Now We All See the Genius of AMD Going Lowend First

[pj-]

Who the hell doesn't?

I would dare argue that the vast majority of people who buy high end cards, do so both with an expectation of better performance at the time of purchase but also with an expectation of more longevity than they would get from a mid range card, so in other words plenty of people buy high end cards for performance 1+ year away.

The problem is that trying to predict which card will necessarily perform best 1+ year from now, is an imprecise science to say the least.[/b[

You cut off half of the sentence. "on what is still a tiny minority of games"

In how many games does fury x beat a reference 980ti? Or a factory OC 980ti? Or my 980ti that runs (a bit loudly) at 1500 mhz on air?

Not many, and certainly not to a degree that negates the performance advantage of the 980ti in the other 99.9999% of games that exist. Fury X and 980ti will be 2+ years old by the time the majority of new games are dx12/vulkan, and they will probably be insufficient to run those games at the desired settings/resolution/fps for people who buy high end stuff.

I don't care that the Fury X will beat my 980ti by 20% in some 2017 game because both will be running < 60fps at 1440p high settings.

 

[boozzer]

The GTX 1060 reviewers guide recently leaked. It contains a comparison of RX 480 with GTX 1060 in a bunch of titles, including AoTS. Assuming reviewers are able to replicate such results across a wide variety of titles, and assuming the relative power draw between RX 480 and GTX 1060 are as claimed, NVIDIA has a lower power, generally better performing product than AMD does.

that is why I asked why are you so sure? :thumbsdown:

 

[Silverforce11]

I don't care that the Fury X will beat my 980ti by 20% in some 2017 game because both will be running < 60fps at 1440p high settings.

Maybe the 980Ti will be subpar in 2017 AAA titles (heck, there's a whole batch of them DX12 AAA games later this year, not even 2017!), but Fury X gains performance over time.

Would not surprise me to see it running > 60 fps @ 1440p with High settings. Actually, that should be easy, considering there's Ultra and likely "Hyper/PCMasterRace/Nightmare" after that.

 

[frozentundra123456]

Who the hell doesn't?

I would dare argue that the vast majority of people who buy high end cards, do so both with an expectation of better performance at the time of purchase but also with an expectation of more longevity than they would get from a mid range card, so in other words plenty of people buy high end cards for performance 1+ year away.

The problem is that trying to predict which card will necessarily perform best 1+ year from now, is an imprecise science to say the least.

Exactly. *But* one does know the performance on games already available when the card is released. That is why I would give more weight to known benchmarks at release rather than some mythical performance on some unknown number of unreleased games. Eventually, major titles will migrate to DX12, but nobody knows how many games will ever be released on Vulcan. That is why I think persoanlly the Vulcan results on one single game are being blown way out of proportion.

 

[TestKing123]

Not with the new Asunc Compute patch. When the game is properly coded for DX-12, GCN gains tremendously.

Even with the new patch it still gets spanked by the 1080/1070 series. A Fury X with the new patch still loses to a 980Ti at 1080p, runs even at 1440p. Both are left in the dust by Pascal. This shows not every DX12 game is an automatic win for AMD.

 

[BlueW1zard]

I'm looking to swap out my hot and loud R9 290 (reference) that I bought at release. I love the card and it's served me well but I've got the upgrade bug now and want to get another GPU.

I usually aim for long term when buying a card (2+ years) so I normally get high end or close to high end. I almost impulse purchased a GTX 1070 when I saw the insane benchmark results. Then I started reading into GCN, DX12, Direct Compute etc and watched this video https://www.youtube.com/watch?v=aSYBO1BrB1I

I'm now leaning towards getting a 3rd party Rx 480 with a good overclock to tide me over until early next year and see what the landscape looks like then. I'd rather take a small hit reselling my RX 480 than spend £400 on a GTX 1070.

I remember when I bought my 290 for £300 and felt sorry for those that had just dropped £900 on a Titan. I don't want that feeling when Vega drops and I've just purchased a 1070

I know that there isn't a huge increase going from a 290 to a 480 but with the overclock, better power management, better cooling and quieter, it's a good stop-gap for me until next year.


