AMD Polaris Thread: Radeon RX 480, RX 470 & RX 460 launching June 29th

[maddie]

because fud.

Which is what I don't understand. What a waste of time and energy.

It's not as if there are alternatives from the opposite camp for any of the new generation designs. There is no choice at the respective price points.

High end buyers have the 1080 and 1070 only.
Mid and low end buyers will have Polaris models only.

In any case, reviews will be out in a few days to give the definitive word.

 

[R0H1T]

The combination effect whould be true under DX12 only, not under DX11, there drivers matters much, much more, im not gona say that consoles support does not matter at all, because specific shaders gona have an effect, althought you can fix that as well on drivers.

We just cant forget that the HD7800 series are still sold today as R7 370, and as a result of that, they still get priority driver optimizations. Compared to GTX600 series that are long forgotten, that what makes the difference, hell we know that old Nvidia cards start to leg behind the newer Nvidia cards as well because of this.

It whould be interesting to see how AMD will handle old GCN on drivers after Polaris launch, AMD today sells GCN 1.0, 1.1 and 1.2... at the same time, as a result of that they all get priority updates.Polaris will be GCN 2.0 i think, reemplacing all old GCN cards, so we cant be sure about what they gona do with the old GCN, neither what they gona do after Polaris get reeplaced. That just guesswork.

They'll get driver updates & optimizations so long as the PS4 & XB1 continue to sell, after that I expect driver updates for a couple of years without much optimization, if at all.

 

[Mikeduffy]

32 ROPs - this means HardOCP was officially wrong about the 480!

They claimed - and still claim - that the 480 is not a successor to the 380, it was conceived as a 390 level replacement. Once AMD realized how pitiful the 480 was, they slashed prices in order to save face.

Anyhow, you don't design a performance level card with 32 ROPs and much lower die size.

Kyle was wrong - plain and simple.

 

[flopper]

There was a dude talking in another forum claiming that every AMD card he ever used gave him near-constant BSODs. And he refused to understand that wasn't normal behavior for AMD drivers, even at their absolute worst, and there was clearly something else wrong with his setup that was causing the crashes.

seems like a former Nvidia user behaviour for me.

 

[Qwertilot]

Which is what I don't understand. What a waste of time and energy.

It's not as if there are alternatives from the opposite camp for any of the new generation designs. There is no choice at the respective price points.

High end buyers have the 1080 and 1070 only.
Mid and low end buyers will have Polaris models only.

In any case, reviews will be out in a few days to give the definitive word.

There will be mid/low end competition, and likely quite soon with it.

Not that it makes all the leaks, insane rumours, FUD etc etc anymore worthwhile!

 

[RussianSensation]

The screenshots I'm seeing at that link indicate 12467 points for standard FireStrike (graphics only), though that's so close that it's really nitpicking.

As someone else mentioned, that's just below R9 390X (12948 points for graphics only). If taken at face value, that would seem to imply a lack of any improvement in perf/TFlop since Hawaii. This is why I believe there's a possibility that the current public drivers are basically treating the Polaris card like an existing GCN product, and that we might realistically see ~10% improvement in the release driver which will have the correct optimizations.

This seems to be one of the most bizarre ways to compare videocards. In modern games, the performance of a videocard is impacted by many factors, not just 1 spec in particular:

1) Shaders (ALUs/shaders)
2) Texture fill-rate (TMUs)
3) Memory bandwidth/compression technology
4) Pixel fill-rate and rasterization (ROPs)
5) Compute (DirectCompute, ASync Compute, Pre-emption, etc.)
6) Cache (L2 caches are continuing to increase)
7) Memory amount (VRAM bottlenecks are becoming more prominent as we enter the 2nd half of the current gen console cycle).
8) Geometry performance (while modern GPUs can handle tessellation better, this is still a key component that impacts GPU performance)

I am not an electrical/GPU engineer, and I can only imagine how complex balancing all of these areas must be when designing a new chip.

