Official AMD Polaris Review Thread: Radeon RX 480, RX 470, and RX 460

[zinfamous]

Possibly only for DX12/Vulkan games, of which are currently a minority of the catalog of games people will be playing.

Posters are being a tad disingenuous when they keep harping about DX12/Vulkan future potential for a product available now that has to deal with 99% of the games not designed to use its strengths.

It's all good and dandy to "future proof" a purchase, but that is basically the sales slogan AMD has been using for almost a half decade. "Just wait and see."

Well, these cards are pretty hard to come by at the moment but when they are more available, is that about when Deus Ex, Watch Dogs, Civ, all those AAA DX12 games are available?

Of course it's also disingenuous to use numbers like "99%" of games when you and I know that this number includes a massive log of games that no one really plays--and a lot of people really just focus on a handful of games at a time, or maybe even just a year. None of those performance is an issue in the vast majority of DX11 games that aren't pushing crazy tech.

It seems to me (As you see with common benchmarks from year to year), that there really is only a few dozen--at most--games each year where performance really matters such that one can start drawing significant comparisons.

To that end, with a dozen DX12 games out right now, and that number doubling within a year's time (moreso, I think), and also representing the majority of new games that people will actually want to play (so, actual sales), DX12 is a far more significant influence than the anti-DX12 crowd will have us believe.

 

[raghu78]

Well, these cards are pretty hard to come by at the moment but when they are more available, is that about when Deus Ex, Watch Dogs, Civ, all those AAA DX12 games are available?

Exactly. If you assume that the average mainstream gamer keeps his card for 2-3 years or more then actually DX12 / Vulkan peformance is going to be more important than DX11. Microsoft pushing for DX12 and Xbox Play anywhere on Microsoft published titles helps improve DX12 adoption.

Of course it's also disingenuous to use numbers like "99%" of games when you and I know that this number includes a massive log of games that no one really plays--and a lot of people really just focus on a handful of games at a time, or maybe even just a year. None of those performance is an issue in the vast majority of DX11 games that aren't pushing crazy tech.

good point. Only a handful of games and game engines really push the envelope in terms of PC graphics. Those games and game engines will benefit the most from the move to DX12. Frostbite is a good example as they are moving to DX12 and async compute with Battlefield 1 being a showcase title for both the Frostbite engine and DX12.

It seems to me (As you see with common benchmarks from year to year), that there really is only a few dozen--at most--games each year where performance really matters such that one can start drawing significant comparisons.

To that end, with a dozen DX12 games out right now, and that number doubling within a year's time (moreso, I think), and also representing the majority of new games that people will actually want to play (so, actual sales), DX12 is a far more significant influence than the anti-DX12 crowd will have us believe.

Actually there are atleast 6 DX12 titles launching in H2 2016 and most of them are highly anticipated and popular franchises. Deus Ex Mankind Divided, Battlefield 1, Watch Dogs 2, Civilization 6, Gears of War 4, Forza Horizon 3.

https://en.wikipedia.org/wiki/List_of_games_with_DirectX_12_support

The first 4 are AMD Gaming Evolved titles so you can bet they will have very good DX12 implementation with async compute. The other 2 should also have excellent DX12 implementation.

 

[railven]

To that end, with a dozen DX12 games out right now, and that number doubling within a year's time (moreso, I think), and also representing the majority of new games that people will actually want to play (so, actual sales), DX12 is a far more significant influence than the anti-DX12 crowd will have us believe.

Actually there are atleast 6 DX12 titles launching in H2 2016 and most of them are highly anticipated and popular franchises. Deus Ex Mankind Divided, Battlefield 1, Watch Dogs 2, Civilization 6, Gears of War 4, Forza Horizon 3.

https://en.wikipedia.org/wiki/List_of_games_with_DirectX_12_support

The first 4 are AMD Gaming Evolved titles so you can bet they will have very good DX12 implementation with async compute. The other 2 should also have excellent DX12 implementation.

You can't argue with numbers. From 6 games to 12 games. What was I thinking.

Guess I'll take that GTX 1060 back. Haha.

 

[boozzer]

I noticed the prices dropping for last generation cards and so I made a performance/price comparison between GTX960, GTX970, RX 480 and GTX 1060.

