4K checkerboard rendering is NOT upscale since it has odd or even rendered pixels at their native position, but it's not full native rendered frame buffer.
4K checkerboard rendering enables the machine to nearly halve the memory bandwidth requirements hence enables 2K/3K GPU to reach very near 4K results. Remember, 4K blu-ray video stream is only a very good approximation from uncompressed video stream as H.265 video format is a lossy compression with temporal and motion pixel reconstruction.
It's still upscale because you are interpolating these other odd or even numbered pixels either from previous frames ( as in the case of temporal reconstruction) or you are guessing the color of the intermediate pixel between two consecutively odd or even pixels, which mathematically is still interpolation. Not to mention the end result whether you use checkerboard or any other upscale method will always have the same ratio of native to constructed pixels.
No one will be suing Sony for saying the PS4 Pro can do native 4k. All they need is one game running with native 4k resolution and they can shout their claims from the highest mountain and suffer no repercussions what so ever.
So if the game says "4K compatible" or something on the box, and it isn't native 4K but upscaled, you are saying there would be no chance of a lawsuit? Even in the US?
4K checkerboard rendering is NOT upscale since it has odd or even rendered pixels at their native position, but it's not full native rendered frame buffer.
4K checkerboard rendering enables the machine to nearly halve the memory bandwidth requirements hence enables 2K/3K GPU to reach very near 4K results. Remember, 4K blu-ray video stream is only a very good approximation from uncompressed video stream as H.265 video format is a lossy compression with temporal and motion pixel reconstruction.
It's still upscale because you are interpolating these other odd or even numbered pixels either from previous frames ( as in the case of temporal reconstruction) or you are guessing the color of the intermediate pixel between two consecutively odd or even pixels, which mathematically is still interpolation. Not to mention the end result whether you use checkerboard or any other upscale method will always have the same ratio of native to constructed pixels.
4K framebuffer checkerboard render is not upscale as odd or even rendered pixels are located at their native position. Only missing odd or even pixels are interpolated.
Checkerboard render's missing pixels has pixel data from sides, top and bottom pixels which is far superior to line based interpolation i.e. missing pixel's sides are also missing which relies on previous frame's rendered line.
Rendering a pixel is all about pixel information.
From EA DICE's POV
Again, 4K blu-ray video stream is only a very good approximation from uncompressed video stream as H.265 video format is a lossy compression which also relies on temporal and motion pixel reconstruction methods.
No one will be suing Sony for saying the PS4 Pro can do native 4k. All they need is one game running with native 4k resolution and they can shout their claims from the highest mountain and suffer no repercussions what so ever.
So if the game says "4K compatible" or something on the box, and it isn't native 4K but upscaled, you are saying there would be no chance of a lawsuit? Even in the US?
Note that 4K blu-ray video stream is only a very good approximation from uncompressed video stream as H.265(aka HEVC) video format is a lossy compression which also relies on temporal and motion pixel reconstruction methods.
H.265 lossy mode video stream is technically inferior to a H.265 lossless mode video stream.
4K framebuffer checkerboard render is not upscale as checkerboard's white pixels are rendered in their native position. The geometry layer would be at 4K with odd or even missing pixels has zero shaders applied.
@NFJSupreme said:
"Faux K"
Just like any H.265 lossy mode video stream from UHD Blu-ray.
@xantufrog: How? How would that not be enough? If games keep their current console settings then they could easily run at 4K on a RX 480. Scorpio will be much faster than that so it would not only run most games at 4K but also increase graphical fidelity.
The RX-480 is already 5.5 TFLOPS, and is a 20-35 FPS card in 4K max/ultra, and those are benchmarks with exceptional rigs (so as to avoid bottlenecking outside the GPU). The Scorpio will be somewhat more powerful, but will also being trying to run games 2+ years from now. It's going to be a 30fps machine that slightly outshines the PS4's performance as a 1080 machine at launch near as I can tell.
No one will be suing Sony for saying the PS4 Pro can do native 4k. All they need is one game running with native 4k resolution and they can shout their claims from the highest mountain and suffer no repercussions what so ever.
