DOOM updated with Vulkan support

[Silverforce11]

Except it isn't semantics, as you claimed that running DX12 meant you were running code from AMD.

Do you still think AMD isn't involved in DX12 when they worked closely with MS for Xbox way back?

Why is it when DX12 was released, the programming guide lifts chunks of paragraphs straight from AMD's published Mantle programming guide? Do you think that's purely coincidental that two different teams creating different APIs arrive at the same code and structure, but also highly similar text and wording for the guide on how to use it?

If you were a teacher grading that document, you would have failed it for plagiarism. -_-

The progenitor is Johan @ DICE, who wanted this for the PC industry but nobody bothered as he approached all the major players, but AMD saw it as an opportunity and the rest is history.

I guess the more correct way is saying, next-gen APIs are modified AMD/Johan code. So nobody should be surprised if AMD GPUs are punching above their tier.

What's the next phase of AMD's strategy? In 2018, as devs get better at DX12 and Vulkan, and PS4 Neo, Xbox Scorpio have time in the limelight, we will arrive to Navi... multi small GPU dies, linked on an interposer, presenting as a single monolithic GPU in DX12/Vulkan (emphasis on all the different mGPU modes within these API's core). The cost effectiveness of that approach will be significant as smaller nodes encounter difficulties. Such a strategy would not be possible if we're still on serial obsolete APIs like DX11 or OpenGL.

Basically NV will be all aboard, because it will benefit them too down the road.

 

[brandonmatic]

Can you tell me what's wrong with AMD's OpenGL for Doom?

http://www.guru3d.com/articles_pages/amd_radeon_r9_rx_480_8gb_review,14.html

Compare the 290, 290X, 390, 390X vs 970 and 980. Around that ballpark.

Certainly not as horrible as Project Cars deficit.

RX 480 gains up to 50% with Vulkan for example with AC enabled. So does Fury Nano (PCGH claims +50%).

https://www.youtube.com/watch?v=GlzKPBIjZPo

Btw, a 50% faster Nano is actually GTX 1080 or faster.

As I posted in the other Vulkan thread, it seems to more closely reflect peak theoretical compute of each of the cards. I know that's probably too simplistic, but it seems vaguely correct:

• 970 -- 3.4 TFlops vs. 60.6 FPS
• 390 -- 5.1 TFlops vs. 77.4 FPS
• 480 -- 5.2 TFlops vs. 79 FPS
• 980ti -- 6.1 TFlops vs. 85 FPS
• 1070 -- 5.8 TFlops vs. 88 FPS
• Fury X -- 8.6 TFlops vs. 110 FPS

It's not perfect correlation, but it seems pretty telling.

https://www.computerbase.de/2016-07...md-nvidia/#diagramm-doom-mit-vulkan-2560-1440
http://www.pcper.com/reviews/Graphi...-RX-480-will-launch-199-more-5-TFLOPS-compute
http://techreport.com/review/28513/amd-radeon-r9-fury-x-graphics-card-reviewed/4

 

[poofyhairguy]

Hmm, wonder why 980 ti bucks that trend?

Sent from my Nexus 7 using Tapatalk

 

[IronWing]

AMD HD6750 1 GB, passive cooling: averaging 4 fps, peaking at 8 fps, default parameters. :biggrin:

I'm waiting for the custom 1060 reviews then buying either the 480 8GB or the 1060 6GB, whichever is quietest.

 

[Bacon1]

Do we know that's fake? This is possible in DX11 as well, if that's what you're going off of.

Async compute is not available in DX11

 

[Shivansps]

As I posted in the other Vulkan thread, it seems to more closely reflect peak theoretical compute of each of the cards. I know that's probably too simplistic, but it seems vaguely correct:

• 970 -- 3.4 TFlops vs. 60.6 FPS
• 390 -- 5.1 TFlops vs. 77.4 FPS
• 480 -- 5.2 TFlops vs. 79 FPS
• 980ti -- 6.1 TFlops vs. 85 FPS
• 1070 -- 5.8 TFlops vs. 88 FPS
• Fury X -- 8.6 TFlops vs. 110 FPS

It's not perfect correlation, but it seems pretty telling.

https://www.computerbase.de/2016-07...md-nvidia/#diagramm-doom-mit-vulkan-2560-1440
http://www.pcper.com/reviews/Graphi...-RX-480-will-launch-199-more-5-TFLOPS-compute
http://techreport.com/review/28513/amd-radeon-r9-fury-x-graphics-card-reviewed/4

You are impling that Doom is using up to 8.6 TFlops of computing? really? it its calculating subatomic particles or something?