I agree with those in here that in 6 - 9 months from now AMD will have a better outlook with the increasing spread of DX12 games (especially for Vega)

 

[AtenRa]

Even with the new patch it still gets spanked by the 1080/1070 series. A Fury X with the new patch still loses to a 980Ti at 1080p, runs even at 1440p. Both are left in the dust by Pascal. This shows not every DX12 game is an automatic win for AMD.

I havent seen how Fury X is performing against GTX 1080/70 with the new patch. I would love to see some benchmarks.

 

[nerp]

You just answered your own question. You have not see the stability of the drives for the 1060. You think that the drivers for the 1060 will be good, because of past drivers. That is to say, that you think Nvidia creates better drivers for their card. What is funny to me, is that if you look at AMD driver updates, they have kept improving them to make their cards faster. If you look at a gtx 680 and a 7970, the AMD drivers have made their card much faster. Really, it looks to be that AMD have kept improving their drivers whereas Nvidia has not once they launch a new card.

So, it sure looks like what you are doing is the very thing you seem to question in others.

I never said anything about AMDs drivers. And you're focusing in on one small thing I said which misses the point of what I was saying: fanboyism is hard to understand, or in other words, the passion and framing the hardware marketplace as if it were political. These are just corporations, not the people who set our tax rates and are responsible for picking up our garbage.

 

[Midwayman]

Will never understand this strange compulsion by hardware enthusiasts to choose sides when it comes to brands.

You'd think even Nvidia fans would be rooting for AMD. Strong competition is better for everyone as consumers. Even if I never buy and AMD card I want nvidia to be very afraid I will.

 

[Bacon1]

You cut off half of the sentence. "on what is still a tiny minority of games"

In how many games does fury x beat a reference 980ti? Or a factory OC 980ti? Or my 980ti that runs (a bit loudly) at 1500 mhz on air?

Not many, and certainly not to a degree that negates the performance advantage of the 980ti in the other 99.9999% of games that exist. Fury X and 980ti will be 2+ years old by the time the majority of new games are dx12/vulkan, and they will probably be insufficient to run those games at the desired settings/resolution/fps for people who buy high end stuff.

I don't care that the Fury X will beat my 980ti by 20% in some 2017 game because both will be running < 60fps at 1440p high settings.

Fury X is already on par with 980 TI in DX11, and surpassing it in Vulkan / DX12 which all future AAA games will be using.

https://tpucdn.com/reviews/AMD/RX_480_CrossFire/images/perfrel_2560_1440.png

And thats before this Tomb Raider patch which raised Fury X perf a lot.

 

[Bacon1]

I never said anything about AMDs drivers. And you're focusing in on one small thing I said which misses the point of what I was saying: fanboyism is hard to understand, or in other words, the passion and framing the hardware marketplace as if it were political. These are just corporations, not the people who set our tax rates and are responsible for picking up our garbage.

The point everyone was trying to make is, you are saying you can't understand fanboys, but then say "I like nvidia drivers" which implies that they have better drivers than AMD, and you are a fan of that, which makes you want to buy their cards over AMD.

 

[vissarix]

not looking so good for fury x here :'(

 

[Monk5127]

How are they running full "Nightmare" settings in that bench? My R9 Nano won't let me set "Nightmare" on shadows due to >5GB VRAM.

 

[krumme]

AMD did get spanked in rise of the tomb raider so let's not assume every DX12 game will favour them.

It does have to be clean dx12 vulcan in core not written on the box only with a little functionality patched on. We have seen plenty of in reality in core dx9 games labelled as dx10 or dx11. Hardly something new.

 

[Nhirlathothep]

One common theme on this forum the last few months is how some people have angst/anxiety about the fact that AMD is ceding the highend market (and some of the midrange) to Nvidia all the way until 2017 (unless the freak show 480X2 oops I mean 490 is real). The concept is that there was a lot of pent up demand for new node GPUs, and Nvidia is sucking up all that demand from gamers who want real upgrades over midrange last node GPUs (like the 290).

