[PcPer] Some mantle news from GDC

[Red Hawk]

So Mantle is officially dead. DX12 won on TKO.

Always with the sensationalist negativity, Shintai.

No, this doesn't mean that Mantle is necessarily dead, it means they're changing the focus of Mantle. What focus that is remains to be seen; if it's a really insignificant focus that ends up with no market adoption, then we'll be able to say that Mantle is dead.

And acting like DX12 "beat" Mantle is disingenuous. It's questionable if DX12 would exist with its focus on close-to-hardware programming if not for Mantle pressing the issue first. Once Microsoft threw its whole DirectX development into close-to-hardware, it was inevitable that it would supersede Mantle and AMD would see little benefit in trying to directly compete. But just the fact that DX12 is so close-to-hardware can be seen as an indirect victory for AMD and the direction they wanted to push PC game development towards.

 

[boozzer]

mantle had alot of influence on dx12. I would say good job to amd.

 

[Gloomy]

http://anandtech.com/show/9038/next-generation-opengl-becomes-vulkan-additional-details-released

In fact Khronos has confirmed that AMD has contributed Mantle towards the development of Vulkan, and though we need to be clear that Vulkan is not Mantle, Mantle was used to bootstrap the process and speed its development, making Vulkan a derivation of sorts of Mantle (think Unix family tree). What has changed from Mantle is that Khronos has gone through a period of refinement, keeping what worked in Vulcan and throwing out portions of Mantle that didn’t work well – particularly HLSL and anything that would prevent the API from being cross-vendor – replacing it with the other necessary/better functionality.

Here's a list of Mantle function names: http://www.geeks3d.com/20140203/amd...ngl-extensions-and-mantle-api-functions-list/

In this PDF you can find a screenshot of a prototype debugger for Vulkan: https://www.khronos.org/assets/uploads/developers/library/overview/2015_vulkan_v1_Overview.pdf

If you compare the function names you'll see that the Vulkan functions are named identical to Mantle, and the only difference is the prefix.

AMD handled Mantle exactly like they said they would by attempting (and succeeding) in creating an industry standard API around the concept. What's with the hate here? You guys need to chill and stop being wrong all the time.

 

[3DVagabond]

Anyone with eyes can see they are the same. People who continue to deny it are just looking foolish.

HLSL is MSFT, so no surprise that's gone and then anything that would prevent it from being multi vendor. Again, no surprise.

 

[Shmee]

Mantle still would be good to be developed as it could be used on Linux or Windows versions that do not get DX12.

 

[Spanners]

You cant be serious. Just because they got similar names you think they are the same API and code?

I think similar is underselling it a little, they are identical besides a new prefix.

 

[Silverforce11]

http://anandtech.com/show/9038/next-generation-opengl-becomes-vulkan-additional-details-released

Here's a list of Mantle function names: http://www.geeks3d.com/20140203/amd...ngl-extensions-and-mantle-api-functions-list/

In this PDF you can find a screenshot of a prototype debugger for Vulkan: https://www.khronos.org/assets/uploads/developers/library/overview/2015_vulkan_v1_Overview.pdf

If you compare the function names you'll see that the Vulkan functions are named identical to Mantle, and the only difference is the prefix.

AMD handled Mantle exactly like they said they would by attempting (and succeeding) in creating an industry standard API around the concept. What's with the hate here? You guys need to chill and stop being wrong all the time.

They're hating it because soon, their hardware will be running secretly AMD code in DX12 or OpenGL. That's a taint within their pure AMD-free system, completely unacceptable.

I really can't see any other reason why there's such rabid hate for Mantle to live on & improved upon in two very open API. It's a win-win situation for gamers and developers.

 

[el etro]

I don't have words to quantify my indignation seeing both Mantle and DX12 reception.


DX12 does the exactly same thing as Mantle does, but only the former gets praised. Many ones describe the latter as pure Hype.

This is a truckload of hypocrisy.