To discard all of the facets that make for a balanced GPU design and focus entirely on ALUs (TFlops) is the same thing as judging how advanced/good a modern camera is strictly based on its Megapixel marketing label. While it can be true that a higher megapixel camera can often be superior (assuming it's a balanced design with great optics), it doesn't mean that there is a direct correlation between camera quality and its mega-pixel rating.

Then, why do some insist on deriving/comparing videocards based on Tflops?

GTX580 = 1.58Tflops
GTX680 = 3.25Tflops

OR

R9 290X = 5.63Tflops
Fury X = 8.6Tflops

OR

GTX 780Ti = 5.35Tflops
GTX 980Ti = 6.06Tflops

Just because in some cases Perf/Tflops may scale nearly linearly, it doesn't mean that Tflops itself is an accurate measure of how fast a videocard is without looking at all the other factors. In the modern era, it has now become clear that even memory bandwidth, ROPs, shaders/CUDA cores cannot be directly compared anymore with any reasonable degree of accuracy. There are no more shortcuts to gauge how fast a new videocard will be based on paper specs.

We even saw way back during HD7970 vs. HD6970 that using only paper specs prior to real world results can lead to erroneous conclusions:

"Even with the same number of ROPs and a similar theoretical performance limit (29.6 vs 28.16), 7970 is pushing 51% more pixels than 6970 is."

In the real world, despite the same number of ROPs as the 6970, 7970 crushed it.

We can just as easily use any another example to prove the opposite. Synthetics show that Fury X and 980 absolutely demolish the R9 290X in pixel fill-rate.

Could we approximate real world gaming performance from pixel pushing power of GTX980 and Fury X relative to the R9 290X? Real world games show that R9 290X trades blows with the GTX 980 and Fury X isn't anywhere close to 2.38X faster in games.

More than ever, we absolutely require benchmarks in real world games. Paper specs alone are starting to prove to be very unreliable when looking at how modern game engines and how graphics cards have evolved.

I am actually surprised so many people insist on comparing graphics cards using Tflops. Even the recently launched GTX1070/1080 cards show that no correlation needs to exist.

GTX1080 has 37.3% higher TFlops paper spec but in the real world is only 21-22% faster.

In conclusion, Performance/Tflops seems like another questionable (made-up?) metric to add to the list because it has no reliability whatsoever.

 

[agfkfhahddhdn]

seems like a former Nvidia user behaviour for me.

Well yeah, it was an Nvidia user explaining why he'd never buy another AMD card, for reasons that don't reflect reality.

 

[iiiankiii]

Here's my minimal expectation for the RX480. I want ~390X performance at stock clocks with the ability to overclock to 1500mhz. I have a feeling I'm going to be disappointed in the RX480. AMD always find a way to fail. Please let me be wrong.

 

[JDG1980]

This seems to be one of the most bizarre ways to compare videocards.
[...]
In conclusion, Performance/Tflops seems like another questionable (made-up?) metric to add to the list because it has no reliability whatsoever.

Perf/TFlops is a more accurate term for what some other people have incorrectly called "IPC" in video cards.

The point is, RX 480 has roughly the same raw computing power as R9 390X. If RX 480 isn't better than R9 390X at gaming, that means either there were no architectural improvements, or those improvements were offset by new bottlenecks (memory bandwidth, ROPs, or something else as of yet unknown), or the drivers are not yet mature enough to make good use of the architecture. This doesn't change the fact that RX 480 will still be a good deal, but from the perspective of a tech enthusiast, it would be disappointing.

 

[Elixer]

... but from the perspective of a tech enthusiast, it would be disappointing.

Then perhaps those tech enthusiast should wait for the enthusiast card, namely Vega?

 

[boozzer]

Perf/TFlops is a more accurate term for what some other people have incorrectly called "IPC" in video cards.