So if the game says "4K compatible" or something on the box, and it isn't native 4K but upscaled, you are saying there would be no chance of a lawsuit? Even in the US?
I wouldn't say "no chance", but given that a huge proportion of the PS3/360 games had "720p" stamped on the box but were upscaled, and a huge proportion of PS4 and X1 games have "1080p" stamped on the box but are upscaled...
@xantufrog: How? How would that not be enough? If games keep their current console settings then they could easily run at 4K on a RX 480. Scorpio will be much faster than that so it would not only run most games at 4K but also increase graphical fidelity.
The RX-480 is already 5.5 TFLOPS, and is a 20-35 FPS card in 4K max/ultra, and those are benchmarks with exceptional rigs (so as to avoid bottlenecking outside the GPU). The Scorpio will be somewhat more powerful, but will also being trying to run games 2+ years from now. It's going to be a 30fps machine that slightly outshines the PS4's performance as a 1080 machine at launch near as I can tell.
@xantufrog: How? How would that not be enough? If games keep their current console settings then they could easily run at 4K on a RX 480. Scorpio will be much faster than that so it would not only run most games at 4K but also increase graphical fidelity.
The RX-480 is already 5.5 TFLOPS, and is a 20-35 FPS card in 4K max/ultra, and those are benchmarks with exceptional rigs (so as to avoid bottlenecking outside the GPU). The Scorpio will be somewhat more powerful, but will also being trying to run games 2+ years from now. It's going to be a 30fps machine that slightly outshines the PS4's performance as a 1080 machine at launch near as I can tell.
I don't want to come off as condescending, but I don't think you read my post properly. I clearly said at console settings.
Let's not forget that PS4 is more closer to a 470 in real world performance (as guesstimated by ROTR) and Scorpio will potentially be sporting a Vega GPU.
RX 480 is extremely ROP limited, it especially shows in higher resolutions against 290/390/390X. I don't think Scorpio will be short on ROPs, besides, MS were ambitious and claimed native 4K countless times.
I don't want to come off as condescending, but I don't think you read my post properly. I clearly said at console settings.
Then what are you arguing with me about? That was what I said in my original posts on the topic as well...
You actually did not. You just said 'moderate level', which is clearly implying PS4 Pro levels of incapability.
Yeaaaah no. It's your own problem if you are so deep in the fan war that you translate my description of the Scorpio running native 4K at "medium settings, typically sub 60fps" (which is what I said. In fact, I led by saying "either it will be very expensive, or...") as "lolz Scorpio just as weak as PSPro". BTW, what the hell are "console settings" if not something like "medium" or "sub 60fps" or... I made specific predictions about how it will perform at a console-gamer friendly price point, while you made a vague assertion that it will run games "fine at console settings" and you want to go after me for being imprecise in my language? I'm not taking sides in this thread - I rarely do. I want all 3 big companies to succeed.
I don't want to come off as condescending, but I don't think you read my post properly. I clearly said at console settings.
Then what are you arguing with me about? That was what I said in my original posts on the topic as well...
You actually did not. You just said 'moderate level', which is clearly implying PS4 Pro levels of incapability.
Yeaaaah no. It's your own problem if you are so deep in the fan war that you translate my description of the Scorpio running native 4K at "medium settings, typically sub 60fps" (which is was I said. In fact, I went on to say "or it will be very very expensive") as "lolz Scorpio just as weak as PSPro". BTW, what the hell are "console settings" if not something like "medium" or "sub 60fps" or... I made specific predictions about how it will perform at a console-gamer friendly price point, while you made a vague assertion that it will run games "fine at console settings" and you want to go after me for being imprecise in my language? I'm not taking sides in this thread - I rarely do. I want all 3 big companies to succeed.
Me neither and so do I. I'm not accusing you of taking sides, either.
But I don't think you're looking at it objectively, since you just chnaged 'RX 480' to 'Scorpio'.
Let's recap what I said, 480 is more than enough to run games at console settings at 4K, since I didn't previously state the frame-rate, I'll add that I meant at 30fps. And what does 'settings' have anything to do with the frame-rate?
