Vulkan and Nvidia

[simonlvschal]

Hey Guys,

After the Vulkan patch for DOOM hit, it looks like Nvidia might be in some trouble as time progresses. I bought a GTX 1070 and I feel like i'm going to regret it when more games support Vulkan. How does a 480 beat a 980? If more developers implement Vulkan into games, Nvidia might be in some serious trouble. I wonder if nvidia even plans for the future.

http://imgur.com/gwFtejD

no not really.. the thing is Vulkan is a direct Child of the so crap Mantle.. Aka its born from Mantle.. aka Amd has a huge lead since it's only optimized well for AMD gpu's and also nvidia already got a Very strong API and optimization is very well done. hence the fps. and no Nvidia isnt in any problems since barely any games support Vulkan or Async and u wont see that for about 3 year's by that time nvidia is far ahead again

- - - Updated - - -

Nvidia being bad at DX12/Vulkan is well known and we've been repeating it over and over this year. This isn't exactly unexpected.

thats a fake.. gtx 980 ti on 2560x1440 gets well over 110 fps with vulkan.. and Fury gets around 90..

- - - Updated - - -

Watching the new digital foundry video i must say i'm pretty damn shocked at those results, that's an incredible performance gain for Amd cards.

because the software and hardware amd has is crap.. well Software has been improved alot.. Mainly that API is the reason u see the huge improvement while nvidia did not improve as much as Vulkan is a direct child of MAntle and also nvidia's api and software is already so well optimized u wont see huge improvements.

- - - Updated - - -

oh and also alot of things u people miss is that Ashe.hitman, and a few other games are heavyli optmized to amd. Not nvidia. And yes Nvidia is letting amd get ahead since nvidia has been winning for over 16 years... and will keep going.

amd is low value because they are using cheap Transisitors,Dies. etc

[mrgreenthump]

oh and also alot of things u people miss is that Ashe.hitman, and a few other games are heavyli optmized to amd. Not nvidia. And yes Nvidia is letting amd get ahead since nvidia has been winning for over 16 years... and will keep going.

Which is most of the current DX12 titles? So we can use the heavily Nvidia optimized results, but should ignore those that are AMD optimized.

Tbh we are just starting to see the benefit AMD has, because they have GCN on the the consoles. Console ports have never been super good and they've already been optimized for AMD hardware, so it doesn't take much to optimize it for their other GCN architectures. Nvidia on the other hand has to enable their driver side pre-emption for each game seperately, which from what we've seen so far seems to be a lot of work.

Also Nvidia can't let AMD gain much if they hope to keep their stock favored, which might be impossible. Speaking of, analysts already downgraded Nvidia stock, because AMD is seemingly doing a lot better at the moment.

thats a fake.. gtx 980 ti on 2560x1440 gets well over 110 fps with vulkan.. and Fury gets around 90..

Everything we've seen so far seems to support this "fake". Granted there is fairly little on the matter, because Doom ain't relevant anymore to a lot of people.

[simonlvschal]

Which is most of the current DX12 titles? So we can use the heavily Nvidia optimized results, but should ignore those that are AMD optimized.

Tbh we are just starting to see the benefit AMD has, because they have GCN on the the consoles. Console ports have never been super good and they've already been optimized for AMD hardware, so it doesn't take much to optimize it for their other GCN architectures. Nvidia on the other hand has to enable their driver side pre-emption for each game seperately, which from what we've seen so far seems to be a lot of work.

Also Nvidia can't let AMD gain much if they hope to keep their stock favored, which might be impossible. Speaking of, analysts already downgraded Nvidia stock, because AMD is seemingly doing a lot better at the moment.


Everything we've seen so far seems to support this "fake". Granted there is fairly little on the matter, because Doom ain't relevant anymore to a lot of people.

well its fake cause i have a 980 ti my self.. and i get 110+ fps with 2560x1440... with no oc.

the reason that Sony goes with amd on consoles is cause they are cheaper and eaiser to make. And also console games are far eaiser to optimize since there is only 1 Type og Gpu,Cpu

[Remilia]

Benchmark is compared to itself, not across different systems. There are also certain settings that enable AC which if you don't have on will skew the results.

[Drunkenvalley]

well its fake cause i have a 980 ti my self.. and i get 110+ fps with 2560x1440... with no oc.

the reason that Sony goes with amd on consoles is cause they are cheaper and eaiser to make. And also console games are far eaiser to optimize since there is only 1 Type og Gpu,Cpu

Like we always say with these kinds of things: Prove it. Show your settings and performance (preferably a video). Until then, nobody cares what you tell us you get for fps.