I used Tomshardware for benchmarks and I eliminated the tests that seem to show poor scaling with the GPUs or skewed results in favor of AMD or nVidia. Out of 8 DX11 games I was left with only 4.
Hitman was showing better performance in general, under AMD cards.
Project CARS had more love for nVidia.
The Division and The Witcher 3 are not scaling well (CPU bottleneck ? dunno ...)

The prices are in euro without vat, cheapest cards on the market. The performances are relative to GTX 970.

         |       |        DX12          |        DX11           |
GPUs     | Price | Perf.   | FPS/Price  | Perf.      | FPS/Price  |
----------------------------------------------------------------
GTX 960  | 131     |  62.91% |   0.205    |  61.33% |   0.425    |
GTX 970  | 212     | 100.00% |   0.201    | 100.00% |   0.428    |
RX 480   | 256     | 112.91% |   0.188    |  90.45% |   0.321    |
GTX 1060 | 259     | 107.75% |   0.177    | 107.73% |   0.378    |

Seems to me this is a good time to buy a GTX 970.

Another thing to mention. Very few models of RX480 were available (3-4 models, but I don't know about actual quantities) and were sold in 2-3 days.
GTX 1060 came with a lot of models and now are almost all out of stock (to be available again in few days).

this ranks amongst the worst advice anyone has ever given on this forum.:thumbsdown::thumbsdown::thumbsdown:

even a 980 @ 200$ is a horrible buy let alone a 3.5 gb 970. :thumbsdown::thumbsdown::thumbsdown:

please stop giving advice that screws over people.

 

[raghu78]

You can't argue with numbers. From 6 games to 12 games. What was I thinking.

Guess I'll take that GTX 1060 back. Haha.

You are never going to buy an AMD GPU no matter what. Don't kid anybody.

 

[Stormflux]

You can't argue with numbers. From 6 games to 12 games. What was I thinking.

Guess I'll take that GTX 1060 back. Haha.

Can you at least argue with some counter facts? Reading this back and forth of DX11 vs DX12/Vulkan games, I see the DX12 camp actually show and confirm games on their side.

Which upcoming AAA(if that) DX11 games are performant critical? Both cards can handle current and yesteryear games better than the last generation.

Personally, and quickly found using this list:
https://en.wikipedia.org/wiki/List_of_games_with_DirectX_11_support#Upcoming_games

Fortnite... which has been in development for 5 years is about the only game I would be looking forward to. But that's UE4, so we'll see when it comes out if it benefits from the engine also updating to the latest APIs.

Starcitizen is already confirmed to be moving towards the new APIs.

I don't see much else on that list.

 

[railven]

Can you at least argue with some counter facts? Reading this back and forth of DX11 vs DX12/Vulkan games, I see the DX12 camp actually show and confirm games on their side.

Which upcoming AAA DX11 games are performant critical? Both cards can handle current and yesteryear games better than their last generation counterparts.

I can rattle a list of DX11 games that will get laughed at because 1) they aren't considered AAA-Devs, at least not by PC-gamer standards 2) they will never be featured in a review site.

Hell the PC game that is most likely going to sell the most copies, any where between 5-7 million if not more is going to be DX11. But you won't see it mentioned because it doesn't support the narrative and it's old as balls.

But I promise you the moment it gets a DX12 update and even more so if it benefits AMD more than Nvidia this place will paste it and its sales number across every single thread possible.

Personally, and quickly found using this list:
https://en.wikipedia.org/wiki/List_of_games_with_DirectX_11_support#Upcoming_games

Fortnite... which has been in development for 5 years is about the only game I would be looking forward to. But that's UE4, so we'll see when it comes out if it benefits from the engine also updating to the latest APIs.

Starcitizen is already confirmed to be moving towards the new APIs.

I don't see much else on that list.

And that's the problem with these arguments. The focus is on AAA-titles that will be featured on review sites. It is nothing about overall performance for the card. When the majority of games from now until 2017 are still DX11 or lower, but "forward thinking" is the current buzz word around here.

Even if I bought all the DX12 games they keep listing, it would still be perhaps 33% if not less of the games I got my eye on. But sadly, they aren't pushing evenlopes, don't have strong following in PC_review sites, or would be acceptable around here as a proper counter argument to the same talking heads.

 

[Stormflux]

I can rattle a list of DX11 games that will get laughed at because 1) they aren't considered AAA-Devs, at least not by PC-gamer standards 2) they will never be featured in a review site.