 

[Fox5]

Yup, if Mantle is truly dead then it benefited everyone for pushing the same features into DX12 but didn't do much for cash strapped AMD.

It puts AMD's cpus back into the running for games.

I think AMD's development on Mantle probably pre-dated DirectX12 though. They probably gave a tech demo to Microsoft early on in Mantle's development (2011/2012), and Microsoft liked it enough that they decided to develop a similar technology. AMD kept going with Mantle and got something out the door first, and was able to get a few major game engines supporting a DX12 like API.

Microsoft had incentive to develop a DX12 like API for the Xbox 360 with its multiple weak cpu cores. They never did, the Xbox 360's api was somewhere between DX9 and DX10. They could have even launched the API with their new console, but they didn't. Sony created such an API with the PS3, Microsoft doesn't have DirectX12 ready until late 2015.

And OpenGL was probably just lagging behind like always.

 

[dacostafilipe]

http://anandtech.com/show/9038/next-generation-opengl-becomes-vulkan-additional-details-released

Here's a list of Mantle function names: http://www.geeks3d.com/20140203/amd...ngl-extensions-and-mantle-api-functions-list/

In this PDF you can find a screenshot of a prototype debugger for Vulkan: https://www.khronos.org/assets/uploads/developers/library/overview/2015_vulkan_v1_Overview.pdf

If you compare the function names you'll see that the Vulkan functions are named identical to Mantle, and the only difference is the prefix.

AMD handled Mantle exactly like they said they would by attempting (and succeeding) in creating an industry standard API around the concept. What's with the hate here? You guys need to chill and stop being wrong all the time.

It's not just the names that are the same, but also the count, type and order of attributes.

Look at how vkQueueSubmit is used, and compare it with this (page 32): http://de.slideshare.net/DevCentralAMD/mantle-introducing-a-new-api-for-graphics-amd-at-gdc14

It's obvious that Vulkan is an improved, HLSL-free Mantle.

PS: Technically, it's a fork of Mantle.

 

[gamervivek]

http://www.anandtech.com/show/9038/next-generation-opengl-becomes-vulkan-additional-details-released

and though we need to be clear that Vulkan is not Mantle

What has changes is that Khronos has gone through a period of refinement, throwing out portions of Mantle that didn’t work well in Vulkan – mainly anything that would prevent it from being cross-vendor – and replacing it with the other necessary/better functionality.

Vulkan is not Mantle.

Mantle was dead on arrival from the start, just accept it.

hahaha holy selective quoting batman.

Yeah Mantle was dead on arrival, but apparently it had an organ donor card. :awe:

 

[NTMBK]

It's not just the names that are the same, but also the count, type and order of attributes.

Look at how vkQueueSubmit is used, and compare it with this (page 32): http://de.slideshare.net/DevCentralAMD/mantle-introducing-a-new-api-for-graphics-amd-at-gdc14

It's obvious that Vulkan is an improved, HLSL-free Mantle.

PS: Technically, it's a fork of Mantle.

This x1000. They've openly told Anandtech that it is a derivative of Mantle- they took Mantle and ran with it.

 

[AnandThenMan]

Can someone explain from a business point of view what AMD is doing with Mantle? I don't get it, they seem to have put in some incredible development into Mantle (and shook up the industry in the process), then basically gave it away. Why?

 

[exar333]

I don't have words to quantify my indignation seeing both Mantle and DX12 reception.


DX12 does the exactly same thing as Mantle does, but only the former gets praised. Many ones describe the latter as pure Hype.

This is a truckload of hypocrisy.

That's because AMD 'is for the people' and M$ (so the hater's recognize) is only out for money. :/ (hope everyone picked-up on the sarcasm here)

I still doubt the real impact Mantle had on DX12, as it was also being worked on for a number of years. Regardless, its a good thing it is gone...vendor-specific APIs are NOT good for the industry. DX12 looks fantastic and should be great.

 

[Gikaseixas]

Can someone explain from a business point of view what AMD is doing with Mantle? I don't get it, they seem to have put in some incredible development into Mantle (and shook up the industry in the process), then basically gave it away. Why?

