace my GPU, since it died.

 

[happita]

And you sir, just got to the 1,000 thanks mark. Congrats, now go get an R9 card to celebrate this wonderous occasion

 

[Steevo]

I wonder if W1zz will take it on trade for a 290 that i can slap my block on. I am down to my laptop and phone till I get the money for one, so no gaming and my rig sits idle.


On a side note whats the easiest way to remove the GPU to make something out of? Oven on foil, flameless torch?

 

[Frag_Maniac]

There needs to be a comprehensive demo other than Star Swarm, which many have already seen, that isn't biased toward RTS titles. Granted RTS is where Mantle can show big performance improvements, but they need to show they can bring big performance gains in the more popular genres.

 

[crazyeyesreaper]

Yea Mantle in Total War would be nice. 50k troops without a 5ghz Intel CPU would be nice hahaha.

 

[erocker]

Yea Mantle in Total War would be nice. 50k troops without a 5ghz Intel CPU would be nice hahaha.

It's funny...

because it's true.

 

[crazyeyesreaper]

Yea Most people are lucky to keep 6-10k troops running at 30 fps with todays CPUs all heavily limited by single thread. 1 Thread for Game 1 for AI thats about it. Animation subsystem needs multi threading per soldier would be impossible but per unit? that could be done. Oh well one can dream haha