Since Scorpio will be more powerful than a 480 (6 TFLOPS, 380+GB/s bandwidth, potentially Vega), it should be even more capable of doing all that.
@nygamespotter: ok. Thanks for the clarification. I feel like we are more-or-less agreeing on what the Scorpio will be and how it performs, just small differences in how we think about it that got us off on the wrong foot.
Is it wrong to call the PS4 a "4k capable" console? Probably not, due it being able to run some games at native 4k, this is what gets Sony out off a false advertisement law suit.
That being said, is it honest? Probably not.
Yep. The problem with the whole thing is the PR and marketing. Sony will undoubtedly print "PS4 Pro 4k capable" on their boxes, whilst a layman would not understand the difference between native 4k and upscaled 4k.
@xantufrog: How? How would that not be enough? If games keep their current console settings then they could easily run at 4K on a RX 480. Scorpio will be much faster than that so it would not only run most games at 4K but also increase graphical fidelity.
The RX-480 is already 5.5 TFLOPS, and is a 20-35 FPS card in 4K max/ultra, and those are benchmarks with exceptional rigs (so as to avoid bottlenecking outside the GPU). The Scorpio will be somewhat more powerful, but will also being trying to run games 2+ years from now. It's going to be a 30fps machine that slightly outshines the PS4's performance as a 1080 machine at launch near as I can tell.
I don't want to come off as condescending, but I don't think you read my post properly. I clearly said at console settings.
Let's not forget that PS4 is more closer to a 470 in real world performance (as guesstimated by ROTR) and Scorpio will potentially be sporting a Vega GPU.
RX 480 is extremely ROP limited, it especially shows in higher resolutions against 290/390/390X. I don't think Scorpio will be short on ROPs, besides, MS were ambitious and claimed native 4K countless times.
RX-480 is not ROPS limited as Tonga's 32 ROPS with delta compression beats R9-290X's 64 ROPS without delta compression. 3DMark's fill-rate benchmarks shows Tonga's advantage with this specific area.
Fury X's 64 ROPS with delta compression murders R9-290X's 64 ROPS without delta compression.
Polaris 10's 32 ROPS with delta compression is superior to Tonga's 32 ROPS with delta compression i.e. design improvements and higher ROPS clock speed (Polaris 10's 32 ROPS at 1266Mhz is effectively like 40 ROPS at 1000 Mhz).
As for RX-480, any overclock editions will be bounded by effective memory bandwidth.
For example
((256 bit x 9000Mhz) / 8) x Polaris's 75.79 percent memory bandwidth efficiency) x Polaris's compression booster 1.36X = 297 GB/s
For reference RX-480
((256 bit x 8000Mhz) / 8) x Polaris's 75.79 percent memory bandwidth efficiency) x Polaris's compression booster 1.36X = 264 GB/s
--
Scorpio's 320 GB/s memory bandwidth
((384 bit x 6700 Mhz) / 8) x Polaris's 75.79 percent memory bandwidth efficiency) x Polaris's compression booster 1.36X = 333 GB/s
Comparison.
The memory bandwidth gap between Fury X and R9-290 = 1.266X (random textures)
With Fury X, it's memory compression is inferior to NVIDIA's Maxwell.
The FLOPS gap between Fury X and R9-290 = 1.48X
The frame rate gap between R9-290X and Fury X is 1.19X.
Random texture memory bandwidth gap's 1.266X factor is closer to frame rate gap's 1.19X.FLOPS gap between R9-290X (5.8 TFLOPS)and Fury X (8.6 TFLOPS) plays very little part with frame rate gap.
With 980 Ti (5.63 TFLOPS), it's superior memory compression enables it to match Fury X's results.
Conclusion: When there's enough FLOPS for a particular workload, effective memory bandwidth is better prediction method for higher grade GPUs.
-------------------
Example of near brain dead Xbox One ports running PC GPUs.
Frame rate difference between 980 Ti and R9-290X is 1.31X
Effective memory bandwidth between 980 Ti and R9-290X is 1.38X
Forza 6 Apex is another example for effective memory bandwidth influencing the frame rate result.
With checkerboard render, it's still has 4K frame buffer with odd or even pixel not being shaded, but the un-shaded pixels are shaded via interpolation method.
