• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

Why AMD’s Mantle API will outlive DirectX 12

~Mike~

~Mike~

~Mike~

Soldato
Joined
17 Jan 2011
Posts
9,033
Location
the King's city
Why AMD’s Mantle API will outlive DirectX 12

We had a chance to sit down with AMD’s Chief gaming scientist, Richard Huddy, outside of Intel’s IDF conference last week and talk about the value that AMD sees in maintaining and promoting its own graphics API. There are three major reasons why AMD believes making continued investments in the Mantle graphics API is good for not only AMD, but for gamers in general.

The biggest issue is legacy support. If Microsoft continues its traditional pattern of not supporting its latest and greatest APIs on older operating systems it’s likely that any additional updates other than the initial DirectX 12 release will never be available on the Windows 7 platform. AMD wouldn’t confirm anything directly but they did point out that Microsoft is EOL’ing Windows 7 next year. Windows 7 currently constitutes just over half of the Windows install base. Obviously for the majority of gamers there’s a lot of value in having a cutting edge graphics API available on your platform, and that’s not something Microsoft can offer them after next year. This is where AMD comes in with Mantle to support the latest advances in graphics on a platform that Microsoft has written off. According to AMD there’s no technical reason that they can’t support Windows 7 with Mantle for many years to come. For Windows 7 users who want the latest graphics APIs but don’t want to move to a more recent OS AMD’s GPUs may eventually become the only solution.

The second biggest reason that AMD will be keeping Mantle around is its value as a tool for moving the industry forward and showing off the best of AMD’s architecture through future revisions to the API. AMD has plans to introduce additional revisions to the Mantle API as it releases new GPU architectures so that developers can take advantage of the most advanced capabilities of AMD hardware. As an example of this in action AMD’s original development and announcement of Mantle kick-started a major new trend in graphics APIs: reduced abstraction. API’s that offer lower level access to the hardware have been created by many important groups like Microsoft, Apple, and Khronos in the wake of Mantle’s debut. Thus AMD sees Mantle as a worthwhile endeavor for allowing them to more rapidly define the pace of innovation in the graphics industry.

The final reason that AMD wants Mantle is for performance. AMD believe that in a majority of cases Mantle will provide better performance than DirectX 12 because Mantle is a less generic API. The gap will be small, 0 to 10 percent, and in some cases AMD believes that DirectX 12 may outperform Mantle. But by and large they feel that they can do a better job of optimizing Mantle for their GPUs than Microsoft can do with DirectX 12 or the Khronos group can with OpenGL. That’s a reasonable assumption to make and it gives us a great insight into AMD budding desire to control its own future rather than wait for the rest of the industry to show them the way forward.

We also had a chance to ask AMD about their thoughts on the probability that Intel, Nvidia, or any of the other graphics vendors would announce support for Mantle when the API is publicly released as an open standard near the end of this year. Largely AMD is unsure of how the other vendors will react although they thought that Nvidia and Intel picking up support for Mantle wasn’t entirely outside of the realm of possibility. They also reminded us that Mantle has been designed from the get go to support any modern DirectX 11 class GPU architecture, with the exception of Tile-based architectures like PowerVR’s and that both companies had expressed interest in at least looking at the API.

If you’re looking for an example of the new culture and programs that AMD’s turnaround has fostered inside the company then look no further than AMD’s efforts to promote and support its own API. Mantle is here to stay and it’s a long-term play by AMD to consistently steer and prod the graphics industry forward.S|A
Click to expand...

http://semiaccurate.com/2014/09/15/amds-mantle-api-going-outlive-directx-12/
 
Good find Mike.

In regards to new gpu support:

AMD has plans to introduce additional revisions to the Mantle API as it releases new GPU architectures so that developers can take advantage of the most advanced capabilities of AMD hardware.
Click to expand...

Might be an indication on Tonga needing a newer version of Mantle.
 
To be honest, I'd much rather both of them co-exist, as it will force both AMD and MS stay competitive and keep on working on improving the standards instead of slacking off. Regardless of what some people say to justify what MS has done (or not done when they could have had done) for the dx for the pass years, the fact was that they were like Creative who monopolised the PC audio market back then, so there was little incentive for them to push things forward. But now with Mantle around, hopefully things will start to change for the better for end-users.