Hell the PC game that is most likely going to sell the most copies, any where between 5-7 million if not more is going to be DX11. But you won't see it mentioned because it doesn't support the narrative and it's old as balls.

But I promise you the moment it gets a DX12 update and even more so if it benefits AMD more than Nvidia this place will paste it and its sales number across every single thread possible.



And that's the problem with these arguments. The focus is on AAA-titles that will be featured on review sites. It is nothing about overall performance for the card. When the majority of games from now until 2017 are still DX11 or lower, but "forward thinking" is the current buzz word around here.

Even if I bought all the DX12 games they keep listing, it would still be perhaps 33% if not less of the games I got my eye on. But sadly, they aren't pushing evenlopes, don't have strong following in PC_review sites, or would be acceptable around here as a proper counter argument to the same talking heads.

So you can't or won't. So please stop with the drive by remarks that bring nothing of value to the discussion. Who cares what others think of the games you want to play. List them and let others identify with your cause. I'm interested.

 

[railven]

So you can't or won't. So please stop with the drive by remarks that bring nothing of value to the discussion. Who cares what others think of the games you want to play. List them and let others identify with your cause. You sound like a baby.

Gotcha. Glad you bring more to the discussion. Have at it!

 

[Stormflux]

I do apologize, I did edit it. I truly am interested. Neither of us is helping our own cause.

 

[zinfamous]

You can't argue with numbers. From 6 games to 12 games. What was I thinking.

Guess I'll take that GTX 1060 back. Haha.

You ignored the entire point.

from "6 games to 12 games" out of maybe 20-30 games that actually matter (people play/performance relevant). and it is well over 12 games in that set by one year out. it's basically 16 total within the next 4 months.

You do know how population percentages work, right?

"99% of all games, again," is pointless because only a handful of those actually use the GPU in any meaningful way and fewer still are games that are market relevant.

If you look at DX12 now--TODAY--you are already talking about 20% of the current AAA market. But go ahead and ignore that games like Watch Dogs, Battlefield, Deus Ex are about to be released and by the end of this year, we will be talking about 50%+ of the market relevant games. No one will play those or care how they perform with modern API?

So good luck with poorer DX12 (and maybe Vulkan) implementation only a year from now.

 

[zinfamous]

I can rattle a list of DX11 games that will get laughed at because 1) they aren't considered AAA-Devs, at least not by PC-gamer standards 2) they will never be featured in a review site.

never denied that. Is this list of games that you have in mind GPU critical? Is it going to actually challenge a new GPU compared to previous generation? Is your argument actually that these brand new GPUs, which individually perform better than last generation's great-performing GPUs in the very same games, should be considered based on "how well beyond great" they now perform? That sounds odd to me, but OK

Hell the PC game that is most likely going to sell the most copies, any where between 5-7 million if not more is going to be DX11. But you won't see it mentioned because it doesn't support the narrative and it's old as balls.

I'm gonna guess you are talking about WoW or some kinda MOBA? Do you think an old-as-balls game still running on tech that is made irrelevant by the performance of nearly 8 year old class of GPUs is somehow relevant for comparing new GPU performance? At what point past 150+ FPS on ultra settings with whatever crappy GPU you throw at it is it even worth discussing these games in such benchmarks? please...

But I promise you the moment it gets a DX12 update and even more so if it benefits AMD more than Nvidia this place will paste it and its sales number across every single thread possible.

I promise you that if WoW improbably gets a DX12 update that somehow manages to benefit AMD significantly more than nVidia, it would be horribly embarrassing for nVidia. There is nothing in that game that could not have been optimized well beyond reasonable for both architectures at this time. You know, this might be a good barometer to test if there really is a huge difference in the hardware here, or merely driver support. ....but that is well beyond Blizzard's strategy of designing for the cheapest, most ubiquitous lowest-common denominator hardware on the market. ...it really doesn't make sense.

And that's the problem with these arguments. The focus is on AAA-titles that will be featured on review sites. It is nothing about overall performance for the card. When the majority of games from now until 2017 are still DX11 or lower, but "forward thinking" is the current buzz word around here.

Even if I bought all the DX12 games they keep listing, it would still be perhaps 33% if not less of the games I got my eye on. But sadly, they aren't pushing evenlopes, don't have strong following in PC_review sites, or would be acceptable around here as a proper counter argument to the same talking heads.