Furthermore, compute shader mode has TMU read and write modes which avoids any ROPS limitations as effective memory bandwidth will be the main limitation.
At 4K, RX-480 is like R9-290X and both has similar effective memory bandwidth i.e. 263 GB/s to 268 GB/s range. R9-390X has 311 GB/s effective memory bandwidth.
Then what are you arguing with me about? That was what I said in my original posts on the topic as well...
You actually did not. You just said 'moderate level', which is clearly implying PS4 Pro levels of incapability.
Yeaaaah no. It's your own problem if you are so deep in the fan war that you translate my description of the Scorpio running native 4K at "medium settings, typically sub 60fps" (which is was I said. In fact, I went on to say "or it will be very very expensive") as "lolz Scorpio just as weak as PSPro". BTW, what the hell are "console settings" if not something like "medium" or "sub 60fps" or... I made specific predictions about how it will perform at a console-gamer friendly price point, while you made a vague assertion that it will run games "fine at console settings" and you want to go after me for being imprecise in my language? I'm not taking sides in this thread - I rarely do. I want all 3 big companies to succeed.
Me neither and so do I. I'm not accusing you of taking sides, either.
But I don't think you're looking at it objectively, since you just chnaged 'RX 480' to 'Scorpio'.
Let's recap what I said, 480 is more than enough to run games at console settings at 4K, since I didn't previously state the frame-rate, I'll add that I meant at 30fps. And what does 'settings' have anything to do with the frame-rate?
Since Scorpio will be more powerful than a 480 (6 TFLOPS, 380+GB/s bandwidth, potentially Vega), it should be even more capable of doing all that.
Vega 10 has 1 TB/s raw memory bandwidth with at least Polaris 10's 1.36X memory compression factor. Vega has 4X perf/watt improvements besides HBM v2 support.
Fury X's HBM v1's 333 GB/s result has 65 percent efficiency with delta compression wasn't able cover the efficiency lost.
The estimated Vega 10's effective memory bandwidth is 884 GB/s based on Fury X's HBM efficiency and Polaris 10's memory compression factor i.e. Vega 10 would blast past GTX 1080 (577 GB/s effective) and rival GTX 1080 Ti (865 GB/s effective estimate i.e. scale from 1080's 256 bit GDDR5X to 384 bit GDDR5X with Pascal's memory compression).
Math
(Vega 10's 1 TB/s x Fury X's 65 percent efficiency) x Polaris memory compression 1.36X factor = 884 GB/s estimate.
Nvidia most likely knows this, hence they are releasing GTX 1080 Ti.
No one will be suing Sony for saying the PS4 Pro can do native 4k. All they need is one game running with native 4k resolution and they can shout their claims from the highest mountain and suffer no repercussions what so ever.
So if the game says "4K compatible" or something on the box, and it isn't native 4K but upscaled, you are saying there would be no chance of a lawsuit? Even in the US?
That's a different thing from Sony saying that the Ps4 Pro is capable of native 4k.
RX-480 is not ROPS limited as Tonga's 32 ROPS with delta compression beats R9-290X's 64 ROPS without delta compression. 3DMark's fill-rate benchmarks shows Tonga's advantage with this specific area.
Fury X's 64 ROPS with delta compression murders R9-290X's 64 ROPS without delta compression.
Polaris 10's 32 ROPS with delta compression is superior to Tonga's 32 ROPS with delta compression i.e. design improvements and higher ROPS clock speed (Polaris 10's 32 ROPS at 1266Mhz is effectively like 40 ROPS at 1000 Mhz).
As for RX-480, any overclock editions will be bounded by effective memory bandwidth.
For example
((256 bit x 9000Mhz) / 8) x Polaris's 75.79 percent memory bandwidth efficiency) x Polaris's compression booster 1.36X = 297 GB/s
For reference RX-480
((256 bit x 8000Mhz) / 8) x Polaris's 75.79 percent memory bandwidth efficiency) x Polaris's compression booster 1.36X = 264 GB/s
--
Scorpio's 320 GB/s memory bandwidth
((384 bit x 6700 Mhz) / 8) x Polaris's 75.79 percent memory bandwidth efficiency) x Polaris's compression booster 1.36X = 333 GB/s
Comparison.
