AMD Hints that rx 480 has gtx 980/r9 nano Level Performance

[Energycore]

15 hours ago, Suika said:

I could be mistaken, but it could be because the GTX 1080 would have destroyed the RX 480 since it has all of those VR optimizations.

But that's what I don't get from AMD's side. Why not bench against their own cards? Compare CF to the Fury X and single card to the 390/390/Fury so we know what to expect instead of having to do military style data analysis.

[Suika]

10 minutes ago, Energycore said:

But that's what I don't get from AMD's side. Why not bench against their own cards? Compare CF to the Fury X and single card to the 390/390/Fury so we know what to expect instead of having to do military style data analysis.

"Hey, you guys see the competition's highest end solution? Well, we have a solution too!"

 

AMD basically won the middle ground, at least until more of Pascal is here in the 1060 and 1050, but I'd imagine they're trying to make a push into the enthusiast / high-end segment with a dual GPU solution while Vega is in the works.

 

Plus since they don't have exact or close to exact performance numbers, it leaves a shroud of mystery over the cards that leaves people speculating and hyping the product.

[Enderman]

7 hours ago, That Norwegian Guy said:

AotS is the only relevant example of DX12, the only game you could infer future performance tiering from in the current market since it's the only game natively developed for DX12. All other "DX12" games incorporated DX12 support only to  their engine midway during development and aren't using the API like AotS is.

 

So your enitre statement is ridiculous. Why should they show benchmarks of DX11 games, which are about to seize to matter anyway?

 

AMD has been more prepared for DX12 than Nvidia is even as far back as 7950/7970, that's their biggest strength and it does make sense to use it for all it's worth.

so you're saying all the DX12 games in these benchmarks are not actually DX12 games and thats why AMD performs worse?

yeah, i dont believe you lol

[Energycore]

3 minutes ago, Enderman said:

so you're saying all the DX12 games in these benchmarks are not actually DX12 games and thats why AMD performs worse?

yeah, i dont believe you lol

Hitman was built to support DX12 from the ground up, right?

[Enderman]

3 minutes ago, Energycore said:

Hitman was built to support DX12 from the ground up, right?

i have literally no idea what he means by " incorporated DX12 support only to  their engine midway during development and aren't using the API like AotS is."

 

like, what the heck, if they're using DX12 for the engine that means its using the DX12 API...

AOTS is just biased to AMD, like heaven is biased to nvidia

[Energycore]

1 minute ago, Enderman said:

i have literally no idea what he means by " incorporated DX12 support only to  their engine midway during development and aren't using the API like AotS is."

 

like, what the heck, if they're using DX12 to the engine that means its using the DX12 API...

I think he means "other games started by being coded for DX11 and only later did they add code support for DX12", which is inline with stuff I heard.

 

But I also thought Hitman had been developed for DX12 first. Is that true?

[Trixanity]

3 minutes ago, Enderman said:

i have literally no idea what he means by " incorporated DX12 support only to  their engine midway during development and aren't using the API like AotS is."

 

like, what the heck, if they're using DX12 for the engine that means its using the DX12 API...

AOTS is just biased to AMD, like heaven is biased to nvidia

While I won't comment on previous statements. It's not like you can flip a switch and it goes from DX11 to DX12. DX12 gives developers a bigger burden and they do need to work to get performance out of it. The Talos Principle switched to Vulkan but it didn't provide any performance boost because it was basically just using OpenGL within a Vulkan wrapping. So it's not like it's a static +50 FPS boost by switching API.

[Enderman]

15 minutes ago, Energycore said:

I think he means "other games started by being coded for DX11 and only later did they add code support for DX12", which is inline with stuff I heard.

 

But I also thought Hitman had been developed for DX12 first. Is that true?

http://wccftech.com/hitman-feature-implementation-dx12-async-compute-amd/

apparently yes

Quote

HITMAN using DirectX 12, in what has been described as the best implementation yet of Async Compute Engines.

 

[That Norwegian Guy]

27 minutes ago, Energycore said:

Hitman was built to support DX12 from the ground up, right?

Nope, Hitman is developed for DX11, BUT it's somewhere in between a full-fledged DX12 title and one that only added DX12 as an afterthought, as Hitman does make some use of DX12's better resource management. But only on a minor level.

Edit: Hitman was developed for mantle before Mantle truly died, which is why it uses DX12 decently compared to DX12 support only titles.