But I think these amazing Vulkan results, combined with good Directx 12 results for the 480, shows that AMD is the playing the hand they have perfectly. What I think we were all missing (me included) is that winning back market share isn't a one-generation task and AMD has a plan to put the generation AFTER Polaris/Vega in the driver's seat.

If we got Vega this month instead of the 480 it would almost certainly be this pig of a card on power usage compared to a 1080 (given 480 power usage), but also a 1080 (or especially the 1080 ti we all know is coming) would probably clean Vega's freaking clock in Directx 11 games. Branded with the scarlet letter of slower and using more power this AMD generation would be a complete flop, and there would be an anti-Halo effect depressing 480 sales when it eventually released.

Instead with the 480 AMD is winning the pure value equation, which is more important to consumers in its price bracket than power usage by definition because they have less money and need to get as much performance as they can with that money to keep up with the market. That means that a 480 power pig should still sell well next to a more expensive (and efficient) 1060, and more importantly it gives developers a newer target for AMD optimizations in Directx 12 (and maybe Vulkan) to set the stage for Vega.

Now when Vega comes in 2017 reviewers will HAVE TO use game benchmarks that include more Directx 12 titles (because we will have more AAA Directx 12 titles by then) and we will see Vega wipe the floor with the 1080 in those titles. Vega will still be a power pig probably, but gaming enthusiasts will take notice of this late round performance victory and give AMD a brand perception boost they sorely need.

In that scenario, when we get to the 1180/590 generation people who otherwise would run to the first card Nvidia sells might be willing to wait to see what AMD has to offer after knowing how well Vega performed the generation before. If AMD can find some way to make whatever comes after Vega less of a power pig then Nvidia's clear advantages (Directx 11, power consumption) will be gone and AMD will have a wind in its sails like we haven't seen since the switch to Directx 9.

I doubted them but AMD apparently does have a plan other than pure desperate survival. Well played AMD, I never saw it coming.

D:
this is your perception

D:

 

[Thinker_145]

oc3d

The fury x still loses at 1080p but edges the 980Ti at 1440p and 4k (on dx12)

That's just a negligible difference between DX11 and 12. What I meant by AMD losing in this game is that it's slower cards are still slower in DX12 meaning the 480 won't magically come closer to the 1070 like it does in hitman and doom.

 

[pj-]

Fury X is already on par with 980 TI in DX11, and surpassing it in Vulkan / DX12 which all future AAA games will be using.

https://tpucdn.com/reviews/AMD/RX_480_CrossFire/images/perfrel_2560_1440.png

And thats before this Tomb Raider patch which raised Fury X perf a lot.

Sure, a ref fury x is on par with a ref 980ti in DX11 games now. I understand why benchmarks are done ref vs ref for fairness, but it doesn't necessarily apply to me.

My 980ti runs at a 25% higher clock in games than a reference one. So for me the fury x is still far behind in dx11 games. That is also not including the year of gaming I've already had with superior performance.

Like I said before, by the time dx12/vulkan are prevalent enough for me to care about a possible 20% deficit of my card vs. a fury X, I will be in the market for a 1080ti or fury xx.

 

[realibrad]

I never said anything about AMDs drivers. And you're focusing in on one small thing I said which misses the point of what I was saying: fanboyism is hard to understand, or in other words, the passion and framing the hardware marketplace as if it were political. These are just corporations, not the people who set our tax rates and are responsible for picking up our garbage.

You said you like Nvidia drivers. Nvidia drivers do not do anything different for their cards, that AMD drivers do for AMD cards. What about the drivers do you like? The only logical conclusion is that you think their drivers are better, but history has shown that they do not support their cards like AMD does with driver update. My point in pointing this out is to show that you have an unfounded preference which is part of fanboyism. There is not a logical reason, but you still prefer one over the other.

As I said, look at a gtx 680, 780 and look at the driver support vs AMD 7970. What is it that you think you prefer in Nvidia drivers?

 

[Bacon1]

Sure, a ref fury x is on par with a ref 980ti in DX11 games now. I understand why benchmarks are done ref vs ref for fairness, but it doesn't necessarily apply to me.

My 980ti runs at a 25% higher clock in games than a reference one. So for me the fury x is still far behind in dx11 games. That is also not including the year of gaming I've already had with superior performance.

