Will AMD make a comeback against Intel once DX12 comes out and mainstream developers kill support for <4 cores

[TorqueS]

Read my response below.

 

Frames per second is completely irrelevant for what a low level API is designed to do. That's where a lot of the community members here on LTT seem to get it wrong. Mantle, DirectX 12, and even OpenGL 5 is not designed to boost frame rates in mind. They were designed to cut out API overhead. DirectX 11 has so many layers between the game engine and the device driver (for "security" reasons) that it's ridiculous. All of these layers waste CPU time and limits hardware potential. The reason you don't see drastic number changes in games today is because of the game itself. Games even such as Battlefield 4 were designed in constraints to the DirectX 11 API so that everyone has the same experience on the desktop platform. If they developed the game to run only on Mantle you would likely see far more content ingame. Unfortunately not everyone (Nvidia) is on board with Mantle and likely never will be due to DirectX 12 making its way to the market.

Yeah, that's roughly what I meant...DX11 and its previous iterations act like a barrier between the game and the GPU's processing power. This barrier is "maintained" through the CPU, so that's why CPU bottlenecking appears. DX12 and Mantle are meant to adress this issue and remove(partially) the barrier. Correct me if I'm wrong.

[Faa]

Mantle wasn't designed to boast performance numbers. Try drawing 10,000 completely separate objects ingame all with their own AI. Watch you get a whole whopping 5 FPS on SLI 980's.

Mantle was designed to boast their sales, main selling point being reducing CPU overhead and refusing to fix their current DX overhead just to bum about Mantle.

[zappian]

Keep in mind that while the pentium is being phased out in games , the i3 still runs games like far cry4 and DAI because its treated like a quad core.

 

The i3 is very good at runing those games , it runs far cry 4 better than a fx 8320.

[Majestic]

In short; no

long answer; noooooooooooooooooooooooooooooooooooooo

 

jk.

 

Low-overhead API's will scale just as well with Intel as they do with AMD. In fact, the increased IPC/synergy of Intel chips makes them scale better with Mantle/DX12.

It's just not showing right now, because they don't bottleneck GPU's as much, so the relative performance difference seems to be in favor of AMD/Low-End chips.

 

But relief some of the GPU bottlenecking and look what happens;

 

DirectX;

 

Mantle;

 

So no, they won't. They will always be subpar/outdated. Sorry. Also, don't expect the trend of killing dual-cores to persist beyond those 2 games that currently don't support it. I'm convinced some foul play is at work over there, it won't be a trend. Killing quad-core support will not happen, not in the next 5-10 years atleast..

[Opcode]

Mantle was designed to boast their sales, main selling point being reducing CPU overhead and refusing to fix their current DX overhead just to bum about Mantle.

In the case of improving their GPU and FX sales certainly. There's no denying that their DirectX ~11 drivers are far from optimized. Tho the need for an low level API had to be pushed sooner or later. The fact that AMD developed Mantle helped push Microsoft into collaborating with AMD to construct DirectX 12. AMD did indeed push it off as a way to improve FX and GPU performance numbers from a sales perspective. Tho it's bread as a whole new technology that even DirectX 12 won't be able to hold a candle to. I think we will see more of Mantle as DirectX 12 makes its debut. It's far easier to port a game from Mantle to DirectX 12 than it is from DirectX 11 to DirectX 12. Along with all next gen consoles supporting the API. AMD has a technology on their hands that could really push a threat for Nvidia regardless to how well Nvidia optimizes their drivers. As it may become standard for games to be written with Mantle in mind and then ported over to DirectX 12 later. Mantle is on its way to Linux as well in which if the development cycle works like that we will see a lot of games for the platform. Even tho AMD's desktop microprocessor sales aren't the greatest right now. They have set themselves up for success in the gaming market better than any other competitor.

[Majestic]

@Opcode I think you forget that the ATX builds are slowly dissapearing aswell. AMD is not particularly good at small-formfactor builds, both their GPU's and CPU's just use too much power. And developers can also read statistics, they won't force or start using a technology that >50% of their users can't use on launchday. That would be ludicrous.

[Faa]

AMD has a technology on their hands that could really push a threat for Nvidia regardless to how well Nvidia optimizes their drivers. 

You don't get it do you? Mantle's main purpose over Directx was to remove CPU bottlenecks which it failed in.

4670K at 4.4GHz with a 290x lightning on Directx omega driver;

This is with the same config as above but with Mantle omega driver;

My 3930K at 4.6GHz with 2 cores & Hyperthreading disabled and a GTX 970;

Mantle is dead and Nvidia is pushing AMD to get their directx drivers right which they even haven't managed to do in a year time. They spent their time on PR'ing Mantle rather than fixing their issues or they give a damn flying fuck to bum Mantle all the time which Richard Huddy seems to be doing in interviews a lot without realizing their competition done it with a single driver update for EVERY game out there unlike AMD who needed a new API and a year of laughable PR.