The point is, RX 480 has roughly the same raw computing power as R9 390X. If RX 480 isn't better than R9 390X at gaming, that means either there were no architectural improvements, or those improvements were offset by new bottlenecks (memory bandwidth, ROPs, or something else as of yet unknown), or the drivers are not yet mature enough to make good use of the architecture. This doesn't change the fact that RX 480 will still be a good deal, but from the perspective of a tech enthusiast, it would be disappointing.

not disappointing at all. it actually feels like back in the old days. a new xx60 bracket gpu matching the previous gen xx80 or xx70 gpu. in this case, 480 matching 970/980/290x/390x. all of that for 225$ msrp.

let hope the real performance holds come 29th.

 

[Thala]

Here's my minimal expectation for the RX480. I want ~390X performance at stock clocks with the ability to overclock to 1500mhz. I have a feeling I'm going to be disappointed in the RX480. AMD always find a way to fail

These are 2 completely different things, one you are being disappointed by over-hyped expectations, the other is AMD failing with RX480.

If RX 480 isn't better than R9 390X at gaming, that means either there were no architectural improvements

I guess you did not understand what RussíanSensation was talking about. If RX480 is able to achieve R9 390X performance, then is does with less bandwidth, less ROPs and less fillrate. There you have the architectural improvement right in front of your eyes!

 

[RussianSensation]

RS is thinking that 1060/ti will perform better than 480 for around 300$.

I actually don't know how 1060/1060Ti will perform but looking back at GTX460/560 vs. HD5850/6850/6870 or 660/660Ti vs. HD7870/7950, NV doesn't need a card as fast or faster to sell more. It's just a matter of time before NV releases lower end cards but it seems it'll take AMD a lot longer to respond to 1070/1080. The one advantage this time is that AMD is going with a 4/8GB configuration for the 480, but if 1060 is a 192-bit card, it'll likely come with a 3/6GB config. I'd be reluctant to recommend a $200 3GB card in 2016.

32 ROPs - this means HardOCP was officially wrong about the 480!

They claimed - and still claim - that the 480 is not a successor to the 380, it was conceived as a 390 level replacement. Once AMD realized how pitiful the 480 was, they slashed prices in order to save face.

Anyhow, you don't design a performance level card with 32 ROPs and much lower die size.

Kyle was wrong - plain and simple.

I am not sure if he is trying to get a job at NV or he doesn't understand that launches can be in a different order (NV's Maxwell went: GTX750/750Ti - low end first, then GTX970/980 - marketing high-end, then GTX960/950 - mainstream/performance segment, then Titan X/980Ti - enthusiast segment).

R9 290/290X were $399/$549 videocards
R9 390/390X were $329/$429 videocards

AMD is not replacing Hawaii with P10.

Think about HD4850/4870/4890 generation. Had AMD first released HD5750/5770, would we call them HD4870/4890 replacements? Of course not. Instead of releasing Top-Down (HD7950/7970 -> 7850/7870 -> 7750/770), AMD is releasing Bottom-Up (RX 460/470/480 -> Vega 10 -> Vega 11 -> Navi).

These are golden:

Kyle:
"AMD - LEADING FROM BEHIND!" is the new battle call. Chris Hook got promoted this week. He will have his hands full, but he has major skills in the GPU PR universe."

Kyle:
"No this is very much a big deal in the overall strategy that seems to be missing."

:sneaky:

AnandTech - March 15, 2016

"With Polaris confirmed to use GDDR5, Vega is notable for being the first AMD architecture to use HBM2, and the first parts in general to use HBM tech since Fiji. I’m presuming these are higher-end GPUs to complement the Polaris GPUs (the smaller of which we know to be a low-power laptop design), which is where HBM would be more cost-effective, at least at current prices.