Beyond that, your claims are frankly irrelevant. Sony went with AMD because they're cheap? Naw, it's probably more to do with them being a company that produces high-end items in all categories necessary, even if they don't beat the competition.

Frankly, AMD have produced really good hardware throughout the years, although their firmware and software have left a lot to be desired at times. So likely that's what both Microsoft and Sony were looking for when their API lets them mostly bypass both of those problems.

[Zenny]

That chart is way out of line with almost everything else, but since it shows AMD in a good light it will be shared over and over again.

http://www.eurogamer.net/articles/di...formance-gains

https://www.youtube.com/watch?v=EyuoQIuEZj0

https://www.youtube.com/watch?v=ZQi6wwSOhZU

That chart would have you believe the Fury X is 30% faster then a 980ti and 1070, which is simply not the case.

[mrgreenthump]

That chart is way out of line with almost everything else, but since it shows AMD in a good light it will be shared over and over again.

http://www.eurogamer.net/articles/di...formance-gains

https://www.youtube.com/watch?v=EyuoQIuEZj0

https://www.youtube.com/watch?v=ZQi6wwSOhZU

That chart would have you believe the Fury X is 30% faster then a 980ti and 1070, which is simply not the case.

Nvidia drivers have updated since the chart was made and before they gained nothing from the switch the Vulcan and now they gain up to 10%.

In the end, it really depends on testing methodology. Even in that video the Fury X was way ahead of the 980ti and he could only match them when he stood still with minimal texture changes in the scene. All in all when GPU performance is measured Fury X is ahead in Doom at the moment. By how much? From what I've seen roughly 10-20%(25% in 4k maxed out) depending on resolution and GPU load.

[Zenny]

Nvidia drivers have updated since the chart was made and before they gained nothing from the switch the Vulcan and now they gain up to 10%.

Wrong, all those tests were done before Nvidia released their latest driver.

[Vash The Stampede]

no not really.. the thing is Vulkan is a direct Child of the so crap Mantle.. Aka its born from Mantle.. aka Amd has a huge lead since it's only optimized well for AMD gpu's

Nvidia has contributed to Vulkan just like AMD and many other companies. While Vulkan was born from Mantle, it isn't just like Mantle. Also Valve is involved too, which was to bring an API to Linux that could compete against DX12.

and no Nvidia isnt in any problems since barely any games support Vulkan or Async and u wont see that for about 3 year's by that time nvidia is far ahead again

A fairly modern GPU like the GTX 970 or R9 290 will get at least 60fps at 1080p with all DX11 games. Most new games being released right now will get DX11 and DX12/Vulkan. Even though Vulkan for Doom is overkill in fps, it shows that there's a lot to be had from the API. Even Nvidia's Pascal GPUs benefit from it, just not as much as AMD GPUs.

thats a fake.. gtx 980 ti on 2560x1440 gets well over 110 fps with vulkan.. and Fury gets around 90..

You benchmark the same GPU on the same system. Jimmy down the road gets 120 fps with Vulkan using a Fury. Does my statement carry value? Of course not. But the benchmarks around the internet are pretty consistently in favor of AMD.

because the software and hardware amd has is crap.. well Software has been improved alot.. Mainly that API is the reason u see the huge improvement while nvidia did not improve as much as Vulkan is a direct child of MAntle and also nvidia's api and software is already so well optimized u wont see huge improvements.

Nvidia doesn't have their own API. AMD did at least open source their Mantle API. You don't see that with Physx or CUDA.

oh and also alot of things u people miss is that Ashe.hitman, and a few other games are heavyli optmized to amd. Not nvidia. And yes Nvidia is letting amd get ahead since nvidia has been winning for over 16 years... and will keep going.

More games are optimized for Nvidia than AMD. Every GameWorks game is a Nvidia optimized game. Fallout 4, GTA V, Witcher 3, and Rise of The Tomb Raider. Guess what games review sites use to benchmark? Games like Hitman and Ashes of Singularity are the few games that AMD has helped develop. Doom was assisted by AMD and Nvidia. Don't forget Nvidia showing off the 1080 with Vulkan Doom? And Tomb Raider finally got Async Compute patch which brought AMD much closer to Nvidia in that GameWorks game.

amd is low value because they are using cheap Transisitors,Dies. etc

I know you're trolling but at least try to sound serious.

[simonlvschal]

Nvidia has contributed to Vulkan just like AMD and many other companies. While Vulkan was born from Mantle, it isn't just like Mantle. Also Valve is involved too, which was to bring an API to Linux that could compete against DX12.