[Opcode]

You don't get it do you? Mantle's main purpose over Directx was to remove CPU bottlenecks which it failed in.

4670K at 4.4GHz with a 290x lightning on Directx omega driver;

-snip-

This is with the same config as above but with Mantle omega driver;

-snip-

My 3930K at 4.6GHz with 2 cores & Hyperthreading disabled and a GTX 970;

 

-snip-

Mantle is dead and Nvidia is pushing AMD to get their directx drivers right which they even haven't managed to do in a year time. They spent their time on PR'ing Mantle rather than fixing their issues or they give a damn flying fuck to bum Mantle all the time which Richard Huddy seems to be doing in interviews a lot without realizing their competition done it with a single driver update for EVERY game out there unlike AMD who needed a new API and a year of laughable PR.

Mantle was not developed to remove "bottlenecks" from the system. It was developed so us developers could have direct access to hardware without any side steps along the way (remove "bottlenecks" at a software level). You saying it failed at reducing CPU time only shows that you've never worked with Mantle nor DirectX for that matter. Mantle cuts CPU time in half compared to DirectX at rendering a single frame. Your comparison of frame rates based on driver optimizations in DirectX optimized games are all completely irrelevant. I have stated this in one of my previous posts if you cared to pay any attention. To put it in layman's terms if a game came out specifically built around the capabilities of Mantle and then got ported to DirectX 11. Nvidia's drivers (and thus hardware) would start looking like massive piles of shit in comparison. The only reason why Nvidia is just as good in comparison to Mantle is because the game itself is optimized specifically for that API. DirectX can only handle 3-5k batches at one time. Mantle can currently handle 10k batches with the goals of extending that to 100k batches. With games like Battlefield 4 you basically got the game pushing a similar number of batches through Mantle as well. The performance gains are a sheer bonus of the low level API. So maybe you're starting to get the idea that Mantle isn't about frame rates. The frame rate bonus is only due to less layers between the engine and the device driver. Along with 50% reduced CPU time at rendering a frame. There's more to Mantle than meets the eye and more to it than you'll understand. So if you want to reply to my post with "omgerd Nvidia FPS is just as good as with Mantle" then you're only proving how extensive your knowledge is in this department. Mantle wasn't developed just for FPS gains.

[Legitsu]

Mantle was not developed to remove "bottlenecks" from the system. It was developed so us developers could have direct access to hardware without any side steps along the way (remove "bottlenecks" at a software level). You saying it failed at reducing CPU time only shows that you've never worked with Mantle nor DirectX for that matter. Mantle cuts CPU time in half compared to DirectX at rendering a single frame. Your comparison of frame rates based on driver optimizations in DirectX optimized games are all completely irrelevant. I have stated this in one of my previous posts if you cared to pay any attention. To put it in layman's terms if a game came out specifically built around the capabilities of Mantle and then got ported to DirectX 11. Nvidia's drivers (and thus hardware) would start looking like massive piles of shit in comparison. The only reason why Nvidia is just as good in comparison to Mantle is because the game itself is optimized specifically for that API. DirectX can only handle 3-5k batches at one time. Mantle can currently handle 10k batches with the goals of extending that to 100k batches. With games like Battlefield 4 you basically got the game pushing a similar number of batches through Mantle as well. The performance gains are a sheer bonus of the low level API. So maybe you're starting to get the idea that Mantle isn't about frame rates. The frame rate bonus is only due to less layers between the engine and the device driver. Along with 50% reduced CPU time at rendering a frame. There's more to Mantle than meets the eye and more to it than you'll understand. So if you want to reply to my post with "omgerd Nvidia FPS is just as good as with Mantle" then you're only proving how extensive your knowledge is in this department. Mantle wasn't developed just for FPS gains.

well if developers would stop being lazy and trying to cut corners they draw call limit would not be issue

mantle is not really that impressive I have looked over the API and whilst it has some advantages its nothing fantastic ... 

whilst I would love to see some API other then DirectX take the lead I don't see that API being mantle 

[Opcode]

well if developers would stop being lazy and trying to cut corners they draw call limit would not be issue

mantle is not really that impressive I have looked over the API and whilst it has some advantages its nothing fantastic ... 

whilst I would love to see some API other then DirectX take the lead I don't see that API being mantle 

Draw call limit is an issue on Microsoft's behalf (DirectX). This is exactly the reason why Mantle came into being. There's so much stuff that is cut out of games because developers are limited to a restricting API. Tho they continue to develop for it because it's the most widely used API. Some companies like DICE and Crytek adopted the use of Mantle for some performance benefit gains. Tho we won't see it's true colors until DirectX 12 games make their way to the market when API restrictions are furthermore lifted.

