guru3dDoom Vulkan Benchmarks

[Shivansps]

Im just trying the Doom Demo on a 750TI, it losses about 10fps by going to Vulkan... nice work ID Tech, what a good patch to improve old card performance, this is how is done.

 

[sontin]

Im just trying the Doom Demo on a 750TI, it losses about 10fps by going to Vulkan... nice work ID Tech, what a good patch to improve old card performance, this is how is done.

Normal for nVidia cards. When the scene is GPU limited you can lose up to 20% of performance: https://youtu.be/d16FlJxp-sw?t=19

It is just a broken Vulkan path. :thumbsdown:

 

[Arachnotronic]

Normal for nVidia cards. When the scene is GPU limited you can lose up to 20% of performance: https://youtu.be/d16FlJxp-sw?t=19

It is just a broken Vulkan path. :thumbsdown:

What do you expect when the lead rendering programmer is out retweeting "#BetterRed"?

Fortunately my 1080 destroys this game in OpenGL and my upcoming Titan X will do an even better job of it, so whatevs.

 

[MajinCry]

Im just trying the Doom Demo on a 750TI, it losses about 10fps by going to Vulkan... nice work ID Tech, what a good patch to improve old card performance, this is how is done.

D'err. Vulkan's fer old CPU performance. And it's not exactly a secret that NVidia's cards are runnin' like dogs on the new explicit APIs.

 

[Bacon1]

What do you expect when the lead rendering programmer is out retweeting "#BetterRed"?

Fortunately my 1080 destroys this game in OpenGL and my upcoming Titan X will do an even better job of it, so whatevs.

Because they never tweeted about 1080 and vulkan.. oh wait

Marty Stratton on stage to unveil #DOOM running on @VulkanAPI & GeForce GTX 1080 #GameReady #FightLikeHell

https://twitter.com/doom/status/729072597774807046

Why don't you guys stop blaming id just like you blamed Oxide, when both id and Oxide work directly with Nvidia daily, and instead blame Nvidia for having shitty drivers.


cursing is not allowed in the technical forums
Markfw900

 

[Shivansps]

Do you think 10 fps less in avg is "getting all out of the hardware"? please, wharever you can do in OGL you could do it better on Vulkan, if you wish to, there is no excuse for the loss of performance, worse case scenario should be equal.

This is a crappy Nvidia rendering path, there is no much else about it. There is no much you could do on drivers with those low end apis, thats the whole point of a low end api.

 

[Arachnotronic]

Why don't you guys stop blaming id just like you blamed Oxide, when both id and Oxide work directly with Nvidia daily, and instead blame Nvidia for having shitty drivers.

Actually, NVIDIA's drivers had excellent support for Doom on day 1, see how well it did in OpenGL relative to the competition which had utterly broken OpenGL drivers.

The (relatively) poor performance on the Vulkan codepath for NVIDIA has nothing to do with NVIDIA's drivers, IMO, as a lot of what the driver handled before is now handled explicitly by the developer.

The Vulkan codepath for Doom is optimized and tweaked for AMD hardware as that codepath likely is a port/derivative of the console codepath.

 

[AtenRa]

Be patient, NV will eventually release a new driver to fix this.

 

[tential]

Actually, NVIDIA's drivers had excellent support for Doom on day 1, see how well it did in OpenGL relative to the competition which had utterly broken OpenGL drivers.

The (relatively) poor performance on the Vulkan codepath for NVIDIA has nothing to do with NVIDIA's drivers, IMO, as a lot of what the driver handled before is now handled explicitly by the developer.

The Vulkan codepath for Doom is optimized and tweaked for AMD hardware as that codepath likely is a port/derivative of the console codepath.

You're seeing the effects of gpu open according to another poster.

Amd unified the libraries between console and desktop so optimizations in console can carry over or something to that effect.

Someone needs to explain that program to me more though because I fail to see how it's open when it's clear amd gains the most benefit from it.

 

[Sweepr]

Do you think 10 fps less in avg is "getting all out of the hardware"? please, wharever you can do in OGL you could do it better on Vulkan, if you wish to, there is no excuse for the loss of performance, worse case scenario should be equal.

Saying it all comes down to drivers is insulting, really shows lack of understanding of the situation here. They were probably 'asked' to release this Vulkan Update before GTX 1060's launch, ready or not. Wouldn't surprise me if future updates (months from now) improve Pascal's performance.

 

[littleg]

You're seeing the effects of gpu open according to another poster.

Amd unified the libraries between console and desktop so optimizations in console can carry over or something to that effect.