Meanwhile AMD has also confirmed the number of GPUs in the Vega stack and their names. We’ll be seeing a Vega 10 and a Vega 11."
http://www.anandtech.com/show/10145/amd-unveils-gpu-architecture-roadmap-after-polaris-comes-vega

What's worse, HardOCP has completely lost its mind without understand that different tiers of GPUs are meant to be compared at different resolutions and yet:

"I think we have made some decisions on how to present our RX 480 review. Given its expectations performance-wise, we are going to step outside our usual comparing equally priced cards a bit and will have a new spin on our highest playable settings that we usually graph. Instead of showing all cards using highest playable settings, we will focus on finding what the best IQ settings are on the RX 480 and then we will set the comparison cards at the same settings. This will give us all "apples to apples" comparisons, but still will be focussed on what the RX 480 can actually achieve at a particular resolution. We are going to focus on 1440p and if we have time, we will hopefully be able to cover some 1080p as well.

Comparison cards Graphed: 480 vs 380X vs 960 - 480 vs 980 v 390X
"
https://hardforum.com/threads/radeon-rx-480-competition-poll.1903083/page-2#post-1042373465

Talk about missing the target market by a country mile. The tiny fraction of 84% of mainstream/performance gamers who are buying a $199-249 graphics card for 1440p gaming with high-end i7 3770K->6800K is almost non-existent. :thumbsdown:

 

[RussianSensation]

Perf/TFlops is a more accurate term for what some other people have incorrectly called "IPC" in video cards.

Absolutely not.

IPC in how it's used on the forums is like this:

GPU 1 clocked at "A" Ghz with a combination of certain fruit (ROPs/TMUs/memory bandwidth) is X% faster than GPU 2 clocked at "B" Ghz with a combination of certain fruit. At least this "IPC" comparison can show a 980Ti user how much more 1070/1080 needs to be overclocked to beat their 1500mhz card. By taking the "combination of fruit" out of the equation, the 980Ti 1500mhz users can easily look at scores of stock 1070/1080 and look at max overclocked scores of 1070/1080 and reasonably estimate how much faster the Pascal cards could be. The Perf/Tflops metric, conversely, is useless.

The point is, RX 480 has roughly the same raw computing power as R9 390X.

Your point? Games aren't only compute limited. That's the entire point I made above but you keep insisting that we should prioritize ALU/shader performance above all other facets that dictate videocard performance in games. We aren't running compute programs on the GPU, which means comparing Tflops on paper is useless unless both GPUs have identical "Combination of fruit" and both videocards are 100% bottlenecked by their ALUs. Since Polaris 10 and Hawaii aren't even the same architecture, your analysis is flawed automatically.

If RX 480 isn't better than R9 390X at gaming, that means either there were no architectural improvements, or those improvements were offset by new bottlenecks (memory bandwidth, ROPs, or something else as of yet unknown), or the drivers are not yet mature enough to make good use of the architecture. This doesn't change the fact that RX 480 will still be a good deal, but from the perspective of a tech enthusiast, it would be disappointing.

You are not making any sense at all here, just like your predictions of GTX 1080 with 4000+ cuda cores and 1080Ti having 5000-6000. If all it takes is a 232mm2 chip with 32 ROPs and 256-bit GDDR5 and 120W power usage to match R9 390, how fast is GCN Gen 4-5 going to be once it's scaled up to 350->600mm2 with 64-96 ROPs, combined GDDR5X and HBM2?

This is like saying because HD5770 failed to smash HD4870/4890, then the entire 14nm line-up from AMD in 2016-2018 is disappointing. How do you not understand that AMD is going to get ~ R9 390 level of performance in a $199-239 price in a 120W power usage? How do you not get that P10 is an R9 380/285 replacement of the new generation and the R9 390/390X replacement will be Vega?

If P10 is disappointing, what was HD5770 according to you? Pure trash? That little card went on to become one of the most popular mainstream cards of that era.

Then perhaps those tech enthusiast should wait for the enthusiast card, namely Vega?

Just think if RX480 outperformed Fury X by even 1% in benchmarks, HardOCP, and some other users on here would still say AMD failed since they'd then claim that AMD's newest flagship is only 1% faster --> to compete they had to slash prices to $199-229 to be able to sell it.

The irony here is the same posters who are bashing on P10 like no tomorrow had no second thoughts recommending the underperforming and overpriced trash that was GTX750/750Ti/950/960 2GB.