 

[Shivansps]

Async compute is not available in DX11

That my biggest issue with Async, it is not avalible on DX11 or on OpenGL, but attemping to use DX12 or Vulkan code on Nvidia results in worse performance, thats odd.
We all know what everyone whould be saying if it was the other way around on a gamework title.

Also you gotta admit the good timing on the Vulkan path in Doom, this could have not come in a better time for AMD.

 

[IllogicalGlory]

Hmm, wonder why 980 ti bucks that trend?

Sent from my Nexus 7 using Tapatalk

Those are the paper specs, but the actual theoretical performance (what?) is likely higher for both because of GPU boost. Using TPU's average clocks, the 1070 has 6.9 TFLOPS and 980 Ti 6.8.

 

[dogen1]

Async compute is not available in DX11

This guy disagrees...

https://forum.beyond3d.com/threads/intel-gen9-skylake.57204/page-7#post-1869983

I think renderstate even mentioned this once before too.

 

[boozzer]

This guy disagrees...

https://forum.beyond3d.com/threads/intel-gen9-skylake.57204/page-7#post-1869983

I think renderstate even mentioned this once before too.

if you don't need hardware queues, why hasn't nvidia release a software solution? it has been 1 year since win 10/dx12 hit retail.

 

[brandonmatic]

Those are the paper specs, but the actual theoretical performance (what?) is likely higher for both because of GPU boost. Using TPU's average clocks, the 1070 has 6.9 TFLOPS and 980 Ti 6.8.

That's a good point.

 

[boozzer]

That my biggest issue with Async, it is not avalible on DX11 or on OpenGL, but attemping to use DX12 or Vulkan code on Nvidia results in worse performance, thats odd.
We all know what everyone whould be saying if it was the other way around on a gamework title.

Also you gotta admit the good timing on the Vulkan path in Doom, this could have not come in a better time for AMD.

not the same as it doesn't degrade performance, nv gamers can stick with ogl. and this is vulkan, not some proprietary garbage like gameworks with unknown codes. also on some threads on reddit, nv users also reporting fps gains. so we don't even know the cause as to why vulkan works for some and not other nv gamers.

 

[Bacon1]

This guy disagrees...

https://forum.beyond3d.com/threads/intel-gen9-skylake.57204/page-7#post-1869983

I think renderstate even mentioned this once before too.

He doesn't offer any proof or even examples, just says he thinks AMD might be able to do something that offers the same functionality....

 

[Silverforce11]

not the same as it doesn't degrade performance, nv gamers can stick with ogl. and this is vulkan, not some proprietary garbage like gameworks with unknown codes. also on some threads on reddit, nv users also reporting fps gains. so we don't even know the cause as to why vulkan works for some and not other nv gamers.

A lot of reviewers test on latest CPU OC to the max. Average joe is likely on older CPUs and not OC. This is where the draw call overhead reduction of Vulkan/DX12 will come into play.

Like in RotTR, NV users on weaker CPUs see nice gains. NV users on beefy CPU, meh.

The core of draw call overhead applies to all. More for AMD because they never bothered to MT their DX11 driver while NV did.

 

[dogen1]

He doesn't offer any proof or even examples, just says he thinks AMD might be able to do something that offers the same functionality....

No, his words were "it's quite possible", nowhere does the post say "might be". He even says that most DX11 drivers already use parallel hardware engines, and I'm pretty sure he'd be one to know, since he's a graphics engineer at intel.

He even briefly explains exactly why it's possible.

If you still disagree, then you can provide a counter argument if you want.

 

[Silverforce11]

No, his words were "it's quite possible", nowhere does the post say "might be". He even says that most DX11 drivers already use parallel hardware engines, and I'm pretty sure he'd be one to know, since he's a graphics engineer at intel.

He even briefly explains exactly why it's possible.

If you still disagree, then you can provide a counter argument if you want.

Have you seen any examples of Intel Skylake running published DX12 games btw?

They keep saying Skylake iGPU is fully DX12 compatible, even supporting FL12_1, but damn, it's so hard to find legit examples, like a reputable site testing whether those iGPU even run DX12 games.

I know I tried and it crashed in Forza, Hitman, Ashes, TW:Warhammer... wouldn't even boot. I don't own Quantum Break so I can't test that one..

I mean we complain about being lied to by NV about Async Compute but dang, Intel's Skylake DX12 must be the biggest con going around.

 

[dogen1]

Have you seen any examples of Intel Skylake running published DX12 games btw?

They keep saying Skylake iGPU is fully DX12 compatible, even supporting FL12_1, but damn, it's so hard to find legit examples, like a reputable site testing whether those iGPU even run DX12 games.

I know I tried and it crashed in Forza, Hitman, Ashes, TW:Warhammer... wouldn't even boot. I don't own Quantum Break so I can't test that one..