I think on the surface this sounds like a good argument. But the reality is that while plenty of other games become popular and people legitimately enjoy them (confession: I don't tend to play tons of AAA titles myself--most of those are boring to me--and if I do, it is years later when I can purchase them for pennies ), a lot of these games are chosen because they are rightfully considered to be the titles that actually push the technology--so they are a legitimate method for testing in these benchmarks. Of course part of it is that people want to know which card generally plays better on the most popular titles, but it is also important to choose variables (the individual games) that are actually relevant to your methodology. If a lot of very popular but low end games don't really push the technology and simply show different levels of excellent performance across all cards, then what is the point?

 

[RussianSensation]

this ranks amongst the worst advice anyone has ever given on this forum.

even a 980 @ 200$ is a horrible buy let alone a 3.5 gb 970. :thumbsdown::thumbsdown::thumbsdown:

please stop giving advice that screws over people.

The key is to ignore all the biased drivel and focus on professional reviews.

RX 480 OC beats GTX970 OC.
https://www.youtube.com/watch?v=-h6GiFqPYN8

As Silverforce11 linked, the RX 480 also matches or beats GTX1060 in 70-80% of current DX12/Vulkan games.

RX 480 > GTX970 3.5GB or 980 4GB. GTX1060 makes 970 and 980 irrelevant as well.

Looking back at history of AMD vs. NV since 2012, no NV card has outlasted its AMD competitor. OTOH, AMD cards are free with mining. As long as the gamer is willing to learn, RX 480 makes > $50 USD a month. NV fans will continue to ignore this AMD feature while objective/tech savvy/open-minded PC gamers will keep upgrading to a next gen AMD GPU such as the RX 480 or 490 for free year after year, 2017 being the 9th consecutive year of free AMD GPU upgrades.

You can't argue with numbers. From 6 games to 12 games. What was I thinking.

Guess I'll take that GTX 1060 back. Haha.

That's right.

$199 RX 480 4GB offers the best value out of all the $200-300 videocards on the market. Mirror's Edge Catalyst got patched now which allows the 4GB RX 480 to slice through it. RX 480 4GB just needs a memory overclock and it'll be almost as fast as the 8GB version since 99% of PC games don't use more than 4GB of VRAM at 1080p right now.

https://www.youtube.com/watch?v=JjTUtUF6gxg

The 8GB 480 is 20% more expensive, and the 1060 is 25% more expensive and yet an AIB 1060 is only 3% faster at 1080p than an AIB 480. That means on average RX 480 and GTX1060 are going to provide a similar level of gaming experience:
https://www.computerbase.de/2016-07/powercolor-radeon-rx-480-red-devil-test/3/

As soon as the RX 480 4GB comes back in stock, it'll be the best value card to recommend as a 1-2 year stop gap. In fact, the RX 470 4GB, if priced at $149-159, could actually become the best budget 1080p gaming card of 2016.

The fact that RX 470 and RX 480 all make $$ monthly is an amazing bonus.

In the real world:

$340 i7 6700K
$0 RX 480 (within 6 months)
$0 FreeSync monitor premium

vs.

$340 i7 6700K
$250 GTX1060
$100-200 GSync premium

The NV eco-system will cost $350-450 USD more over 6 months for less than 10% more performance but you save $10 in electricity though! :hmm:

Extra bonus: Once a PC gamer learns to mine and make $ with an AMD card, as long as he keeps mining, the next AMD GPU/CPU upgrade could be financed with $ earned on the 480 or the next card. As long as profitable mining exists, budget gamers buying sub-$300 GPU should only buy AMD. 90% of the performance for free is better than 100% of performance for $250-300.

 

[railven]

You ignored the entire point.

from "6 games to 12 games" out of maybe 20-30 games that actually matter (people play/performance relevant). and it is well over 12 games in that set by one year out. it's basically 16 total within the next 4 months.

You do know how population percentages work, right?

"99% of all games, again," is pointless because only a handful of those actually use the GPU in any meaningful way and fewer still are games that are market relevant.

If you look at DX12 now--TODAY--you are already talking about 20% of the current AAA market. But go ahead and ignore that games like Watch Dogs, Battlefield, Deus Ex are about to be released and by the end of this year, we will be talking about 50%+ of the market relevant games. No one will play those or care how they perform with modern API?

So good luck with poorer DX12 (and maybe Vulkan) implementation only a year from now.