NV released GTX760 in June 2013 and replaced it with a card 11-14% faster by January 2015. That means in a span of ~1.5 years, NV improved performance for PC gamers by less than 15% when looking at the mainstream/performance price bracket.

Conversely, the June 2016 RX 480 is going to be close to 70% faster than the January 2015 GTX960. GTX960 went on to become one of the best selling cards of last generation, but RX 480 for $199-249 is disappointing? The brand bias is insane, just insane. I am convinced even if AMD released a card 50% faster than GTX 1080 for $199, some people here would still find something wrong with it.

 

[TrantaLocked]

This seems to be one of the most bizarre ways to compare videocards. In modern games, the performance of a videocard is impacted by many factors, not just 1 spec in particular:

1) Shaders (ALUs/shaders)
2) Texture fill-rate (TMUs)
3) Memory bandwidth/compression technology
4) Pixel fill-rate and rasterization (ROPs)
5) Compute (DirectCompute, ASync Compute, Pre-emption, etc.)
6) Cache (L2 caches are continuing to increase)
7) Memory amount (VRAM bottlenecks are becoming more prominent as we enter the 2nd half of the current gen console cycle).
8) Geometry performance (while modern GPUs can handle tessellation better, this is still a key component that impacts GPU performance)

I am not an electrical/GPU engineer, and I can only imagine how complex balancing all of these areas must be when designing a new chip.

To discard all of the facets that make for a balanced GPU design and focus entirely on ALUs (TFlops) is the same thing as judging how advanced/good a modern camera is strictly based on its Megapixel marketing label. While it can be true that a higher megapixel camera can often be superior (assuming it's a balanced design with great optics), it doesn't mean that there is a direct correlation between camera quality and its mega-pixel rating.

Then, why do some insist on deriving/comparing videocards based on Tflops?

GTX580 = 1.58Tflops
GTX680 = 3.25Tflops

OR

R9 290X = 5.63Tflops
Fury X = 8.6Tflops

OR

GTX 780Ti = 5.35Tflops
GTX 980Ti = 6.06Tflops

Just because in some cases Perf/Tflops may scale nearly linearly, it doesn't mean that Tflops itself is an accurate measure of how fast a videocard is without looking at all the other factors. In the modern era, it has now become clear that even memory bandwidth, ROPs, shaders/CUDA cores cannot be directly compared anymore with any reasonable degree of accuracy. There are no more shortcuts to gauge how fast a new videocard will be based on paper specs.

We even saw way back during HD7970 vs. HD6970 that using only paper specs prior to real world results can lead to erroneous conclusions:

"Even with the same number of ROPs and a similar theoretical performance limit (29.6 vs 28.16), 7970 is pushing 51% more pixels than 6970 is."

In the real world, despite the same number of ROPs as the 6970, 7970 crushed it.

We can just as easily use any another example to prove the opposite. Synthetics show that Fury X and 980 absolutely demolish the R9 290X in pixel fill-rate.

Could we approximate real world gaming performance from pixel pushing power of GTX980 and Fury X relative to the R9 290X? Real world games show that R9 290X trades blows with the GTX 980 and Fury X isn't anywhere close to 2.38X faster in games.

More than ever, we absolutely require benchmarks in real world games. Paper specs alone are starting to prove to be very unreliable when looking at how modern game engines and how graphics cards have evolved.

I am actually surprised so many people insist on comparing graphics cards using Tflops. Even the recently launched GTX1070/1080 cards show that no correlation needs to exist.

GTX1080 has 37.3% higher TFlops paper spec but in the real world is only 21-22% faster.

In conclusion, Performance/Tflops seems like another questionable (made-up?) metric to add to the list because it has no reliability whatsoever.

TIL for a GPU metric to be "reliable" it has to equate to 1:1 FPS performance gains, as if paper specs were ever meant to be 1:1 because GPUs are the only factor for gaming FPS. Yeah let's go ahead and call TFLOPS and every other metric unreliable bullshit.