Anyway, standing by 4...3...2...1...

...let the usual bashing begins :rolleyes:
 
Last edited:
Oh dear.

A few observations.

The biggest issue is legacy support, oh you mean like supporting your own previous cards, they moan at Microsoft for plans to not support something from two generations before when they cannot even support their own previous generation of cards. very hypocritical.

New revisions of mantle to support newer hardware, ok fair enough. But where is it, the new hardware is here the R9 285, I bet we wont be seeing a new version of mantle any time soon. If industry is to move forward as AMD want it too, then they had better pull their finger out because making people wait a few months isn't moving the industry forward it is holding it back.

Better performance because mantle is less generic, ok, but doesn't that completely contradict the above point. If mantle is going to need new revisions for every new card than it is going to have to become very generic for AMD hardware or else it just wont work at all.

Designed to work with any modern directX 11 hardware, well except for their own 5000 and 6000 series which both support directX 11, let alone other companies hardware.


Now I know my observations are not necessarily what they were meaning but I do find the whole thing just a bit of a joke.
 
I gave up after the second paragraph...

"DX12 will die because future revisions will not be supported on legacy OSes"

"Mantle fills this gap..."

It fills a gap only available on a specific set of AMD GPUs....and he has the audacity to claim EOL Legacy operating systems are a barrier for DX12. Not to mention it will only fill this very small gap, on a small set of AMD GPUs on games that ACTUALLY implement Mantle. Hey, what about the fact MOST games will have a DirectX render path and that, shock horror, it will more than likely have a fallback to DX11 also.

AMD sure seems to know a lot about which OSes will get DX12 support.

This guy blows nothing but hot air.
 
Last edited:
Ayahuasca said:
It's quite clear anyone working in PR gets the job by being a good BS artist.
Click to expand...

this is so true,

lets look at mantle for eg.


utter dross in reality. done nothing much and we have loads of none titles to show you how it performs :p

it was all a sales pitch for basically just a upgrade of a driver.

funny thing is they brought into loads of the worst recent engine in times frostbite to show how good mantle is :D

i could have told them how bad frostbite was before they did this for a small price instead of investing millions in it :p

the cut and short of it is dx 12 will be bigger and better than mantle.


if the amd people on here actually want to know what games to use ask someone who actually knows lol.
 
Martini1991 said:
Didn't Mantle just show it had a problem when it comes to newer architecture's? And DX12's the one that's got problems.........
Click to expand...
Yea it has a problem, but got blown out of proportion with some people more interested in bashing Mantle and looking at the situation/issue subjectively.

My interpretation of the problem is that Mantle could potentially perform worse than dx in games when running games that are older than the cards due to the reality of developers won't be updating/patching for a title that's over a 6 months to 1 year old (not blaming the developers or anything, just pointing out the reality of how things is).

So to put things into perspective, when it means older Mantle cards like 290/290x won't have "Mantle performance issue" with Mantle titles that are release after it. When 390x is eventually launched, it would probably have the same problem as the 285, and its Mantle performance in games that pre-date it would would probably be worse than running with dx11 (due to games would no longer get updated). But does it really matter when 390x's dx performance will still be faster than 290x's Mantle performance in the old Mantle titles (due to being a faster new gen card), and Mantle performance will be better than dx performance in all Mantle titles released after the launch of the 390x?

The foundmental problem with the 285 is that it is not really faster than the card it is replacing.
 
Last edited:
Mantle kick-started a major new trend in graphics APIs: reduced abstraction. API’s that offer lower level access to the hardware have been created by many important groups like Microsoft, Apple, and Khronos in the wake of Mantle’s debut. Thus AMD sees Mantle as a worthwhile endeavor for allowing them to more rapidly define the pace of innovation in the graphics industry.
Click to expand...