The memory bandwidth gap between Fury X and R9-290 = 1.266X (random textures)
With Fury X, it's memory compression is inferior to NVIDIA's Maxwell.
The FLOPS gap between Fury X and R9-290 = 1.48X
The frame rate gap between R9-290X and Fury X is 1.19X.
Random texture memory bandwidth gap's 1.266X factor is closer to frame rate gap's 1.19X.FLOPS gap between R9-290X (5.8 TFLOPS)and Fury X (8.6 TFLOPS) plays very little part with frame rate gap.
With 980 Ti (5.63 TFLOPS), it's superior memory compression enables it to match Fury X's results.
Conclusion: When there's enough FLOPS for a particular workload, effective memory bandwidth is better prediction method for higher grade GPUs.
-------------------
Example of near brain dead Xbox One ports running PC GPUs.
Frame rate difference between 980 Ti and R9-290X is 1.31X
Effective memory bandwidth between 980 Ti and R9-290X is 1.38X
Forza 6 Apex is another example for effective memory bandwidth influencing the frame rate result.
With checkerboard render, it's still has 4K frame buffer with odd or even pixel not being shaded, but the un-shaded pixels are shaded via interpolation method.
Furthermore, compute shader mode has TMU read and write modes which avoids any ROPS limitations as effective memory bandwidth will be the main limitation.
At 4K, RX-480 is like R9-290X and both has similar effective memory bandwidth i.e. 263 GB/s to 268 GB/s range. R9-390X has 311 GB/s effective memory bandwidth.
While I do think AMD's recent advances in bandwidth efficiency is nothing short of remarkable. The 480 actually IS ROPS limited. So much that it interferes with real world performance, like I previously said, (I was wrong about 200x, I guess) the 390, 390X still edges the 480 out by a significant margin, at least in my opinion.
On paper, 390 and 480 should be somewhat equal, in fact, the 480 is even more powerful (on paper) than the 390 if we factor in the boost clock improvements. And I would say ROPS are definitely a factor in this.
Regardless, I'm expecting Scorpio to sport 64 ROPS especially as it will be targeting 4K, and since it will also be potentially Vega, it should have no problems with ROPerations.
Also, a bit off-topic but the Apex benchmark is a tad outdated. A recent benchmark by Glen.de shows significant performance improvements and an AMD lead (as expected). It will be interesting to see how Horizon 3 fares on the 480.
They never said what kind of 4K, they just said "4K." If people get upset by this they should research before they buy. They aren't technically lying but it does border some unethical marketing.
I'm still excited, but I wonder what will be the target resolution most games will aim to render at. I guess something like 1440p and then checkerboard upscale to 4K. I hope the target is a bit higher though.
LMAO!! So all those 720P and 900P XB1 games that are upscaled to 1080P are ok to say they are 1080P games right? Sony has spoken. Andrew can't be wrong can he? lol!!
4K checkerboard rendering is NOT upscale since it has odd or even rendered pixels at their native position, but it's not full native rendered frame buffer.
4K checkerboard rendering enables the machine to nearly halve the memory bandwidth requirements hence enables 2K/3K GPU to reach very near 4K results. Remember, 4K blu-ray video stream is only a very good approximation from uncompressed video stream as H.265 video format is a lossy compression with temporal and motion pixel reconstruction.
It's still upscale because you are interpolating these other odd or even numbered pixels either from previous frames ( as in the case of temporal reconstruction) or you are guessing the color of the intermediate pixel between two consecutively odd or even pixels, which mathematically is still interpolation. Not to mention the end result whether you use checkerboard or any other upscale method will always have the same ratio of native to constructed pixels.
4K framebuffer checkerboard render is not upscale as odd or even rendered pixels are located at their native position. Only missing odd or even pixels are interpolated.
Checkerboard render's missing pixels has pixel data from sides, top and bottom pixels which is far superior to line based interpolation i.e. missing pixel's sides are also missing which relies on previous frame's rendered line.
Rendering a pixel is all about pixel information.