[Legitsu]

Draw call limit is an issue on Microsoft's behalf (DirectX). This is exactly the reason why Mantle came into being. There's so much stuff that is cut out of games because developers are limited to a restricting API. Tho they continue to develop for it because it's the most widely used API. Some companies like DICE and Crytek adopted the use of Mantle for some performance benefit gains. Tho we won't see it's true colors until DirectX 12 games make their way to the market when API restrictions are furthermore lifted.

for those who have been around long enough that statement is false a lot of it(not all mind you but a fair chunk) is developers being lazy or working on impossible timetables(looking at you EA) 

and not spending enough time writing proper occlusion culling/optimization into the engine .. 

speaking of crytech you notice that the Cry Engine  doesn't have a issue with drawcalls and still manages to look better and run better then say .. assins crap unity or battlefold 4 despite it having higher quality assets and competing "map sizes" hell even the unreal 3 engine doesn't do terrible when properly handled and its a pig 

[Opcode]

for those who have been around long enough that statement is false a lot of it(not all mind you but a fair chunk) is developers being lazy or working on impossible timetables(looking at you EA) 

and not spending enough time writing proper occlusion culling/optimization into the engine .. 

speaking of crytech you notice that the Cry Engine  doesn't have a issue with drawcalls and still manages to look better and run better then say .. assins crap unity or battlefold 4 despite it having higher quality assets and competing "map sizes" hell even the unreal 3 engine doesn't do terrible when properly handled and its a pig 

Time tables should be put aside just look how many games even bothered with PhysX back in the day. Instead of optimizing a game for the best performance to visual quality ratio around DirectX they should of just implemented what they wanted from the start and go from there. With Mantle there's so much freed up CPU time for doing other stuff (more content, more complex AI, etc). Instead developers are forced to optimize their games with what DirectX offers as it's the most utilized API for games. I would rather see games developed for a low level API where more trees, buildings, and other content that can be implemented in game without sacrificing frame rates below what DirectX currently offers. A prime example is Faa's AMD numbers posted a few posts back. There's a lot of room for handling other stuff before your frame rates fall below what is standard for their DirectX driver. We don't know how well Nvidia's cards would scale with Mantle as they refuse to use it (even tho they could, AMD offered it). We'll see more on this scaling when DirectX 12 makes it debut as you should still be able to force certain games back to using a previous API version. Nvidia on DirectX 11 vs DirectX 12 would be somewhat of a direct comparison to Mantle. Games like Minecraft are literally perfect for Mantle scaling. Instead developers have to limit your FOV and other aspects of the game because it would turn into a slide show otherwise with that many more batches. My point is there's a lot of improvements that can be made in games with a low level API other than frame rate gains.

[Legitsu]

Time tables should be put aside just look how many games even bothered with PhysX back in the day. Instead of optimizing a game for the best performance to visual quality ratio around DirectX they should of just implemented what they wanted from the start and go from there. With Mantle there's so much freed up CPU time for doing other stuff (more content, more complex AI, etc). Instead developers are forced to optimize their games with what DirectX offers as it's the most utilized API for games. I would rather see games developed for a low level API where more trees, buildings, and other content that can be implemented in game without sacrificing frame rates below what DirectX currently offers. A prime example is Faa's AMD numbers posted a few posts back. There's a lot of room for handling other stuff before your frame rates fall below what is standard for their DirectX driver. We don't know how well Nvidia's cards would scale with Mantle as they refuse to use it (even tho they could, AMD offered it). We'll see more on this scaling when DirectX 12 makes it debut as you should still be able to force certain games back to using a previous API version. Nvidia on DirectX 11 vs DirectX 12 would be somewhat of a direct comparison to Mantle. Games like Minecraft are literally perfect for Mantle scaling. Instead developers have to limit your FOV and other aspects of the game because it would turn into a slide show otherwise with that many more batches. My point is there's a lot of improvements that can be made in games with a low level API other than frame rate gains.

again I would love to see a low level API that offers competing functionality to DirectX/OpenGL

but mantle is not that API nor will it ever be for a variety of both technical and business reasons 

if a lower level API is going to happen then it needs to be the following

1.not tied nor created by any one vendor 

2. fully open standard including complete source code and full technical schismatics available to ANYBODY

3. its feature set must rival or exceed the competing standards such as opengl,directx,mantle ect ect

4. ability to be implemented into any architecture 

[Faa]

Mantle was not developed to remove "bottlenecks" from the system. It was developed so us developers could have direct access to hardware without any side steps along the way (remove "bottlenecks" at a software level). 