A fairly modern GPU like the GTX 970 or R9 290 will get at least 60fps at 1080p with all DX11 games. Most new games being released right now will get DX11 and DX12/Vulkan. Even though Vulkan for Doom is overkill in fps, it shows that there's a lot to be had from the API. Even Nvidia's Pascal GPUs benefit from it, just not as much as AMD GPUs.


You benchmark the same GPU on the same system. Jimmy down the road gets 120 fps with Vulkan using a Fury. Does my statement carry value? Of course not. But the benchmarks around the internet are pretty consistently in favor of AMD.



Nvidia doesn't have their own API. AMD did at least open source their Mantle API. You don't see that with Physx or CUDA.

More games are optimized for Nvidia than AMD. Every GameWorks game is a Nvidia optimized game. Fallout 4, GTA V, Witcher 3, and Rise of The Tomb Raider. Guess what games review sites use to benchmark? Games like Hitman and Ashes of Singularity are the few games that AMD has helped develop. Doom was assisted by AMD and Nvidia. Don't forget Nvidia showing off the 1080 with Vulkan Doom? And Tomb Raider finally got Async Compute patch which brought AMD much closer to Nvidia in that GameWorks game.


I know you're trolling but at least try to sound serious.

why do u think i am trolling? cause your a kid? who think he knows everything? its common sense that cheaper products uses lower quaility componenets.. just look at PSU,Gpu coolers etc.. cheaper = lower quaility. and ye so what if nvidia has more ? that just mean's they are better.. Devs would likely go nvidia cause of their superior Quality and product management..

and also yes Nvidia showed their potential in Doom so what? its not heavyli optimized towards nvidia or amd.. it's using OpenGl which amd doesn't really do a good job at supporting. Nvidia hasn't made the switch yet since Vulkan did not show great potential at launch.. and Nvidia took at bet on OpenGl which has up to now been superior but obviously Vulkan is far better

[Thunderball]

why do u think i am trolling? cause your a kid? who think he knows everything? its common sense that cheaper products uses lower quaility componenets.. just look at PSU,Gpu coolers etc.. cheaper = lower quaility. and ye so what if nvidia has more ? that just mean's they are better.. Devs would likely go nvidia cause of their superior Quality and product management..

and also yes Nvidia showed their potential in Doom so what? its not heavyli optimized towards nvidia or amd.. it's using OpenGl which amd doesn't really do a good job at supporting

It's ok to be an NVIDIA fanboy but it's a completely different thing to make claims that have no connection to the reality.

[mrgreenthump]

its common sense that cheaper products uses lower quaility componenets.. just look at PSU,Gpu coolers etc.. cheaper = lower quaility.

So a proffessional headset costing $100 is worse than a gaming headset that costs $50 more.. Branding plays a big role in pricing and currently Nvidia brand holds a premium and they abuse it. AMD can't price their products any higher, because Nvidia has such a strong brand, much like Iphone had a few years ago.

Talking about those PSUs you mentioned, a Seasonic PSU is ~$50 more expensive than a rebranded Seasonic. They really are the same PSU, but the other one says Seasonic on the side of it.

[simonlvschal]

So a proffessional headset costing $100 is worse than a gaming headset that costs $50 more.. Branding plays a big role in pricing and currently Nvidia brand holds a premium and they abuse it. AMD can't price their products any higher, because Nvidia has such a strong brand, much like Iphone had a few years ago.

Talking about those PSUs you mentioned, a Seasonic PSU is ~$50 more expensive than a rebranded Seasonic. They really are the same PSU, but the other one says Seasonic on the side of it.

i know branding plays alot.. But u do need to know that A Titan Class Card from Nvidia is not for gaming it's for heavy 3D rendering, aka Design and alot of compute power.

Titan Cards are professional grade aka they go through alot of inspection to make sure that the card meets the standards and set goals.. etc hence why a Titan is around that 1000-2000$ and a 980 ti isn't and plus they use higher quaility components that can take more Heat, more Punching aka they last longer under the right preassure.

same goes for Headsets 100$ vs 50$ doesn't always count towards build quaility it can mean that it uses less sound drivers, or worse inner components.
Just look at AudioTechnica? i believe they are called they have low,mid,high end headsets and there is a Clear Difference on price and Quaility.

[Drunkenvalley]

I don't think this guy has the common sense he boasts to have. Common sense is that you don't buy the cheapest, not that things less expensive than others are automatically worse.

The quality increase rises extremely fast just a few tiers up with almost everything, but peters out fast.

And seriously, take your fanboyism elsewhere.