Someone needs to explain that program to me more though because I fail to see how it's open when it's clear amd gains the most benefit from it.

Open? As in Open Source? Anyone can use the resources there, modify and distribute them to their hearts content. Practically the very definition of open. It's really not that hard to understand.

 

[Arachnotronic]

Open? As in Open Source? Anyone can use the resources there, modify and distribute them to their hearts content. Practically the very definition of open. It's really not that hard to understand.

Do you think that the AMD exclusive GCN shader intrinsic are "open" to optimization on NVIDIA hardware?

 

[sirmo]

You're seeing the effects of gpu open according to another poster.

Amd unified the libraries between console and desktop so optimizations in console can carry over or something to that effect.

Someone needs to explain that program to me more though because I fail to see how it's open when it's clear amd gains the most benefit from it.

Everyone benefits from it. You can learn how to best utilize the GPU hardware by reading the code, you can use the code in your project.. Nvidia can inspect and make sure there are no benchmark cheating shenanigans in it, unlike what we can do with blackbox Gameworks middleware which is completely closed.

AMD is sharing its code with the community.. which allows us to peer review it, use it, improve it, share it..

Just one more reason to support AMD over Nvidia imo.

 

[MajinCry]

Do you think that the AMD exclusive GCN shader intrinsic are "open" to optimization on NVIDIA hardware?

Uh. Does your browser have some sort of GPU vendor check? If not, then it's as open as can be. https://github.com/GPUOpen-LibrariesAndSDKs

 

[IllogicalGlory]

What do you expect when the lead rendering programmer is out retweeting "#BetterRed"?

Fortunately my 1080 destroys this game in OpenGL and my upcoming Titan X will do an even better job of it, so whatevs.

Weren't you one of the people (correctly) laughing off claims of bias before the Vulkan patch? Now the tables turn and you do the exact same thing, huh?

It's really simple:
-legacy NV products don't properly support DX12/Vulkan and often lose performance as a result.
-AMD products gain a lot of performance under them, often beating their NV competitors as a result.
-iD obviously has marketing deals with both companies.

Is the solution always to throw more money at the problem? It's nice for a $1200 expenditure to be 'whatevs', but not everyone can say the same. A problem that has a $1200 solution is a pretty big problem, especially when the competition got their customers far along the road for no additional cost. NV unfortunately gets to take its customers for granted. Spend $1000 on an OG Titan X, only to find out it's poor at DX12? NV doesn't seem to value its customers very much.

 

[Arachnotronic]

Uh. Does your browser have some sort of GPU vendor check? If not, then it's as open as can be. https://github.com/GPUOpen-LibrariesAndSDKs

Not what I'm talking about. See:

The GCN architecture contains a lot of functionality in the shader cores which is not currently exposed in current APIs like Vulkan™ or Direct3D® 12. One of the mandates of GPUOpen is to give developers better access to the hardware, and today we’re releasing extensions for these APIs to expose additional GCN features to developers.

Shader Extensions
With those shader extensions, we provide access to wavefront-wide functions, which is an important building block to exploit the SIMD execution model of GPUs. For instance, the use of mbcnt and ballot can replace atomics in various cases, drastically boosting performance. The wavefront-wide instructions also include swizzles, which allow individual lanes to exchange data without going through memory.

Additionally, we expose readfirstlane and other functions which enable the compiler to move data from VGPRs into SGPRs. Especially for VGPR heavy code, marking variables as wavefront-uniform can reduce the VGPR count significantly.

Another often-requested feature which is getting exposed today is direct access to the barycentric coordinates. This is again an important building block for various algorithms.

Finally, we also provide various utility functions. In this release, we’re providing the 3-parameter min, max and med functions which map directly to the corresponding GCN opcodes.

http://gpuopen.com/gcn-shader-extensions-for-direct3d-and-vulkan/

There's no way for GCN specific shader intrinsic functions designed to expose functionality exclusive to the GCN shader cores to deliver benefits (or even work frankly) on NVIDIA hardware.

 

[Arachnotronic]

Weren't you one of the people (correctly) laughing off claims of bias before the Vulkan patch? Now the tables turn and you do the exact same thing, huh?

It's really simple:
-legacy NV products don't properly support DX12/Vulkan and often lose performance as a result.
-AMD products gain a lot of performance under them, often beating their NV competitors as a result.
-iD obviously has marketing deals with both companies.

This is exactly what the Vulkan render path in Doom was designed to create the impression of. Seriously, go listen to what Robert Duffy was saying in the RX 480 marketing/launch video and it becomes immediately clear that they did a lot of extra optimization work for AMD GCN hardware in the Vulkan rendering path than they did for the NVIDIA cards.