How exactly is this useful if you still have to write it for Directx anyways? Mantle wastes more time than it does anything good, it's even wasting our time here arguing about a dead API. Also I never said Mantle was made to lift CPU bottlenecks, it's main purpose was dealing with CPU bottlenecks. Without that purpose Mantle is pretty useless for the consumers which currently is as it doesn't do any better than Directx.

 

 

Mantle cuts CPU time in half compared to DirectX at rendering a single frame. Your comparison of frame rates based on driver optimizations in DirectX optimized games are all completely irrelevant. 

And? Reducing your API overhead can give you the same results. Example; http://docs.nvidia.com/gameworks/content/gameworkslibrary/graphicssamples/opengl_samples/multidrawindirectsample.htm

 

 

Your comparison of frame rates based on driver optimizations in DirectX optimized games are all completely irrelevant. 

 

Because it made Mantle irrelevant?

 

Mantle cuts CPU time in half compared to DirectX at rendering a single frame. 

Sounds PR.

 

 

Along with 50% reduced CPU time at rendering a frame. 

 

Now we have a number for our PR. Also that number which you probably made up was compared against their OWN Directx which is completely irrelevant.

 

To put it in layman's terms if a game came out specifically built around the capabilities of Mantle and then got ported to DirectX 11. Nvidia's drivers (and thus hardware) would start looking like massive piles of shit in comparison. 

 

That's a claim. Not really worth mentioning this though when the API didn't even survive a month or two or the fact that this will never happen.

 

 

Mantle wasn't developed just for FPS gains.

Mantle is nothing more or less than a vibrator on a razor. Customers who are gamers obviously aren't interested to use a tool thats only making developers life easier.

 

So if you want to reply to my post with "omgerd Nvidia FPS is just as good as with Mantle" then you're only proving how extensive your knowledge is in this department. 

Let's just look at what you managed to C/P, apparently all PR.

 

The fact that AMD developed Mantle helped push Microsoft into collaborating with AMD to construct DirectX 12. 

Mantle had nothing to do with Directx12, just nothing. Sure Mantle pushed Microsoft to come with a presentation, that's all about it. Directx12 is under 4 years of development (march 2014) when Microsoft already presented some features of Directx12 which was using a Titan black, Mantle came out around January, do you really think AMD is going to push anything of Mantle in Directx12? 

According to Nvidia; NVIDIA claims that work on DX12 with Microsoft “began more than four years ago with discussions about reducing resource overhead. For the past year, NVIDIA has been working closely with the DirectX team to deliver a working design and implementation of DX12 at GDC.” This would indicate that while general ideas about what would be in the next version of DX, the specific timeline to build and prepare it started last spring.

The CPU overhead improvements had nothing to do with Mantle.

According to PCPER; NVIDIA is currently the only GPU vendor to have a DX12 capable driver in the hands of developers and the demo that Microsoft showed at GDC was running on a GeForce GTX TITAN BLACK card. (UPDATE: I was told that actually Intel has a DX12 capable driver available as well leaving AMD as the only major vendor without.)

They didn't even have a driver ready yet for DX12 and you're claiming AMD helped Microsoft to construct Directx12. Also I would like to see you having some evidence of AMD helping Microsoft out constructing Directx12 and as far as I'm aware Richard Huddy was against DirectX.

 

 

 

It's far easier to port a game from Mantle to DirectX 12 than it is from DirectX 11 to DirectX 12. 

But you said earlier Nvidia would perform shit if the game was made firstly for Mantle and later on being ported to Directx. Also I've read this somewhere here; http://en.wikipedia.org/wiki/Mantle_%28API%29#Other_claims

Also a claim.

 

[Opcode]

again I would love to see a low level API that offers competing functionality to DirectX/OpenGL

but mantle is not that API nor will it ever be for a variety of both technical and business reasons 

if a lower level API is going to happen then it needs to be the following

1.not tied nor created by any one vendor 

2. fully open standard including complete source code and full technical schismatics available to ANYBODY

3. its feature set must rival or exceed the competing standards such as opengl,directx,mantle ect ect

4. ability to be implemented into any architecture 

That's the thing is Mantle offers competitive functionality to DirectX.

1. I would agree to an API being open source and capable of being contributed to by any source (including you and me).
2. As stated above an open standard would be great tho I guess it would face issues regarding commits that target a specific vendors hardware (competitive optimization).
3. I don't think we need a more advanced API just yet as Mantle, DirectX and OpenGL all offer comparative features.
4. DirectX, OpenGL, and Mantle support all architectures as well. Mantle can be used on Nvidia and Intel graphics architecture tho both companies refuse.