Like I said before, by the time dx12/vulkan are prevalent enough for me to care about a possible 20% deficit of my card vs. a fury X, I will be in the market for a 1080ti or fury xx.

And that is perfectly fine. You like to spend $600+ per year on GPUs.

Most people don't. Heck I'm excited to get my fury air for $310 which is going to have some great support for upcoming games.

 

[3DVagabond]

oc3d

The fury x still loses at 1080p but edges the 980Ti at 1440p and 4k (on dx12)

Except for 1080 where the CPU bottleneck would be the most extreme, nVidia is still suffering performance regression in DX12. The game is a mess.

 

[3DVagabond]

Will never understand this strange compulsion by hardware enthusiasts to choose sides when it comes to brands. I'm probably going to buy a 1060 because it will be fast and relatively inexpensive. I like NVIDIA's drivers. I've had plenty of AMD cards over the years. I'll probably buy another someday. Or, if when the time comes for me to buy a 1060, there is something with better price/perf. available. Why is this so hard for some people?

People like underdogs

Look at the way the two companies do business if you want to find the real reason some people prefer AMD.

 

[linkgoron]

not looking so good for fury x here :'(

Those are great results. The MSI lightning at stock is 21% faster than a stock 980ti at 1440p and the stock fury X is still beating it at 1440p.

They had to OC the 980ti even more for it to actually beat the Fury X.

That's just a negligible difference between DX11 and 12. What I meant by AMD losing in this game is that it's slower cards are still slower in DX12 meaning the 480 won't magically come closer to the 1070 like it does in hitman and doom.

Those are not negligible gains for DX12 vs DX11.
The minimums are 20% higher at 1080p and 55%(!!!) higher at 1440p which really affects playability.

Anyway, I made a mistake. You originally said "AMD did get spanked in rise of the tomb raider so let's not assume every DX12 game will favour them.". They are definitely not getting spanked, although they are on par now. However, the correct graph that I should've shown was the comparison between the original DX12 version and the current version:

20% gain at very high settings at 1440p, and if you check the review, you can also see that they gained 25% in 1080p (didn't want to spam the thread with pics). Those are some major strides. You can't say that there's a negligible difference when you look at that chart.

 

[zinfamous]

This is AMD's problem on the high end.

Who the hell buys a high end card for performance 1+ years away, on what is still a tiny minority of games?

I held off building my computer for months in anticipation of the fury x. I ended up going with a 980ti and don't regret it. Would I buy a 980ti over a fury x now? No, I wouldn't buy either of them.

The second AMD has a card in the $600-800 tier that is faster (on release) than the comparable nvidia card, I will switch. Until then nvidia will remain the only option in the high end single GPU market.

It seems AMD isn't all that interested in ~3% of the actual market looking at their strategy, so you might be waiting a while.

 

[Jay-Sharp]

The Amd vs Nvidia landscape has been pretty brutal lately, and while it's great to debate video cards, most of the last year of recommendations has been based on biased information and complete misinformation. Async compute is now a mythical thing, and Nvidia is a terrible company that just happens to have the fastest gpu out, which only uses 2560 cores, has a 2 ghz clock rate, and includes several smart software innovations, which are all bad things depending on who you ask.

Meanwhile, Amd, who has been underselling over-engineered hardware for years is finally seeing appropriate throughput, and of course it's the new 'sliced bread'.

The truth seems to lie in the middle, with both companies now competing at given tflop metrics, and both stepping up to the plate to make consumers happy. I can attest that the 1080 is an incredible card. So is the 480.

It's very easy to do these companies a disservice by losing sight of the technology at play.

 

[pj-]

It seems AMD isn't all that interested in ~3% of the actual market looking at their strategy, so you might be waiting a while.

A small % of the market can get you big profits if you gouge consumers enough. I wouldn't be surprised if nvidia made more money from 1070 and up cards than 1060 and below, even though the lower end stuff has an order of magnitude or two greater sales.

Apple makes 90-something percent of all smartphone industry profits with 15% market share.

I also doubt that the 1080 and up market is only 3% of all dGPU buyers