Hilbert gave credit to WCCFT, but WCCFT like to claim articles as their own. I'd rather take performance gains across all DX11 CPU limited titles than a select few with Mantle baked in.

Who wouldn't but these graphs so nothing to that point. They are using cpu that really has no bottleneck involved in case like star swarm. Where these drivers will prove themselves is with weaker cpu's paired with gpus like 780 and 780 ti. We have to wait and see. But considering nvidia can do this, I would guess amd can do this too with dx11 while still being slower then mantle cause dx11 will have an overhead bigger then mantle it does not matter what nvidia or amd does in driver side it does not make the API magically lose all it's overhead.

Cant wait for these drivers. No wonder nvidia is so mum on mantle.

I hardly doubt it will be this huge of improvement over Mantle... Remember when Nvidia cheated with 3Dmark ect making you think it was better.

I noticed that is basically a 290x vs 780ti... Surely that performance difference is mostly the 780ti being the most powerful card on the market. Also what use is investing into DX11 right now? Surely just work with Microsoft on DX12?

I hardly doubt it will be this huge of improvement over Mantle... Remember when Nvidia cheated with 3Dmark ect making you think it was better.

Huge ?No!But it will make Mantle pointless,because all DX11 games will benefit from this new driver.If those drivers whould not be better the Mantle I`m pretty sure that Nvidia would not show those slides.

I hardly doubt it will be this huge of improvement over Mantle... Remember when Nvidia cheated with 3Dmark ect making you think it was better.

lol so did amd(ati)so its all the same crap out of both of these company's Like some have said they are keeping it close and thats it i think for a few reasons 1st and formost....how can they show the newer stuff when it gets here being alot faster?

I think that they had this ace on them but didn't had to take it out untill they need it.

Good to hear, even better to see the competition. A shame it's not the same in the CPU department, where Intel clearly dominates.

Buuuut, systems with slow CPUs won't see this improvement through a graphics driver update right? The tests were run on a 3930K, which isn't your average CPU. I predict that Mantle will still own on CPU bottlenecked systems.

thumbs up i love this-agreed......"Proves big point. Nvidia doesn't want to beat AMD by huge margin, but only to stay competitive. Mantle forced their hand, seem's they probably have been sitting on this kinda performance update until required":banana:

Buuuut, systems with slow CPUs won't see this improvement through a graphics driver update right? The tests were run on a 3930K, which isn't your average CPU. I predict that Mantle will still own on CPU bottlenecked systems.

Just because they tested the driver with a 3930K don`t mean that other CPU`s will not see a improvement.

The techreport Mantle article already showed that Nvidia's CPU driver overhead was significantly better than AMD's even prior to Mantle.

This is desperately needed! I find that DX11 games are still not performing as well as they ought to by this point. I can't think of one DX11 game (or game with DX11 features) that don't have microstuttering or some other notable performance inhibitors. I still can't play AC4 even on medium setting more or less, though that might just be the game engine.

I don't want to ruin everyone's fun, but... DirectX libraries are owned, compiled and distributed by M$. If there is high CPU overhead in library like d3dx11_43.dll, nVidia/AMD has no way to reduce such overhead by other means than compiling their own alternative with same imports/exports and replacing such library. Therefore if nVidia reduced any kind of overhead for DX11 procedures they reduced driver level overhead and not number of CPU cycles required by d3dx11_43.dll and such. If they are overriding code in default M$ libraries, we are getting back nearly 20 years when stuff behaved and looked differently on ATi/nVidia systems. At those times performance was treasured way over standardization. And I do not think they are aiming for this. But on other hand if they reduced this greatly Driver level overhead and gained such performance boost, imagine those thin drivers on Mantle or in 2 years on DX12. I do not want to over-cheer here. As in lowest they could have taken that heavy motion blur effect in star swarm, made detection for it and used some other way to reproduce same/similar effect with lower CPU overhead. And they could have do same for Thief. Therefore I would like to see those heavy effects in screenshots and video comparison. Because everyone knows here how PR works.

I don't want to ruin everyone's fun, but... DirectX libraries are owned, compiled and distributed by M$. If there is high CPU overhead in library like d3dx11_43.dll, nVidia/AMD has no way to reduce such overhead by other means than compiling their own alternative with same imports/exports and replacing such library. Therefore if nVidia reduced any kind of overhead for DX11 procedures they reduced driver level overhead and not number of CPU cycles required by d3dx11_43.dll and such. If they are overriding code in default M$ libraries, we are getting back nearly 20 years when stuff behaved and looked differently on ATi/nVidia systems. At those times performance was treasured way over standardization. And I do not think they are aiming for this. But on other hand if they reduced this greatly Driver level overhead and gained such performance boost, imagine those thin drivers on Mantle or in 2 years on DX12. I do not want to over-cheer here. As in lowest they could have taken that heavy motion blur effect in star swarm, made detection for it and used some other way to reproduce same/similar effect with lower CPU overhead. And they could have do same for Thief. Therefore I would like to see those heavy effects in screenshots and video comparison. Because everyone knows here how PR works.

This is where you are wrong. http://www.pcper.com/reviews/Graphics-Cards/NVIDIA-Talks-DX12-DX11-Efficiency-Improvements

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 shows Nvidia has been given source code access. Personally I think that not only will we get a new driver but we will also get a DirectX 11 update.

So looking forward to this if it's true. I love "free" upgrades.

Me too. Sure I am cpu bottlenecked somewhat with my powerful sli with a 3 year old sandy bridge even at4.7ghz. Cant wait for the driver and all dx11 games out of the box is a big plus

Pmsl 😀 😀 😀

I doubt that old sandy bridge are bottle-necked. From first generation i7 nothing has drastically changed ( darstically like from c2q to i7 ). This new drivers will change little. Nvidia Fans will stay like that and Ati Fan they will say mantle is still better. :) new story, old story. At least the driver update has not to be developer supported, is something you get for free everywhere ( maybe i'll get even pcsx2 performance improvementes 🤓 ) And that's why they annonunced that old products are going to be erased from new drivers in short time!!!

This is where you are wrong. http://www.pcper.com/reviews/Graphics-Cards/NVIDIA-Talks-DX12-DX11-Efficiency-Improvements This shows Nvidia has been given source code access. Personally I think that not only will we get a new driver but we will also get a DirectX 11 update.

Article is known to me. But look at it this way: If they improved DirectX11 itself and M$ will release new DX11 libraries, it will give boost to AMD systems too and therefore nVidia showing 290x in graphs should get boost too. There is no such boost visible therefore I presume This Boost is not on DX11 libraries side or is very misrepresented by nV PR team. Other possibility is to include into libraries alternative code path for nV systems where special procedures would be called (like that GPU will offload part of math done by CPU, or GPU will handle some stuff completely, but that is possible only for limited instruction set). Such scenario is again very unlikely as it would mean M$ giving "unfair" advantage to nV against AMD and such case would likely end up in court with result that nV would have to pay to AMD part of money they make from such boost. (unless AMD got same offer/opportunity)