From EA DICE's POV
Again, 4K blu-ray video stream is only a very good approximation from uncompressed video stream as H.265 video format is a lossy compression which also relies on temporal and motion pixel reconstruction methods.
Yeah, that's what I said, it's still upscale. No matter how you com to the final image, the ratio of native to constructive pixels will always remain same in all upscale methods. The only control you have is on the position of the pixels or the number of native pixels influencing the color of constructed pixel, that's it.
From what I've read, the Pro will make very good use of a 4K display using the cutting edge checkerboard upscaling technique. This article mentions that "it's one of a number of new custom features built into the PS4 Pro's GPU and as such comes with zero cost to game developers."
"I first took a look at the results in Days Gone, the open world survival horror title from Sony Bend. I observed the pixel structure on a 65-inch Sony 4K display from just two feet away, and then I moved closer. It looked good, seriously good. There is a slight softness compared to the pin-sharp precision of a native 4K presentation, but even close-up, the effect works well - in a living room environment, it should work just fine. In common with the other titles using this technique, the demo code we saw can switch in real-time between 1080p and '4K' at the press of a button, with HDR on and off also selectable. There is a clear, unambiguous night and day difference between 1080p and the 4K mode, which is clearly resolving more than the basic 2x increase in pixel throughput being generated at the base level. In fact, the detail increase is almost revelatory - and that applies equally to both Days Gone and Horizon Zero Dawn."
That's impressive that even at 2 feet from a 65 inch display it looked close to native 4K, and who sits that close to a large screen like that when playing on a console anyway?
"But the key takeaway is this - while the PlayStation 4 Pro GPU lacks the horsepower to render out challenging content at native 4K, the presentation we've seen on a number of titles clearly shows a worthwhile, highly desirable increase in fidelity over 1080p - one that does put a 4K screen to good use. Switching between full HD and checkerboard 4K, the increase in detail is simply stunning."
I don't see any problem with 4K being touted as one of the features of the PS4. Even though most games won't be native 4K, there's still a clear benefit there for people with 4K TVs.
No one will be suing Sony for saying the PS4 Pro can do native 4k. All they need is one game running with native 4k resolution and they can shout their claims from the highest mountain and suffer no repercussions what so ever.
So if the game says "4K compatible" or something on the box, and it isn't native 4K but upscaled, you are saying there would be no chance of a lawsuit? Even in the US?
Yes. They'll just remaster Journey, it will get an 11 here, and be native 4K.
This is just becoming great now lol. SONY fan boys all gen was attacking, trashing and belittling the xbox one for not being nactive 1080p when Sony had the higher resolution with a lot of their games and now of the bosses of Playstation comes out and says the same thing about their newest console that all xbox fan boys have being saying about the xbox one LMAO. You can't make this stuff up and now even die hard Sony fan boys are starting to turn away from Sony and the others are digging up all arrow Greenberg twits to damage control this LMAO. I remember I posted a few months back when your a fan boy whatever you say will come back and bite you in the ass at some point cause Sony won't be the stronger forever and then what, will now that time is upon us and you'll are starting to look dumb with your response seeing how you said the complete opposite before lol.
This is just becoming great now lol. SONY fan boys all gen was attacking, trashing and belittling the xbox one for not being nactive 1080p when Sony had the higher resolution with a lot of their games and now of the bosses of Playstation comes out and says the same thing about their newest console that all xbox fan boys have being saying about the xbox one LMAO. You can't make this stuff up and now even die hard Sony fan boys are starting to turn away from Sony and the others are digging up all arrow Greenberg twits to damage control this LMAO. I remember I posted a few months back when your a fan boy whatever you say will come back and bite you in the ass at some point cause Sony won't be the stronger forever and then what, will now that time is upon us and you'll are starting to look dumb with your response seeing how you said the complete opposite before lol.
Should i quote you damage controlling 720p games in the same way you damage control bad score on xbox one..?
Lemmings are graphics whore but since they loss this gen they don't care about graphics or frames,after making threads after thread comparing the PS3 and xbox 360 all of the sudden it didn't matter.