My unofficial theory: They pushed and persuaded Microsoft into the Mantle train and basically offered the basis for DX12 for free in exchange of a industry-wide adoption of a API that puts their cpu division in a better light.

 

[NTMBK]

Can someone explain from a business point of view what AMD is doing with Mantle? I don't get it, they seem to have put in some incredible development into Mantle (and shook up the industry in the process), then basically gave it away. Why?

Because better APIs means games which make better use of new GPUs, which means more GPU sales. "Giving it away" means that a larger number of GPUs will support it, which means more developers will add support in their games, which means more games which make good use of new GPUs, which again means more GPU sales.

Yes, it also helps NVidia, but that is the price you pay for openness. It's the sort of move NVidia should have made with CUDA years ago. It's one of those "rising tide lifts all boats" situations.

 

[AnandThenMan]

It benefits Nvidia as much, or possibly even more this is not a good business plan. The CPU angle, maybe but that seems like a convoluted and expensive way (and risky) to try and make your CPUs look better.

 

[geoxile]

That's because AMD 'is for the people' and M$ (so the hater's recognize) is only out for money. :/ (hope everyone picked-up on the sarcasm here)

I still doubt the real impact Mantle had on DX12, as it was also being worked on for a number of years. Regardless, its a good thing it is gone...vendor-specific APIs are NOT good for the industry. DX12 looks fantastic and should be great.

Work on mantle started in 2008. DX12, most likely around 2010.

 

[dacostafilipe]

Can someone explain from a business point of view what AMD is doing with Mantle? I don't get it, they seem to have put in some incredible development into Mantle (and shook up the industry in the process), then basically gave it away. Why?

It certainly helped in terms of image.

Having Mantle run on real Hardware so early, could have shaped future hardware development. This could lead to hardware that is best suited for Mantle-like API's (ex: removing possible bottlenecks). We already saw AMD improving the ACEs on PS4/X1 to allow a lot more command batches.

Also, because DX12/Vulkan makes the usage of compute a lot easier this will certainly show the compute superiority of GCN.

Disclaimer: No, I have no source for this. Just random stuff that crossed my mind.

 

[Headfoot]

Can someone explain from a business point of view what AMD is doing with Mantle? I don't get it, they seem to have put in some incredible development into Mantle (and shook up the industry in the process), then basically gave it away. Why?

Risk management. AMD could not rely on DX12 and OpenGL to meet what they saw as the design they needed -- thus make proof-of-concept Mantle drawing upon concepts successfully used in the consoles. If DX12 and OpenGL bombed, DX12 ended up just being DX11 + new features like all the other DX's before it, then Mantle would have stood above. But Microsoft stepped up its game, and because Mantle set the precedent, DX12 must at least do the major good things from Mantle or it would be seen as a failure. AMD can't predict the future, so the only way to ensure that such an API would exist would be to build one.

End result? AMD has an API in the marketplace that has the console like features and reduced CPU overhead they want whether by Mantle or DX12 or OpenGL

 

[n0x1ous]

Thank you AMD for getting this ball rolling and doing us all a favor.

 

[Headfoot]

Mantle was dead on arrival from the start, just accept it.

Vulkan is a fork of Mantle. You understand what a fork is, right? You understand a "multi-vendor ready" release of Mantle called Mantle would still be a fork, right? You realize that what name you give a fork is 100% irrelevant as to what the actual written code's history is, right?

Let me guess, if its not the exact same code, it's "not Mantle." In which case, you haven't used Windows since 95. Or since the first patch tuesday for Windows... And Unix and Linux are not related at all under that logic.

 

[96Firebird]

So many people just can't let Mantle go... Let it die in peace.

 

[caswow]

So many people just can't let Mantle go... Let it die in peace.

we get mantle under a different name. thats right :thumbsup:

 

[AnandThenMan]

So many people just can't let Mantle go... Let it die in peace.

Mantle is not dead, in fact it ended up in multiple APIs.