Tho AMD took it upon themselves to develop an API to bring PC gaming back to a console level. To be quite frank if Mantle never made its appearance I have questionable doubt in my mind that DirectX and OpenGL would of ever made the change to a low level API. It seems that Mantle has influenced this trend and it's beneficial for both us the developers and you the consumers.

 

How exactly is this useful if you still have to write it for Directx anyways? Mantle wastes more time than it does anything good, it's even wasting our time here arguing about a dead API. Also I never said Mantle was made to lift CPU bottlenecks, it's main purpose was dealing with CPU bottlenecks. Without that purpose Mantle is pretty useless for the consumers which currently is as it doesn't do any better than Directx.

Because DirectX 12 and Mantle have very similar functionality. As said in my previous post that Mantle -> DirectX 12 and vise versa will be far easier than DirectX 11 -> DirectX 12. With that being said I honestly have doubts that DirectX 12 titles will retain previous iterations of DirectX support below version 11.2 as Microsoft already stated they will bring DirectX 12 feature set to DirectX 11.2 minus the low level support.

 

And? Reducing your API overhead can give you the same results. Example; http://docs.nvidia.com/gameworks/content/gameworkslibrary/graphicssamples/opengl_samples/multidrawindirectsample.htm

That's working with an extended version of OpenGL. Quite similar to Mantle being the backport of several game engines. It's not something that will be widely adopted as it doesn't come as standard core functionality. The most humors part of this particular quote is that OpenGL extension was created by an AMD developer.

 

Because it made Mantle irrelevant?

It doesn't make Mantle irrelevant as I have already stated frame rates are irrelevant. Continuing on with frame rates as being the only important aspect of game design is just idiocy at its finest. High frame rates don't matter when the game looks like CS 1.6. Polygon count is critical for higher tessellation.

 

Sounds PR.

Have you even bothered writing some example code for testing both Mantle and DirectX 11.

 

Now we have a number for our PR. Also that number which you probably made up was compared against their OWN Directx which is completely irrelevant.

The percentage can even exceed that as there are no driver threads. CPU overhead is cut down by over 60% tho unlike DirectX Mantle creates very small GPU overhead. Still combined the time it takes to render a single frame is roughly 50% shorter. With full Mantle support games can spit out up to twice as many frames as DirectX with the current iteration.

 

That's a claim. Not really worth mentioning this though when the API didn't even survive a month or two or the fact that this will never happen.

It's not a claim Nvidia is bound by DirectX limitations and they have no way out of that other than adopting Mantle as they should. I don't know if you're pro-Nvidia or what but Mantle is actually very widely adopted. There are over 22 games in development as we sit here and talk about it that leverages Mantle. It's not going away any time soon. The hype for Mantle has slowed down as it's no longer a new subject. Tho that doesn't mean that the project is dead.

 

Mantle is nothing more or less than a vibrator on a razor. Customers who are gamers obviously aren't interested to use a tool thats only making developers life easier.

Mantle isn't an API designed for making writing games easier. In fact it's not any easier than working with DirectX as you have direct hardware access. This is why a majority of developers baked Mantle directly into their game engines.

 

Let's just look at what you managed to C/P, apparently all PR.

As just a casual gamer with no game programming knowledge your just left with biased opinions. I don't C/P anything, either I know what I'm talking about or I don't even bother conforming a reply. It saves the hassle of looking like a fool later on.

 

Mantle had nothing to do with Directx12, just nothing. Sure Mantle pushed Microsoft to come with a presentation, that's all about it. Directx12 is under 4 years of development (march 2014) when Microsoft already presented some features of Directx12 which was using a Titan black, Mantle came out around January, do you really think AMD is going to push anything of Mantle in Directx12? 

According to Nvidia; NVIDIA claims that work on DX12 with Microsoft “began more than four years ago with discussions about reducing resource overhead. For the past year, NVIDIA has been working closely with the DirectX team to deliver a working design and implementation of DX12 at GDC.” This would indicate that while general ideas about what would be in the next version of DX, the specific timeline to build and prepare it started last spring.

The CPU overhead improvements had nothing to do with Mantle.

According to PCPER; NVIDIA is currently the only GPU vendor to have a DX12 capable driver in the hands of developers and the demo that Microsoft showed at GDC was running on a GeForce GTX TITAN BLACK card. (UPDATE: I was told that actually Intel has a DX12 capable driver available as well leaving AMD as the only major vendor without.)