Downplaying the PS4 Pro without taking a huge dump on the xbox one first is impossible,the xbox one S may do 4k for movies but still is a 720p shitty box that can't get 1080p on most games,if you didn't die by it no one will die for PS4 Pro games using checker board rendering which isn't upscaling or native.
Either way a 4 million pixel frame buffer >>>>>>>>>>>>>>> 900k one which you and many lemmings are ok with.
My advice keep enjoying those 720p games on xbox one you have 14 months to go. lol
@blackace said:
LMAO!! So all those 720P and 900P XB1 games that are upscaled to 1080P are ok to say they are 1080P games right? Sony has spoken. Andrew can't be wrong can he? lol!!
Enough say fool they already did it before the console was even launch they damage control 720p games after Albert Panello deny on its own neogaf account that sony had a 40% power advantage.
@blackace said:
They didn't use the cloud for graphics with Titanfall you moron. Pretty much everyone knows that. It was used for the A.I. in Titanfall. It was used for Drivatar in Forza 6 & Horizon 2. Titanfall looks great on the XB1. You have no clue because you don't own one. You idiots were the ones saying DX12 would do nothing on the XB1 hardware. lol!! Benchmarks are already proving you fools wrong. Only a matter of time and all the devs will be talking.
The podcast is up. https://www.youtube.com/watch?v=47cnFWK0dRM
3-5x's faster with DX12. Greatness is coming...
Hahahahaaaaaaaaaaaaaaaaaaaa.... 3 to 5X with DX12...hahahahaa
You are deluded and anything you say should be laugh at...
How does quoting me "damage controling" 720p games, which I never have, have anything to do with what I just posted idiot lol. Your comprehension is as funny as this post cause I've said it in the beginning I don't put resolution of a game over the actually fun factor of a game, you do and you don't even play any games, all you do is Google numbers in your mom's basement and spit the same nonsense over and over again. Before you respond to someone's post, actually try and understand what they are saying clown
“I thought it was quite intriguing given that we’d seen a year of stories about “a new and different Microsoft” in terms of their attitude towards the competition. That was rather short lived, wasn’t it?”
RX-480 is not ROPS limited as Tonga's 32 ROPS with delta compression beats R9-290X's 64 ROPS without delta compression. 3DMark's fill-rate benchmarks shows Tonga's advantage with this specific area.
Fury X's 64 ROPS with delta compression murders R9-290X's 64 ROPS without delta compression.
Polaris 10's 32 ROPS with delta compression is superior to Tonga's 32 ROPS with delta compression i.e. design improvements and higher ROPS clock speed (Polaris 10's 32 ROPS at 1266Mhz is effectively like 40 ROPS at 1000 Mhz).
As for RX-480, any overclock editions will be bounded by effective memory bandwidth.
For example
((256 bit x 9000Mhz) / 8) x Polaris's 75.79 percent memory bandwidth efficiency) x Polaris's compression booster 1.36X = 297 GB/s
For reference RX-480
((256 bit x 8000Mhz) / 8) x Polaris's 75.79 percent memory bandwidth efficiency) x Polaris's compression booster 1.36X = 264 GB/s
--
Scorpio's 320 GB/s memory bandwidth
((384 bit x 6700 Mhz) / 8) x Polaris's 75.79 percent memory bandwidth efficiency) x Polaris's compression booster 1.36X = 333 GB/s
Comparison.
The memory bandwidth gap between Fury X and R9-290 = 1.266X (random textures)
With Fury X, it's memory compression is inferior to NVIDIA's Maxwell.
The FLOPS gap between Fury X and R9-290 = 1.48X
The frame rate gap between R9-290X and Fury X is 1.19X.
Random texture memory bandwidth gap's 1.266X factor is closer to frame rate gap's 1.19X.FLOPS gap between R9-290X (5.8 TFLOPS)and Fury X (8.6 TFLOPS) plays very little part with frame rate gap.
With 980 Ti (5.63 TFLOPS), it's superior memory compression enables it to match Fury X's results.
Conclusion: When there's enough FLOPS for a particular workload, effective memory bandwidth is better prediction method for higher grade GPUs.
-------------------
Example of near brain dead Xbox One ports running PC GPUs.