And this is why I can't take some of this arguments seriously. 50%+ of the relevant market is DX12/Vulkan? Who determines relevance?

Hell DOOM wasn't "relevant" until it got the Vulkan patch around these forums. Prior to that it was a shill game by incompetent devs that NV bought.

The amount of flip flopping from both sides makes some arguments hard to counter/follow.

I'll just leave it at that. DX12/Vulkan is now 50%+ of the relevant market. Gotcha.

 

[Arachnotronic]

And this is why I can't take some of this arguments seriously. 50%+ of the relevant market is DX12/Vulkan? Who determines relevance?

Hell DOOM wasn't "relevant" until it got the Vulkan patch around these forums. Prior to that it was a shill game by incompetent devs that NV bought.

The amount of flip flopping from both sides makes some arguments hard to counter/follow.

I'll just leave it at that. DX12/Vulkan is now 50%+ of the relevant market. Gotcha.

Great post.

 

[railven]

In the real world:

$340 i7 6700K
$0 RX 480 (within 6 months)
$0 FreeSync monitor premium

vs.

$340 i7 6700K
$250 GTX1060
$100-200 GSync premium

Does no one see how this is absurd? This is the real world?

So basically if an AMD user is not mining and rocking a freesync monitor, you aren't part of the real world.

I'm also not part of it, I don't own a G-Sync monitor

 

[zinfamous]

And this is why I can't take some of this arguments seriously. 50%+ of the relevant market is DX12/Vulkan? Who determines relevance?

Hell DOOM wasn't "relevant" until it got the Vulkan patch around these forums. Prior to that it was a shill game by incompetent devs that NV bought.

The amount of flip flopping from both sides makes some arguments hard to counter/follow.

I'll just leave it at that. DX12/Vulkan is now 50%+ of the relevant market. Gotcha.

I didn't say this, I said will be by the end of the year and certainly within a year's time.

And you already agreed with me: you complained that these benchmarks are dominated by AAA, super popular games that apparently aren't very representative to what you play. I only defined relevance: useful performance benchmarking/popularity, and pointed out that these well-established AAA DX12 games are about to be here, more will be here after those, and added to what we have NOW, TODAY, we will certainly be seeing 50% and greater relevance of DX12 when it comes to addressing GPU performance, and certainly when it comes to what the market determines as games a lot of people want to play (regardless of whether you or I are in that equation).

How anyone can deny what is so obvious is confusing to me; but it also confuses me that in agreeing with me, then not agreeing with me, you have managed to backtrack on your arguments about which games should be tested and why: games that don't show any relevant GPU influence today--at which point you have to wonder why bother claiming one is better than the other--does it actually matter?

 

[Arachnotronic]

Does no one see how this is absurd? This is the real world?

So basically if an AMD user is not mining and rocking a freesync monitor, you aren't part of the real world.

You're not alone.

 

[zinfamous]

Great post.

If he has to put words in my mouth to make his point, then I guess so?

 

[railven]

I didn't say this, I said will be by the end of the year and certainly within a year's time.

You already counted it, whats the difference between you counting them now and then finally releasing?

And you already agreed with me: you complained that these benchmarks are dominated by AAA, super popular games that apparently aren't very representative to what you play. I only defined relevance: useful performance benchmarking/popularity, and pointed out that these well-established AAA DX12 games are about to be here, more will be here after those, and added to what we have NOW, TODAY, we will certainly be seeing 50% and greater relevance of DX12 when it comes to addressing GPU performance, and certainly when it comes to what the market determines as games a lot of people want to play (regardless of whether you or I are in that equation).

When did I say that? Just about all the games on all review sites I've played. I'm not the one hoping that more DX12 games pop up on them to change the overall tone of reviews in AMDs favor. If they add more DX12/Vulkan games, kudos that's more data points. I think you may have missed what I even argued from the start.

How anyone can deny what is so obvious is confusing to me; but it also confuses me that in agreeing with me, then not agreeing with me, you have managed to backtrack on your arguments about which games should be tested and why: games that don't show any relevant GPU influence today--at which point you have to wonder why bother claiming one is better than the other--does it actually matter?

Find it odd that you think we agreed. You argue relevance and who dictates it, I argue relevance doesn't matter if it's as subjective as a game being denounced today but then heralded simply because of the benchmark results.