I mean we complain about being lied to by NV about Async Compute but dang, Intel's Skylake DX12 must be the biggest con going around.

No, I've never looked it up.

Have you tried dota 2?

 

[redvapor]

That my biggest issue with Async, it is not avalible on DX11 or on OpenGL, but attemping to use DX12 or Vulkan code on Nvidia results in worse performance, thats odd.
We all know what everyone whould be saying if it was the other way around on a gamework title.

Also you gotta admit the good timing on the Vulkan path in Doom, this could have not come in a better time for AMD.

It could have hit at RX 480 launch and would have been much better. For AMD its 10 days too late.

 

[Bacon1]

No, his words were "it's quite possible", nowhere does the post say "might be". He even says that most DX11 drivers already use parallel hardware engines, and I'm pretty sure he'd be one to know, since he's a graphics engineer at intel.

He even briefly explains exactly why it's possible.

If you still disagree, then you can provide a counter argument if you want.

"It's quite possible they are doing it" meaning, "they could be". Anyway offtopic so made a thread here

http://forums.anandtech.com/showthread.php?t=2480034

 

[ThatBuzzkiller]

Have you seen any examples of Intel Skylake running published DX12 games btw?

They keep saying Skylake iGPU is fully DX12 compatible, even supporting FL12_1, but damn, it's so hard to find legit examples, like a reputable site testing whether those iGPU even run DX12 games.

I know I tried and it crashed in Forza, Hitman, Ashes, TW:Warhammer... wouldn't even boot. I don't own Quantum Break so I can't test that one..

I mean we complain about being lied to by NV about Async Compute but dang, Intel's Skylake DX12 must be the biggest con going around.

Gen 9 can run DX12 games like Quantum Break so you can't exactly claim that their not DX12 compatible ...

AMD and Nvidia drivers are very forgiving when it comes to developers abusing corner cases in DirectX compared to Intel drivers asking for stricter conformance ...

I hoped that lower level APIs would bring stricter practices leading to less external bugs but it looks like that's not the case as far as Intel is concerned and that especially applies to them ...

 

[dogen1]

"It's quite possible they are doing it" meaning, "they could be". Anyway offtopic so made a thread here

[/QUOTE] Yeah, and that's exactly ...g it or not, just that it's possible in DX11.

 

[Silverforce11]

Gen 9 can run DX12 games like Quantum Break so you can't exactly claim that their not DX12 compatible ...

That would be awesome if confirmed. Do you have a link or source for that?

 

[RampantAndroid]

Do you still think AMD isn't involved in DX12 when they worked closely with MS for Xbox way back?

Why is it when DX12 was released, the programming guide lifts chunks of paragraphs straight from AMD's published Mantle programming guide? Do you think that's purely coincidental that two different teams creating different APIs arrive at the same code and structure, but also highly similar text and wording for the guide on how to use it?

If you were a teacher grading that document, you would have failed it for plagiarism. -_-

The progenitor is Johan @ DICE, who wanted this for the PC industry but nobody bothered as he approached all the major players, but AMD saw it as an opportunity and the rest is history.

I guess the more correct way is saying, next-gen APIs are modified AMD/Johan code. So nobody should be surprised if AMD GPUs are punching above their tier.

What's the next phase of AMD's strategy? In 2018, as devs get better at DX12 and Vulkan, and PS4 Neo, Xbox Scorpio have time in the limelight, we will arrive to Navi... multi small GPU dies, linked on an interposer, presenting as a single monolithic GPU in DX12/Vulkan (emphasis on all the different mGPU modes within these API's core). The cost effectiveness of that approach will be significant as smaller nodes encounter difficulties. Such a strategy would not be possible if we're still on serial obsolete APIs like DX11 or OpenGL.

Basically NV will be all aboard, because it will benefit them too down the road.

Except this isn't a 4th grade assignment, and should not be treated as such. For all we know, those sections were written in collaboration with AMD. We have FAR from the whole picture, and so I'm not going to jump to extreme conclusions like you're doing here. And I haven't claimed there was no involvement. I have called into doubt your statement that running a DX12 game means you're running AMD written code (and you weren't implying that the driver was running...)

Either way, it's incredibly off topic...

 

[Bacon1]

Yeah, and that's exactly what I was saying. I wasn't talking about whether AMD is doing it or not, just that it's possible in DX11.

There is no proof that it is able to be done in DX11, hence why AMD created Mantle to make it possible to do so. He was saying that "maybe AMD is doing it somehow" without offering how or proof that they are, or are even capable of doing it.

 

[ThatBuzzkiller]

That would be awesome if confirmed. Do you have a link or source for that?

Here you go ...

https://www.youtube.com/watch?v=8DrPgY2Xjoc