Frame rate difference between 980 Ti and R9-290X is 1.31X
Effective memory bandwidth between 980 Ti and R9-290X is 1.38X
Forza 6 Apex is another example for effective memory bandwidth influencing the frame rate result.
With checkerboard render, it's still has 4K frame buffer with odd or even pixel not being shaded, but the un-shaded pixels are shaded via interpolation method.
Furthermore, compute shader mode has TMU read and write modes which avoids any ROPS limitations as effective memory bandwidth will be the main limitation.
At 4K, RX-480 is like R9-290X and both has similar effective memory bandwidth i.e. 263 GB/s to 268 GB/s range. R9-390X has 311 GB/s effective memory bandwidth.
While I do think AMD's recent advances in bandwidth efficiency is nothing short of remarkable. The 480 actually IS ROPS limited. So much that it interferes with real world performance, like I previously said, (I was wrong about 200x, I guess) the 390, 390X still edges the 480 out by a significant margin, at least in my opinion.
On paper, 390 and 480 should be somewhat equal, in fact, the 480 is even more powerful (on paper) than the 390 if we factor in the boost clock improvements. And I would say ROPS are definitely a factor in this.
Regardless, I'm expecting Scorpio to sport 64 ROPS especially as it will be targeting 4K, and since it will also be potentially Vega, it should have no problems with ROPerations.
Also, a bit off-topic but the Apex benchmark is a tad outdated. A recent benchmark by Glen.de shows significant performance improvements and an AMD lead (as expected). It will be interesting to see how Horizon 3 fares on the 480.
No, RX-480 is mostly memory bandwidth bound.
RGBA32F format, RX-480's 1266 Mhz x 32 ROPS x 16 bytes = 648 GB/s Note why better memory compression and HBM v2 or 384 bit GDDR5X are needed.
RGBA16F format, RX-480's 1266 Mhz x 32 ROPS x 8 bytes = 324 GB/s. Scorpio just maximized this color format.
R9-390X still has the higher effective memory bandwidth over RX-480.
Anyway... AMD's 4K benchmarks for R9-390X
VEGA 10 is the large VEGA chip with 1 TB/s of four HBM v2 modules and 4X perf/watt improvements. The lesser VEGA that replaces Polaris 10 is TBA.
Yeah, that's what I said, it's still upscale. No matter how you com to the final image, the ratio of native to constructive pixels will always remain same in all upscale methods. The only control you have is on the position of the pixels or the number of native pixels influencing the color of constructed pixel, that's it.
Again, upscale is pure pixel copy to target resolution. EA DICE has stated their solution doesn't fit the definition for upscale and native i.e. it's custom.
Note that lossy compression 4K UHD-blu-ray videos are not pixel perfect when compared to lossless video formats.
LOL! You keep bringing the past because you can't handle that Sony is lying about it's Faux-K ability! Besides, you don't even own or plan on getting a PS4 Pro, so don't even bother!
Looking very forward to the first upscaled Scorpio games. You lems can't help yourselves when it comes to smacking a hornets nest with a stick. Just don't get angry when you get stung.
Yeah, that's what I said, it's still upscale. No matter how you com to the final image, the ratio of native to constructive pixels will always remain same in all upscale methods. The only control you have is on the position of the pixels or the number of native pixels influencing the color of constructed pixel, that's it.
Again, upscale is pure pixel copy to target resolution. EA DICE has stated their solution doesn't fit the definition for upscale and native i.e. it's custom.
Note that lossy compression 4K UHD-blu-ray videos are not pixel perfect when compared to lossless video formats.
I don't care what developers want to call their solution. It's all PR talk. Technically if the original information was not present and guessed using any method is the definition of upscale.
@tdkmillsy: You'll be able to tell the difference even on 1080p tv, thanks to supersampling, you'll get better IQ than normal 1080p even on 1080p screen, obviously not as good as it would look on screen with the rendered resolution but still better than your normal 1080p
Yes it is lol. I've seen some really good 4k TV offers here, you could buy a 55" 4k LG at 599 euros ($668). Might be getting one sooner rather than later.
Log in to comment