They didn't even have a driver ready yet for DX12 and you're claiming AMD helped Microsoft to construct Directx12. Also I would like to see you having some evidence of AMD helping Microsoft out constructing Directx12 and as far as I'm aware Richard Huddy was against DirectX.

From my sources (not Google) Microsoft has reached out to AMD to help them with DirectX 12. One of the reasons as to why DirectX has been high level for so long is that Microsoft has no clue how to secure it. Reaching out to AMD gave them all the resources that they needed to bring DirectX 12 back down (DirectX was low level before) to a low level API. To add weight to that Microsoft is not a hardware vendor. Their GPU drivers alone is nothing more than a joke.

 

If you read that statement it's just Nvidia claiming that they've worked with Microsoft to ensure their devices and drivers are compliant. Just like AMD's own statement below.

“AMD strongly believes in the benefits gamers and game developers can realize from lower-overhead API development,” said Matt Skynner, corporate vice president and general manager, Graphics Business Unit, AMD. “With the Mantle API, AMD has shown the world our commitment to incredible performance, and we look forward to enabling the same performance gains by supporting the industry-standard DirectX® 12.”

 

Nvidia has no footprint in low level API's otherwise they would have designed one of their own. AMD has DirectX 12 compliant drivers as they were shown off a EMGP.

 

I would suggest you stick to what you know instead of relying on Google'd sources (not all sources are credible).

 

But you said earlier Nvidia would perform shit if the game was made firstly for Mantle and later on being ported to Directx. Also I've read this somewhere here; http://en.wikipedia.org/wiki/Mantle_%28API%29#Other_claims

Also a claim.

It's not a claim Nvidia has good driver optimization for DirectX 11 but if a game was built around a low level API in mind and then was ported to DirectX 11. You do the math, the game would have to be stripped of content or either take the performance hit. DirectX is only an hourglass.

[Faa]

From my sources (not Google) Microsoft has reached out to AMD to help them with DirectX 12. One of the reasons as to why DirectX has been high level for so long is that Microsoft has no clue how to secure it. Reaching out to AMD gave them all the resources that they needed to bring DirectX 12 back down (DirectX was low level before) to a low level API. To add weight to that Microsoft is not a hardware vendor. Their GPU drivers alone is nothing more than a joke.

That's just pure bullshit. AMD, Intel, Nvidia were ALL involved in DX12 development.

"The day after the D3D12 keynote, I got on the phone with Tony Tamasi, Nvidia's Senior VP of Content and Technology. Tamasi painted a rather different picture than Corpus. He told me D3D12 had been in in the works for "more than three years" (longer than Mantle) and that "everyone" had been involved in its development. As he pointed out, people from AMD, Nvidia, Intel, and even Qualcomm stood on stage at the D3D12 reveal keynote."

http://techreport.com/review/26239/a-closer-look-at-directx-12

3 years in development, Mantle comes out and DX12 gets presented and you think they managed to get the lower CPU overhead from Mantle (actually Mantle doesnt offer anything over DX as being proven) in 2-3 months time ready? Give me a good reason why we should take your word over a guy who's working in that field? You know what's logical? AMD stole the DX12 ideas and went for her own glory. Why am I saying it? Richard huddy said in 2011 so 4 years ago he wanted the API go away.

'I certainly hear this in my conversations with games developers,' he says, 'and I guess it was actually the primary appeal of Larrabee to developers – not the hardware, which was hot and slow and unimpressive, but the software – being able to have total control over the machine, which is what the very best games developers want. By giving you access to the hardware at the very low level, you give games developers a chance to innovate, and that's going to put pressure on Microsoft – no doubt at all.' 

http://www.bit-tech.net/hardware/graphics/2011/03/16/farewell-to-directx/1

He admitted a low-level API would put pressure on Microsoft, he made one and failed and now you're trying to claim DX12 is the new Mantle orsomething? AMD is refusing to fix their DirectX overhead issues simply because they want to advertise the shit out of Mantle. A year ago it was a known fact after 337.50 came out that AMD's driver overhead were extremely terrible and we still see the same difference in the latest and even in the Mantle titles.

 

I would suggest you stick to what you know instead of relying on Google'd sources (not all sources are credible).

I've proven you wrong.

 

 

It's not a claim Nvidia has good driver optimization for DirectX 11 but if a game was built around a low level API in mind and then was ported to DirectX 11. You do the math, the game would have to be stripped of content or either take the performance hit. DirectX is only an hourglass.

Prove it or it remains a claim. You can't really prove something that never happened or is going to happen or even use yourself as a source.

 

 

That's working with an extended version of OpenGL. Quite similar to Mantle being the backport of several game engines. It's not something that will be widely adopted as it doesn't come as standard core functionality. The most humors part of this particular quote is that OpenGL extension was created by an AMD developer.