 

[medi04]

Perf/TFlops is a more accurate term for what some other people have incorrectly called "IPC" in video cards.

The point is, RX 480 has roughly the same raw computing power as R9 390X. If RX 480 isn't better than R9 390X at gaming, that means either there were no architectural improvements, or those improvements were offset by new bottlenecks....

It sounds like you didn't quite read the post you are responding to.

TFlops only tells you about shaders.
They do 2 floating point operations per clock, that times clockrate gives 5.8 for RX 480.

But uh oh, RX 480 somehow consumes 2.5-ish less power (node shrink only gives about 1.7), uses 256 bit instead of 384 bit mem bus, etc. That alone is one hell of an architectural improvement.

 

[crisium]

As far as "IPC" goes, what I want to see is a comparison of 470 vs Tonga. If 470 is either 2048SP or 1792SP it can compare to Tonga if you equalize the frequencies via downclocking.

Like this:

http://www.hardware.fr/articles/926-24/tonga-vs-tahiti.html

I'm really hoping 470 is either of those 2 configurations.

 

[JDG1980]

As far as "IPC" goes, what I want to see is a comparison of 470 vs Tonga. If 470 is either 2048SP or 1792SP it can compare to Tonga if you equalize the frequencies via downclocking.

Like this:

http://www.hardware.fr/articles/926-24/tonga-vs-tahiti.html

I'm really hoping 470 is either of those 2 configurations.

Good point. That would also eliminate fewer ROPs and a narrower memory bus as a factor, since Tonga had the same 32 ROPs and a 256-bit memory bus (in all shipping configurations, I know it was actually 384-bit on-die).

Let's suppose a RX 470 is 2048 shaders. In that case, I'd expect RX 470 downclocked to 970 MHz core and 1425 MHz RAM clock (same as R9 380X) to beat the 380X by at least 10%. Otherwise this is a technical fail, just a glorified die-shrink with no meaningful architectural improvements. Again, it will likely still be a good deal, but I had really hoped that Polaris meant an end to GCN stagnation.

 

[Mopetar]

As far as "IPC" goes, what I want to see is a comparison of 470 vs Tonga. If 470 is either 2048SP or 1792SP it can compare to Tonga if you equalize the frequencies via downclocking.

I honestly don't think there's a huge improvement in the shaders in terms of performance from what AMD has talked about. Any changes were likely there to make the shaders more efficient at existing levels of performance.

Performance improvements are going to come from stuff like their improved memory compression or the primitive discard capabilities that they added to the card.

 

[Ranulf]

Comparison cards Graphed: 480 vs 380X vs 960 - 480 vs 980 v 390X
"[/I]
https://hardforum.com/threads/radeon-rx-480-competition-poll.1903083/page-2#post-1042373465

Talk about missing the target market by a country mile. The tiny fraction of 84% of mainstream/performance gamers who are buying a $199-249 graphics card for 1440p gaming with high-end i7 3770K->6800K is almost non-existent. :thumbsdown:

To be fair, he does seem to take advice and a few posts later says they'll favor 1080p over 1440p. Also says they'll be testing the 4gb version... so yeah. 1440p with the 4gb card, in 2016.

Fun times in the tech sites/forums for sure anyway!

 

[TrantaLocked]

I honestly don't think there's a huge improvement in the shaders in terms of performance from what AMD has talked about. Any changes were likely there to make the shaders more efficient at existing levels of performance.

Performance improvements are going to come from stuff like their improved memory compression or the primitive discard capabilities that they added to the card.

Will the new compression algorithms result in any decrease in visual quality?

 

[RussianSensation]

Otherwise this is a technical fail, just a glorified die-shrink with no meaningful architectural improvements.

You seem to be very much devoted to comparing videocards on paper rather than actually using them (remember how you defended the worthless FurMark as somehow relevant to discussing a videocard's power usage).