AMD believe that in a majority of cases Mantle will provide better performance than DirectX 12 because Mantle is a less generic API. The gap will be small, 0 to 10 percent, and in some cases AMD believes that DirectX 12 may outperform Mantle. But by and large they feel that they can do a better job of optimizing Mantle for their GPUs than Microsoft can do with DirectX 12 or the Khronos group can with OpenGL. That’s a reasonable assumption to make and it gives us a great insight into AMD budding desire to control its own future rather than wait for the rest of the industry to show them the way forward.
Click to expand...

Good, The last thing i want to see is AMD taking a "mission accomplished approach" to Mantle.

I'm of the firm belief that we would still be stuck with DX11 if it wasn't for Mantle.
Mantle needs to stick around, it needs to evolve, it needs to inovate. Microsoft should never get the impression that they have put AMD back in their box.
 
Last edited:
Only cross platform compatibility will keep mantle relevant.

At the moment, mantle is still beta, doesnt work very well and supported by a very small amount of titles.
 
Titles older than the card- use DX
Titles newer than the card- use Mantle

It is that simple. Realistically...how many people would care by the time of 390x released in say...6-9 months time that its Mantle performance (fps) in BF4/Thief is no better than 290x Mantle performance, when its dx performance will be faster than both?

This Mantle problem in reality terms is probably about as real as Nvidia's "dx12 advantage over AMD" due to more cards would support it, while realistically by the time of dx12's release, most people, if not everyone would have already long moved on from the GTX400/500 series.
 
Last edited:
humbug said:
Good, The last thing i want to see is AMD taking a "mission accomplished approach" to Mantle.
Click to expand...


Totally agree, as much as I don't like the way AMD are going about things the last thing we want is for them to say ok its done we are moving on to something else now.
 
Marine-RX179 said:
Yea it has a problem, but got blown out of proportion with some people more interested in bashing Mantle and looking at the situation/issue subjectively.

My interpretation of the problem is that Mantle could potentially perform worse than dx in games when running games that are older than the cards due to the reality of developers won't be updating/patching for a title that's over a 6 months to 1 year old (not blaming the developers or anything, just pointing out the reality of how things is).

So to put things into perspective, when it means older Mantle cards like 290/290x won't have "Mantle performance issue" with Mantle titles that are release after it. When 390x is eventually launched, it would probably have the same problem as the 285, and its Mantle performance in games that pre-date it would would probably be worse than running with dx11 (due to games would no longer get updated). But does it really matter when 390x's dx performance will still be faster than 290x's Mantle performance in the old Mantle titles (due to being a faster new gen card), and Mantle performance will be better than dx performance in all Mantle titles released after the launch of the 390x?

The foundmental problem with the 285 is that it is not really faster than the card it is replacing.
Click to expand...

Isn't it the current mantle version implementation that needs changing? So anything with current Mantle in development would still have an issue with Tonga.

But if Mantle's giving the best performance, and you've upgraded to a newer GPU to get better performance, how's it acceptable that you have to compromise with DirectX as Mantle's not performing? It's not really. Your stance is somewhat surprising.
 
Martini1991 said:
Isn't it the current mantle version implementation that needs changing? So anything with current Mantle in development would still have an issue with Tonga.

But if Mantle's giving the best performance, and you've upgraded to a newer GPU to get better performance, how's it acceptable that you have to compromise with DirectX as Mantle's not performing? It's not really. Your stance is somewhat surprising.
Click to expand...
I'm just being realistic here. It's unrealistic to expect developers to write games and engine in a way that it would be able to future-proof for all future Mantle GPU releases (thought if AMD from their could come up with a solution, then of course I'd be happy).

As much as "I would like" Mantle to be great even for old games, but between titles that's over 6-9 months old and all current/future releases, I'd rather have the Mantle performance issue lies with the old games that I probably won't have time to revisit than current and future titles.
 
Last edited:
I think it's unrealistic to expect developers to fix it months and months and months after launch for a new GPU. That doesn't detract from it being a flaw as much as you try to downplay it.

Then we're missing the other side. What happens to older GCN when they need to update Mantle? Do we then suffer? What will happen to an R9 290X's Mantle performance when they fix Tonga's?
 
You must log in or register to reply here.
Back
Top Bottom