[Misanthrope]

1 hour ago, Prysin said:

Which got their source from WCCFTech, which got THEIR source from Kotaku, which got THEIR source from the anal cavity of a disgrunteled employee

Source? Kotaku? Requesting a source from them would label you and evil minion working for the patriarchy. 

[Energycore]

5 minutes ago, That Norwegian Guy said:

Nope, Hitman is developed for DX11, BUT it's somewhere in between a full-fledged DX12 title and one that only added DX12 as an afterthought, as Hitman does make some use of DX12's better resource management. But only on a minor level.

Edit: Hitman was developed for mantle before Mantle truly died, which is why it uses DX12 decently compared to DX12 support only titles.

So apart from ashes, any titles that were developed for DX12 from the very start?

 

Perhaps some of the new ones coming out, like BF1

[That Norwegian Guy]

11 minutes ago, Energycore said:

So apart from ashes, any titles that were developed for DX12 from the very start?

 

Perhaps some of the new ones coming out, like BF1

I would assume BF1 will be yeah.

[DocSwag]

6 hours ago, Sakkura said:

You seem to have forgotten about the R9 Fury, which also costs $500.

I thought about that, but considering that it's actually supposed to be a 550 dollar card and it's mail in rebates, sales, etc. that are making its price lower I thought the Nano would be more likely to be what AMD is referring to (if they are referring to one of their own cards at all.

 

1 hour ago, Energycore said:

Please do the thing to make the quotes fine for night theme. Please and thank you.

I know it's messed up, and I did that just for everyone to see the original. I have a version with formatting removed right above the messed up one that is 100% identical except for formatting.

[BuckGup]

2 hours ago, TidaLWaveZ said:

It's true, he showed me the article on huffington post.

Yeah I saw this too!

[TidaLWaveZ]

1 minute ago, SamStrecker said:

Yeah I saw this too!

Told ya.

[Dackzy]

2 hours ago, Enderman said:

AOTS is just biased to AMD, like heaven is biased to nvidia

Just because something has good DX12 support, does it not mean that it is AMD biased, it just means that Nvidia needs to get their shit together.

[Enderman]

1 minute ago, Dackzy said:

Just because something has good DX12 support, does not mean that it is AMD biased, it just means that Nvidia needs to get their shit together.

you could say the same about tessellation "oh heaven is not biased, nvidia is just better at it"

 

but you know what?

normal games perform well on both AMD and nvidia

the fury X was pretty much on par with the 980ti at stock in almost every game

 

then there are things like heaven which have a large advantage on nvidia, or AOTS which has a large advantage on AMD

that is why they are called "____ biased"

and that is why you do not use them to compare the two brands, rarely a game is nvidia biased like heaven, or AMD biased like AOTS

 

benchmarks should give a neutral performance representation of any GPU

because games are meant to work well on any PC, not just one or the other

[Dackzy]

17 minutes ago, Enderman said:

you could say the same about tessellation "oh heaven is not biased, nvidia is just better at it"

 

but you know what?

normal games perform well on both AMD and nvidia

the fury X was pretty much on par with the 980ti at stock in almost every game

 

then there are things like heaven which have a large advantage on nvidia, or AOTS which has a large advantage on AMD

that is why they are called "____ biased"

and that is why you do not use them to compare the two brands, rarely a game is nvidia biased like heaven, or AMD biased like AOTS

 

benchmarks should give a neutral performance representation of any GPU

because games are meant to work well on any PC, not just one or the other

You see DX12 is something Nvidia could get good support for if they did not fuck up this gen.

When we talk about something being biased we talk about something where one brand has something in a game that only that brand have and that the other brand never can get aka property technology, not when it is something they both can get.

[Prysin]

11 minutes ago, Enderman said:

you could say the same about tessellation "oh heaven is not biased, nvidia is just better at it"

 

but you know what?

normal games perform well on both AMD and nvidia

the fury X was pretty much on par with the 980ti at stock in almost every game

 

then there are things like heaven which have a large advantage on nvidia, or AOTS which has a large advantage on AMD

that is why they are called "____ biased"

and that is why you do not use them to compare the two brands, rarely a game is nvidia biased like heaven, or AMD biased like AOTS

 

benchmarks should give a neutral performance representation of any GPU

because games are meant to work well on any PC, not just one or the other

the irony in your statement is that Tesselation was introduced by AMD, and later on adapted and refined by Nvidia.