Your comparison on audio is awful. I got some Audio Technica ATH-MX50. Were they more expensive than the junk you get at cheap stores? Yes. Were they expensive? No. Not compared to other audio gear. They're quite cheap in fact. But they're excellent.

[simonlvschal]

I don't think this guy has the common sense he boasts to have. Common sense is that you don't buy the cheapest, not that things less expensive than others are automatically worse.

The quality increase rises extremely fast just a few tiers up with almost everything, but peters out fast.

And seriously, take your fanboyism elsewhere.

Your comparison on audio is awful. I got some Audio Technica ATH-MX50. Were they more expensive than the junk you get at cheap stores? Yes. Were they expensive? No. Not compared to other audio gear. They're quite cheap in fact. But they're excellent.

no not really.. i am not a fanboy if anyone u just need to learn.. instead of pushing away thats how things works and no i am not saying cheaper products are worse not at all.. But in most common things Headsets, Desktop speakers,mobile phones etc.. all of these things follow that princip Lower grade phones tends to use lower grade components. hence their Price same goes for gpu,cpu,motherboards,ssd,hdd and so on.. and obviously branding plays a role but not as much as u think.. Unless u can come up with clear proof that lets say a 500$ headset vs a 100$ headset.. clearly the 500$ one is better right? nope doesn't have to it depends on who made it.. But if its within one brand like Audio techinca. then the 500$ is most likely going to be far superior in quaility and well everything else. Just look at kingstons super cheap headsets they are utterly crap and cost 100$ or so and they use low grade components. but the brand is draggin that price up

[Alindra]

Vulkan is a direct Child of the so crap Mantle

This is a bit of a tangent, but I'm confused how Mantle was "so crap" considering it revolutionized graphics APIs all by itself. Is it because it didn't have a lot of games written for it? It's hardly surprising, it's a lot of work to write an engine for an API that only supports one hardware vendor and/or OS (I can't remember if there was Linux support or if it was Windows only). The fact there was any support for it at all is saying a lot.

[simonlvschal]

I don't think this guy has the common sense he boasts to have. Common sense is that you don't buy the cheapest, not that things less expensive than others are automatically worse.

The quality increase rises extremely fast just a few tiers up with almost everything, but peters out fast.

And seriously, take your fanboyism elsewhere.

Your comparison on audio is awful. I got some Audio Technica ATH-MX50. Were they more expensive than the junk you get at cheap stores? Yes. Were they expensive? No. Not compared to other audio gear. They're quite cheap in fact. But they're excellent.

Audio Technica ATH-MX50 uses worse components then most of their headsets.. you bought a Considered Low end model from their line of products.. were as the more expensive once has better build quaility, drivers, sound etc..

- - - Updated - - -

This is a bit of a tangent, but I'm confused how Mantle was "so crap" considering it revolutionized graphics APIs all by itself. Is it because it didn't have a lot of games written for it? It's hardly surprising, it's a lot of work to write an engine for an API that only supports one hardware vendor and/or OS (I can't remember if there was Linux support or if it was Windows only). The fact there was any support for it at all is saying a lot.

Amd pretty much lied about how good mantle was.. it brought nearly no performence gains.. and it was only going to work for a few games.. and it's no longer being made.. cause of Vulkan

[Alindra]

Amd pretty much lied about how good mantle was.. it brought nearly no performence gains.. and it was only going to work for a few games.. and it's no longer being made.. cause of Vulkan

If there was no performance gains, then it would have been a failure and these low level APIs would have never been created and we would have stuck with DirectX 11 / OpenGL. Since that is not the case, there were obviously performance gains. Star Swarm has been one of the poster childs of how Mantle improved performance.

The thing is, these APIs are very different from the older ones we've been using for ages and engines can't just be retrofitted with them and expect to reap all the benefits. Engines need to be built around them to truly utilize their potential. For example, older APIs are pretty much single threaded where as these new APIs are built on multi-threading. Taking a single threaded task and trying to multi-thread it is not a trivial problem! Engines have to keep track of thing they didn't have to before (what state is this texture in, is it in the upload state, shader read state, etc).

Mantle, while short lived, was revolutionary.

[Drunkenvalley]

Yeah, no, what simon is saying is still mostly just junk. Mantle wasn't the best thing since sliced bread in terms of actual performance gains, etc, but it displayed massive potential due to core principles that AMD were trying to push. These core principles have made their way into both DX12 and Vulkan, because they're good ideas.

[Vash The Stampede]

no not really.. i am not a fanboy if anyone u just need to learn..

The chips themselves very rarely go bad. The chips are manufactured by other companies. The components are chosen by card makers like XFX and ASUS. You need to learn a lot about who makes what.