That's covering the overhead drivers cause on the CPU and how you can mitigate it - that doesn't equal an API. Seems like you haven't read anything of it while it's stating what I just said literally there.

The main goal of MultiDrawIndirect and other AZDO features is to reduce driver overhead; while they may not always increase frame rate directly, they can generally reduce the CPU load, providing more CPU "headroom" for the application itself.

 

 

The percentage can even exceed that as there are no driver threads. CPU overhead is cut down by over 60% tho unlike DirectX Mantle creates very small GPU overhead. Still combined the time it takes to render a single frame is roughly 50% shorter. With full Mantle support games can spit out up to twice as many frames as DirectX with the current iteration.

You mean Mantle cuts CPU overhead by 60% over their own wonky Directx drivers?

 

 

Still combined the time it takes to render a single frame is roughly 50% shorter. 

 

Against their own Directx I guess.

 

 

With full Mantle support games can spit out up to twice as many frames as DirectX with the current iteration.

 

One more claim.

 

 

It doesn't make Mantle irrelevant as I have already stated frame rates are irrelevant. Continuing on with frame rates as being the only important aspect of game design is just idiocy at its finest. High frame rates don't matter when the game looks like CS 1.6. Polygon count is critical for higher tessellation.

We aren't discussing here development or game design, I had to prove your lies wrong, we're discussing here how useful Mantle is for the end-customer.

[CDHoward]

Currently, the only Intel processors available with 4 cores or more are their old Core 2 Quad series and anything with i5 and up. Celeron, Pentium, i3s, and even some Xeons will become obsolete once the DX12 comes out and AMD currently has more multi-core processors than Intel on the market, including the 4,6, and 8 core FX series, 4 core Athlon, and 4+ core APUs up for offer. So, it's easy to see that while Intel has been spending their resources on getting performance per core, AMD has just gone ahead and added more cores to their CPUs to improve performance. When DX12 comes out, core# and the multithreaded performance will be more important and performance per core not so much.

When DX12 comes out and more games pull support for dual core processors (it is already happening with FC4) in favour of quad cores and up, do you see AMD coming back and blowing Intel out of the water? Do you see them surpassing Intel?

I've been pondering the very same thing, mate.

DirectX12 will take advantage of multiple cores, and so it follows that the FX 8 cores will gain a huge performance increase.

The only reason Intel are better is because DirectX11 runs everything on one core (core 0). DirectX12 will completely change this.

[Godlygamer23]

I've been pondering the very same thing, mate.

DirectX12 will take advantage of multiple cores, and so it follows that the FX 8 cores will gain a huge performance increase.

The only reason Intel are better is because DirectX11 runs everything on one core (core 0). DirectX12 will completely change this.

DX11 is not single threaded. It's just inefficient.

https://www.google.com/url?q=http://msdn.microsoft.com/en-us/library/windows/desktop/ff476891(v=vs.85).aspx&sa=U&ei=EE-fVNu7IseigwTE9INo&ved=0CAsQFjAA&sig2=cveVjHwsP64ygI9qE0hsAg&usg=AFQjCNHNieszGG94YWBpXBxWGW5NYmKbkA

[Opcode]

I've been pondering the very same thing, mate.

DirectX12 will take advantage of multiple cores, and so it follows that the FX 8 cores will gain a huge performance increase.

The only reason Intel are better is because DirectX11 runs everything on one core (core 0). DirectX12 will completely change this.

DirectX 11 actually supports threading. Nvidia's latest drivers are believed to finally leverage this as that's the only logical explanation for such a performance gain. That or Mantle tech got out into the wild with the DirectX 12 collaboration and Nvidia is using some of it in their latest drivers. DirectX 11 creates a lot of API overhead on Core #0 so you end up with poor frame times among other things because they are reliant on the thread latency on that core. With DirectX 12 the API is spread out more evenly among the cores reducing overall frame time. It also helps cut out CPU time by removing a bunch of layers between the game engine and the device driver. Unfortunately DirectX 12 will not be bare down to the bone as Mantle is. So you could expect similar to less performance out of DirectX 12 in comparison. What you see right now with a FX-8350 and R9 290X with Mantle is what you'll typically see out of the same setup using DirectX 12. Tho you'd better off sticking with Mantle due to it being much closer to the hardware. DirectX 12 will be a big help tho for Nvida who is quite good at optimizing their DirectX drivers.