GTX960 is an engineering marvel compared to the GTX760, outperforming it in nearly every metric that matters. As an end consumer product, in educated tech circles and longevity wise, it was a terrible product. Perf/mm2, perf/Tflops, all of these metrics are useless if in the end a $149-179 RX 470 and $199-249 RX 480 are the best consumer products in the $150-300 price range. You continue to discredit P10 by coming out with some obsure metrics no one else buy you cares about while ignoring that P10 is about to bring 70% more performance than a GTX960 in the same performance class. It absolutely doesn't matter if AMD used a V12, twin-turbo V6, supercharged V8, twin-scroll turbo inline-6, or a combination of twin-turbo and supercharging. All that matters to the 84% target market is the end result to the consumer.

Right now NV has increased prices of GTX970 -> 1070 by $70-120 ($399-$449) and increased the prices from GTX680 -> 1080 by roughly 24-40% ($499->$620-$699). Use any engineering metric you want in the world, AMD is replacing HD7850 at a similar price, while NV is jacking up prices at the cost of consumer value. Consumers actually buying RX 480 will not care about obscure metrics like Perf/mm2 or Perf/Tflops.

To be fair, he does seem to take advice and a few posts later says they'll favor 1080p over 1440p. Also says they'll be testing the 4gb version... so yeah. 1440p with the 4gb card, in 2016.

Fun times in the tech sites/forums for sure anyway!

Did you read this thread?

https://hardforum.com/threads/from-ati-to-amd-back-to-ati-a-journey-in-futility-h.1900681/page-39

"We are simply trying to make sure you make an informed decision before you plunk down your hard earned cash on hardware."

To earn credibility and trust, a review site MUST be objective for all videocard reviews, not only high-end cards.

ComputerBase clearly showed that GTX960 was the worst x60 videocard in the last 5 generations.
http://www.computerbase.de/2015-03/geforce-gtx-460-560-660-760-960-vergleich/2/

HardOCP gave GTX960 Gold Awards and ignored the amazing value of R9 280X/290 that crushed GTX960 in price/performance for close to a year of that generation. Such fails from "professionals" absolutely cannot be forgiven when European sites made no such mistakes.

It's going to take nothing short of 100% objective reviews, expanding game testing to 10-15 games per review, getting rid of GameWorks settings (some reviews having 80% GameWorks games!) and blaming poor CF support in GWs titles on AMD for HardOCP to even be on the map. HardOCP should just stick to 1440p/4K/5K multi-monitor gaming for $350+ GPUs and become the best it at. For HardOCP to even start comparing GTX1070/1080 to RX 480 earlier without understanding the product positioning in the first place was already eye opening and shockingly embarrassing. It would be akin to expecting a $249 HD7850 to compete with a $399 GTX670 and $499 GTX680. It would be somewhat forgivable had AMD not communicated many times that their high-end cards are Vega, not Polaris.

 

[imported_bman]

Will the new compression algorithms result in any decrease in visual quality?

No, these are all lossless algorithms (delta colour).

 

[Mikeduffy]

Russian Sensation -
You may have a point that he's trying to get a job with Nvidia, his forum posts are clearly anti-AMD. I find it very funny though, before the Macau event he was openly posting that we was going to take his wife along with him for the trip. 😄

Anyways, I dont think you can use prior history of the x60 vs AMD's mainstream any longer - not with dx12 and the console-effect. 1060 at best won't top P10 in dx11 - evidence is there considering we know how the 1080 performs. Nvidia will have to undercut to compete with a 1060.

BTW - the latest rumours are that wafers are pretty scarce at TSMC - why would they bother with a 1060 now when the 1080 is selling well?

 

[Mopetar]

Will the new compression algorithms result in any decrease in visual quality?

No it won't (or it shouldn't unless AMD made changes) as their current memory compression doesn't. Both AMD and Nvidia already use lossless compression, which means when you decompress something, you get the same result as what you had prior to compressing it.

Think of it like the FLAC audio codec which is a way of compressing audio files. All of the information is retained and you can get back exactly the same data as was input.