Like I said before, these devs have to do a LOT of low level optimization to get framerates on consoles to acceptable levels and AMD is exposing a lot of this hardware functionality to developers with DX12/Vulkan (with extensions that go beyond even what DX12/Vulkan provide). The devs can, as in the case of Doom, take their optimizations/tweaks and bring them to GCN with low level APIs like DX12 and Vulkan, which certainly helps AMD hardware look better than NVIDIA's.

The AMD marketing has worked, and from a biz perspective AMD did a very good job here. Props to them on that front.

 

[Shivansps]

With a hand on your heart just tell me, this looks like "extracting all performance out of the hardware" to you?

This is 100% ID Tech fault, trying to blame it on Nvidia drivers is a pittyfull justification by AMD lovers, thats no how low apis work. If 40 fps can be archived by a old, inefficient high level api, them the same can be archived by a new, highly efficient low level api with the right devs work, there is no excuse for the loss of performance, i can fully accept that AMD cards gain MORE fps can Nvidia ones, but not this.

Its VERY funny how Gameworks bashers are defending this ID Tech dissaster and trying to blame it on Nvidia drivers, its VERY funny.

 

[jpiniero]

Well, 2 GB of memory is your problem right there. What res is that, 1080p?

 

[MajinCry]

With a hand on your heart just tell me, this looks like "extracting all performance out of the hardware" to you?

This is 100% ID Tech fault, trying to blame it on Nvidia drivers is a pittyfull justification by AMD lovers, thats no how low apis work. If 40 fps can be archived by a old, inefficient high level api, them the same can be archived by a new, highly efficient low level api with the right devs work, there is no excuse for the loss of performance, i can fully accept that AMD cards gain MORE fps can Nvidia ones, but not this.

Its VERY funny how Gameworks bashers are defending this ID Tech dissaster and trying to blame it on Nvidia drivers, its VERY funny.

Uh, yeah, that's how APIs work. These API's interface through a driver to the GPU, low level or otherwise. And it seems only AMD has a reliable one fer Vulkan. Zlatan states as such 'ere:

http://forums.anandtech.com/showthread.php?p=38227826&highlight=#post38227826

Entirely possible for there to be a performance regression, if the hardware developer has shoddy drivers. Great example; Direct3D 9 -> OpenGL on AMD hardware. OpenGL just performs like a dog. Why? Inadequate drivers.

 

[Arachnotronic]

Well, 2 GB of memory is your problem right there. What res is that, 1080p?

VRAM doesn't change in going from OGL to Vulkan.

 

[jpiniero]

VRAM doesn't change in going from OGL to Vulkan.

Yeah, but devs have typically been using more memory as opposed to nVidia's driver. It's not just OpenGL vs Vulcan either; you see the same thing with DX11 vs DX12.

 

[itsmydamnation]

Is it ID fault? Prove it, show me the traces! DX12/ Vulkan aren't anywhere near as "to the metal" as people think, they just allow you to hang your self in certain area's the driver managed for you before.

Its funny the average render dev is constantly hand tuning Assembly to get maximum IPC best register usage etc yet somehow manically become incompetent when it comes to executing code on a GPU....... funny that.

Or maybe Nvidia's tight lips on the way there uarch actually works comes to bite them in the arse in DX12 vs AMD's openness? Throughput lives and dies on memory management.

 

[krumme]

Salt mine explosion
Dooms vulcan performance coresponds to the cards theoretical throughput. Its not about drivers but what an api can do in an fps if all ressources is used.

 

[Shivansps]

Well, 2 GB of memory is your problem right there. What res is that, 1080p?

Well, yes, but the fact is, it works better with OGL, they cant even do memgr right?

Uh, yeah, that's how APIs work. These API's interface through a driver to the GPU, low level or otherwise. And it seems only AMD has a reliable one fer Vulkan. Zlatan states as such 'ere:

http://forums.anandtech.com/showthread.php?p=38227826&highlight=#post38227826

Entirely possible for there to be a performance regression, if the hardware developer has shoddy drivers. Great example; Direct3D 9 -> OpenGL on AMD hardware. OpenGL just performs like a dog. Why? Inadequate drivers.

You are trying to compare a High Level API Driver, like OpenGL/DX9 to a Low Level API driver like Vulkan, ITS NOT THE SAME, the driver is there to expose the features to applications, but ITS NOT THE SAME, and that is the point actually. There is not much, if any, you can do to improve performance by drivers here, you can expose new features, new extensions, but no much else.