[CDHoward]

DirectX 11 actually supports threading. Nvidia's latest drivers are believed to finally leverage this as that's the only logical explanation for such a performance gain. That or Mantle tech got out into the wild with the DirectX 12 collaboration and Nvidia is using some of it in their latest drivers. DirectX 11 creates a lot of API overhead on Core #0 so you end up with poor frame times among other things because they are reliant on the thread latency on that core. With DirectX 12 the API is spread out more evenly among the cores reducing overall frame time. It also helps cut out CPU time by removing a bunch of layers between the game engine and the device driver. Unfortunately DirectX 12 will not be bare down to the bone as Mantle is. So you could expect similar to less performance out of DirectX 12 in comparison. What you see right now with a FX-8350 and R9 290X with Mantle is what you'll typically see out of the same setup using DirectX 12. Tho you'd better off sticking with Mantle due to it being much closer to the hardware. DirectX 12 will be a big help tho for Nvida who is quite good at optimizing their DirectX drivers.

Really? Well, I hope you're wrong, mate.

[Opcode]

Really? Well, I hope you're wrong, mate.

It is true, DirectX 12 won't bring any more performance to the table over Mantle. Which for existing GCN owners might seem like a drag. Tho these cards will be quite dated by the time DirectX 12 games make it to the market anyways. R9 300 series is coming out next year with some massive improvements. And I wouldn't expect DirectX 12 to become a "thing" until at least 2016. It usually takes a long while for a new API to catch on. By the time DirectX 12 games flood the market existing GCN 1.0/1.1 owners will be wanting to upgrade anyways.

[patrickjp93]

Time tables should be put aside just look how many games even bothered with PhysX back in the day. Instead of optimizing a game for the best performance to visual quality ratio around DirectX they should of just implemented what they wanted from the start and go from there. With Mantle there's so much freed up CPU time for doing other stuff (more content, more complex AI, etc). Instead developers are forced to optimize their games with what DirectX offers as it's the most utilized API for games. I would rather see games developed for a low level API where more trees, buildings, and other content that can be implemented in game without sacrificing frame rates below what DirectX currently offers. A prime example is Faa's AMD numbers posted a few posts back. There's a lot of room for handling other stuff before your frame rates fall below what is standard for their DirectX driver. We don't know how well Nvidia's cards would scale with Mantle as they refuse to use it (even tho they could, AMD offered it). We'll see more on this scaling when DirectX 12 makes it debut as you should still be able to force certain games back to using a previous API version. Nvidia on DirectX 11 vs DirectX 12 would be somewhat of a direct comparison to Mantle. Games like Minecraft are literally perfect for Mantle scaling. Instead developers have to limit your FOV and other aspects of the game because it would turn into a slide show otherwise with that many more batches. My point is there's a lot of improvements that can be made in games with a low level API other than frame rate gains.

No, there isn't. Look at Nvidia's DX 11 benches vs. AMD's Mantle! AMD just won't fix its DX 11 driver! The turth is Mantle fixed almost nothing that couldn't have been done in DX 11... Now, there are some good concepts, but the reality is the execution fell short.

[Opcode]

No, there isn't. Look at Nvidia's DX 11 benches vs. AMD's Mantle! AMD just won't fix its DX 11 driver! The turth is Mantle fixed almost nothing that couldn't have been done in DX 11... Now, there are some good concepts, but the reality is the execution fell short.

There's a lot of controversy about Nvidia stealing Mantle code from AMD through the collaboration on DirectX 12. That or they finally threaded their drivers to utilize DirectX 11 properly. Them are the two only ways that they could of obtained that much of a performance boost over their previous driver package. Hence why I wrote a reply to Faa but haven't submitted it until I know what's going on with Nvidia's current drivers.

[patrickjp93]

There's a lot of controversy about Nvidia stealing Mantle code from AMD through the collaboration on DirectX 12. That or they finally threaded their drivers to utilize DirectX 11 properly. Them are the two only ways that they could of obtained that much of a performance boost over their previous driver package. Hence why I wrote a reply to Faa but haven't submitted it until I know what's going on with Nvidia's current drivers.

WOW! Now that's some conspiracy theory... Can Nvidia do no right in your eyes? Yes Jen Hsun Huang is a greedy asshat but Nvidia makes great software and hardware. You really think Nvidia had to have stolen Mantle code? Nvidia doesn't hold back unless it can. Currently it can't.

[Opcode]

WOW! Now that's some conspiracy theory... Can Nvidia do no right in your eyes? Yes Jen Hsun Huang is a greedy asshat but Nvidia makes great software and hardware. You really think Nvidia had to have stolen Mantle code? Nvidia doesn't hold back unless it can. Currently it can't.

Indeed it is but you would never imagine what companies will do in order to obtain a competitive edge. I own Nvidia hardware so again drop the fanboyism. You're so pro-Intel and pro-Nvidia that it makes me even sound like an AMD fanboy.