Heaven is excessively tesselation bound, but it is NOT impossible to run on AMD, nor is it impossible to get a good score on AMD.

 

A test is not "biased", as a "bias" indicates that the test itself was designed specifically FOR a specific purpose.

 

Both AMD and Nvidia CAN run Async Compute. AMD just happens to be better at it. Just like Nvidia happens to be better at Tesselation.

In all honesty, both vendors can shake hands on their mistake.

 

Tesselation is the way forward with VR and AMD is pushing hard for VR. Perhaps they are finally going to try catch up to Nvidia in tesselation? (you cannot use bump-maps or parrallax maps in VR because our brain will easily realize its fake and percieve it as flat. So tesselation is the only way forward with VR).

 

Pascal allows Nvidia to empty large parts of the pipeline in order to prioritize compute workloads. Thus this will make Pascal perform a whole lot better in async, as they wont have to wait around for so long before doing the compute work and going back to graphics.

That being said, we still do not know if Nvidia has increased the queue depht, which can be crucial for Async performance.

 

Either way, calling a test biased only means you are either wanting to, or trying to persuade others that Nvidia is the victim here. Nvidia is not a victim in this test. They chose to not design their GPU architecture for compute workloads and instead they chose to skew their design towards 3D graphics.

 

AMD on the other hand has been too poor, or too lazy (or both) to improve the tesselation performance on their end by any notable amount.

 

There is no victims here, both companies made a decision. Whether or not that was the right decision is anyones guess. Nvidias lack of good Async Compute in maxwell certainly hasnt seem like it has slowed down sales for them.

[Energycore]

1 hour ago, SamStrecker said:

Yeah I saw this too!

Ok that's pretty funny, but you do know children whose parents are infected with HIV also have a high chance of contracting it and dying right?

[Enderman]

3 hours ago, Prysin said:

 

There is no victims here, both companies made a decision. Whether or not that was the right decision is anyones guess. Nvidias lack of good Async Compute in maxwell certainly hasnt seem like it has slowed down sales for them.

Oh, but my point is that most games perform well on both AMD an nvidia

its just that AOTS and heaven stand out for performing much better on one or the other

and its better to use neutral benchmarks to get an idea of relative performance

 

there is no reason to compare cards using small cases like AOTS or heaven

there is no reason for a game to use excessive tessellation to perform better on nvidia, or excessive async compute to perform better on AMD

 

hitman is a good DX12 game which doesnt do either of those things, giving a good idea of performance with no bias to either side

and as expected, a overlocked 980ti performs slightly better than a fury X at stock

 

this is an example of a good game, that is not unrealistically hurting either AMD or nvidia

 

[Trixanity]

2 minutes ago, Enderman said:

Oh, but my point is that most games perform well on both AMD an nvidia

its just that AOTS and heaven stand out for performing much better on one or the other

and its better to use neutral benchmarks to get an idea of relative performance

 

there is no reason to compare cards using small cases like AOTS or heaven

there is no reason for a game to use excessive tessellation to perform better on nvidia, or excessive async compute to perform better on AMD

 

hitman is a good DX12 game which doesnt do either of those things, giving a good idea of performance with no bias to either side

and as expected, a overlocked 980ti performs slightly better than a fury X at stock

 

this is an example of a good game, that is not unrealistically hurting either AMD or nvidia

 

How do you determine if a game is neutral?

[Briggsy]

This sounds like the same marketing as NVidia, where the VR performance is amazeballs compared to regular gaming performance. At least, it better be for AMD or their marketing dept will look like idiots, even more than they did with the FuryX at launch.

[Enderman]

3 hours ago, Dackzy said:

Just because something has good DX12 support, does not mean that it is AMD biased, it just means that Nvidia needs to get their shit together.

its neutral if it lines up with 90% of the other games

 

for example, if the 980ti overclocked performs a few % better than a stock fury X in almost every game, except for game X, then game X wouldnt be considered neutral, and is either biased to one side or the other

 

so you base a purchasing decision based on how a product will almost always perform, not based on how it performs in 10% of games

(unless that 15% is all you play)

[Big-G]

I just laugh when people state DX11 performance isn't important and that DX12 is the only relevant measure of performance. Do you actually think all game developers will immediately jump on DX12? It takes time for engines to mature on new API's... I reckon you'll still see DX11 titles release up until end of 2018...