The bolted on DX12 games have been horrendous. But a list of "soon to be patched DX12 games" is some how suppose to be inspiring? When the DX12 native titles are still a small subset of the overall library of games...actually never mind you already declared that library irrelevant.

 

[sirmo]

When the DX12 native titles are still a small subset of the overall library of games...actually never mind you already declared that library irrelevant.

Why keep reiterating this same point over and over again. Are you really that big of a luddite?

We all know DX12 and Vulkan is new no one is disputing that.. they are here however, DX11 is legacy, on its way out. If you want a GPU that performs better with old tech get an Nvidia card.


Insulting other members is not allowed
Markfw900

 

[Headfoot]

Lot of rhetoric slinging back and forth.

I think it's entirely fair to inspect performance against multiple population segments. First population segment is obviously all PC games.

Given that these are high performance cards, the next fair population segment is "current games where I need performance" and "future games during the life of my card where I need performance" That is vague -- so lets define it. Game that does not need performance = game that can be run at 120hz at 1440p with lower end last gen hardware e.g. GTX 960 / R7 380 level hardware. These are things like League of Legends, CSGO, TF2, any 2d game, many 3d indie games, etc. Feel free to change out the 120hz or 1440p to whatever your personalized demands are.

The third population is thus games that do need more power. There are two sub populations here: current and future within the expected life span of the card being purchased. There are plenty of DX11 games from the last 2 - 3 years that still cant be pushed to 1440p144 on midrange hardware, so they could use more power to reach maximum picture quality.

Let's say the card lifespan is ~2 years, as that seems to be reasonable. That means the most relevant population of games is games that require more power (per above) now and up to 2 years from now. A ~5 year spread of games (~2013 games through 2018 games).

So it makes sense to analyze what trends are going on and attempt to extrapolate. It's pretty fair to say DX12 and Vulkan adoption will increase through 2018. By how much - no one knows. But certainly more than 12 games. All except the very most recent past games will likely not change from their current API, the vast majority being DX11.

 

[railven]

Why keep reiterating this same point over and over again. Are you really that big of a luddite?

Why are you not chasing the posts. I don't just drop into a thread and go "hey guys, DX11 is still King LOLZ"

I actually don't even remember what brought me to have to restate my position... let me check:

Ah yes, how Power consumption will completely be changed once more DX12 titles are used in reviews and thus making RX 480 more efficient than GTX 1060 (just ignore the other titles).

EDIT:

To revisit it:
Railven: Polaris 10 came out as a pig, not the best foot forward.
Poster: That changes with DX12/Vulkan.
Railven: LOLz, DX11 is still king guys!

 

[AtenRa]

Ah yes, how Power consumption will completely be changed once more DX12 titles are used in reviews and thus making RX 480 more efficient than GTX 1060 (just ignore the other titles).

Power consumption will not get lower with DX-12 on the RX 480, its the perf/watt that will get higher.

Im giving GTX 1060 until December this year, until then Deus Ex, Civilization 6, BF1 and Watch Dogs II will be released. If GTX 1060 will lose in those vs RX 480, then it will be safe to say it will be less future proof than RX 480 because then Sniper 4 and more Microsoft DX-12 GCN/XB1 optimized games will follow in 2017.

 

[nurturedhate]

To revisit it:
Railven: Polaris 10 came out as a pig, not the best foot forward.
Poster: That changes with DX12/Vulkan.
Railven: LOLz, DX11 is still king guys!

To be honest it does tend to read that way but would need to change "Poster: That changes with DX12/Vulkan." to Poster: That changes with DX12/Vulkan, cuz DX11 is DED!" to be more accurate.

There does however need to be change in the way graphics cards are reviewed but no sites want to make the effort and they all have their reasons, good and bad.

Cards need to be tested in actual, used setups. No open benches, no 2 min runs. No one games like that. Some older games need to be thrown out. A shift to to DX12/Vulkan needs to be made as that is where the market it at today and is going tomorrow.

On the flip side if a reviewer is testing an RX 460/470 and whatever the 1050 and lower end up being I would like to see a test with an i5 2400 or an sandy-haswell i3 and test things like LoL, WoW (in an actually demanding area, even a 5 man or LFR), CS:GO, stuff people actually play and are buying the card for.

Credence needs to be given for the current (mix of dx11/12) future (mostly dx12), and what people actually play in a test setting that actually resembles what people actually have. There's no point in testing crysis 3 and metro for another 5 years.

A person can dream right?