Post Xbox One Two Scorpio, what should Sony do next? *spawn* (oh, and Nintendo?)

[KOF]

This misconception about "Human eyes can't resolve 4K resolution in distance." is born from several misguided assumptions. SMPTE has calculated and made such claim and viewing distance recommendation based on several factors.

- used the same way Dr. Hermann Snellen has used in his famous eye chart to calculate vision acuity
- used regular SD/HD broadcast
- does not take into account video processing variable

The problem with using Snellen's method is that his eye test is performed with stationary target. Real video contents do not. Not only that, but current displays behave very differently from real world when resolving moving objects. What we conceive as motion on displays aren't really motion. They are just pixels turning themselves on and off, with flickering in between. Real life moving lines do actually move continuously, but on displays, they do not because they are only comprised of array of fixed pixels, and since some displays have huge gap between pixels, (low pixel fill rate) and some do not, we can't make a uniform conclusion based on very few TVs and projectors samples.

In Snellen's eye test, static acuity indeed hold majority of share, but when starting to resolve real life motion, dynamic acuity begins to be important as well, and when such tests are done on displays, importance of dynamic acuity increases even more because of how display's pixel array structure is quite different than that of real world. This point has been raised by Pixelworks in SID and they've made two conclusions.


1. “When the sampling-phase effects are taken into account, the perceivable pixel density limit increases from 60 ppd to nearly 90 ppd”.

2. “Common video processing techniques can cause the fixed- pixel structure of the display to be visible regardless of the display resolution. Fortunately, alternative approaches are available that can greatly reduce the effect of fixed- pixel sampling, as will be discussed in the next paper in this series”.


Another thing to consider is pixel quality, because not all pixels are born equal. Even among 2160p/4K materials, there are vast differences, ranging from 8 terabytes of Hollywood RAW camera capture all the way down to Youtube 4K. Even some of Netlix's 4K materials aren't really 4K. Breaking Bad for example, has been discovered to be only using vanilla HD Bluray, upscaled to 4K. Heck, the Netflix version actually loses out to the 1080p Bluray original, despite having 4 times the resolution, how? Because the Netflix 4K version has less data than the BD original. Well...not quite, because Netflix uses HEVC(H265). Since HEVC is said to bring at least 40% compression improvement to the H.264, (which current HD BD uses) compression efficiency should have covered small bitrate deficiency, but it didn't, why? One Bluray compressionist actually makes a claim all Netflix compressions are automated due to huge catalog, while for Blurays, painstaking care is given to each and every titles.

So, now we have two more variables: bitrate, and compression quality. Unfortunately, a video guru, Stacey Stanley (president of Stencil & Muncil/ the guy who has done work on Xbox One's AV software interface) makes a remark that such 4K upscales are actually detrimental to compression efficiency because output resolution is compressed as 4K. He uses Martian UHD Bluray as an example. That movie's DI (Digital Intermediate) is only 2K, but upscaled to 4K so that they won't be litigated when every one of UHD Bluray discs already have "4K UHD Bluray" logo.

So, if we were to use 4K upscales such as Netflix 4K Breaking Bad as testing materials, then nobody will be able to spot picture quality improvement between HD and UHD displays because it uses HD Bluray as a source. Change the testing material to Apple ProRes 4:4:4 DI that Hollywood uses and many people will change their tune. And games are especially different from movies because games already use the most pristine uncompressed RAW quality Bluray movie collectors can only dream of. I've seen my share of reference quality Bluray movies on my plasma TV and every one of them did not fare well next to games. While not HDR, even among SDR titles, there are quite a big difference with dynamic range, (my plasma TV has very good contrast ratio to actually allowing me to distinguish) and on average, games have far superior dynamic range than movies because games do not have real world constraints such as movies (like the choice of movie set, weather, lighting condition, etc) So when people say "People can't distinguish between 1080p and 4K on such and such distance", you should understand such claim is only intended for video contents and never for games.

 

[Rangers]

I'm not sure why MS is only targetting 6tf to be honest considering we are 18 months away from the release window. Reports suggest that the Neo is using tech that can achieve figures in this range now.

If you go by DF's analysis/other things known/reported, Neo is using 36 CU's (Polaris 10 whatever it's called in RX 480, releasing soon), Scorpio is using 60 (AMD's big 64 CU part Vega, releasing later). Plus a 384 bit bus vs 256. They dont really seem in the same ballpark. And for all the talk sony can do this and that, as long as they're releasing first MS has the "late upgrade" upper hand. The reason the relative flops are lower than the CU difference is likely just down to clocks. (speculated ~800 for MS part, 911 for Sony's).

For Sony to increase flops just to the 5.5 rumored of the desktop 480 part, they'd have to ramp clocks unacceptably high for a console part. The desktop R 480 will likely be clocked at 1.1-1.3 ghz,

18 months away just shows there's a lot of lead time for a console. The Neo is supposed to be about a year away as well.

 

[dobwal]

I agree and think Scorpio is reasonably last minute. There have been reports of the Neo from March, even the specs were banded around from April onwards, so I'm sure MS were looking at their next console, but maybe stepped it up after hearing these reports.

There have been rumors regarding two new Xbox devices swirling around since January. I hardly doubt it was a last minute decision.

They wanted hardware that could current gen titles and play it 4K, which is probably why they choose a late 2017 release as 6 Tflop gpu will be cheaper to source after AMD is well within the production cycle of its next gen tech.

Furthermore, I doubt Neo is slated for this year as Sony said it didn't have anything to show. A Neo launching this holiday season should have software well within its development cycle as you need titles that will work with Neo at launch.

 

[temesgen]

If you go by DF's analysis/other things known/reported, Neo is using 36 CU's (Polaris 10 whatever it's called in RX 480, releasing soon), Scorpio is using 60 (AMD's big 64 CU part Vega, releasing later). Plus a 384 bit bus vs 256. They dont really seem in the same ballpark. And for all the talk sony can do this and that, as long as they're releasing first MS has the "late upgrade" upper hand. The reason the relative flops are lower than the CU difference is likely just down to clocks. (speculated ~800 for MS part, 911 for Sony's).

For Sony to increase flops just to the 5.5 rumored of the desktop 480 part, they'd have to ramp clocks unacceptably high for a console part. The desktop R 480 will likely be clocked at 1.1-1.3 ghz,

18 months away just shows there's a lot of lead time for a console. The Neo is supposed to be about a year away as well.

If Neo is a year out I'm not sure it makes sense to bother, release should be within 6 months of VR IMO.... Any longer and might as well wait for PS5.

 

[RancidLunchmeat]

If Neo is a year out I'm not sure it makes sense to bother, release should be within 6 months of VR IMO.... Any longer and might as well wait for PS5.

A couple of things here, first of which is that Sony had no idea MS was going to do a generational leap that they did with the Scorpio. That's the first bit.

The second bit is that Sony doesn't have PSVR ready yet. There's no release date on that. This is a problem for Sony because they want to release and sell the PSVR to their current user base (PS4 owners), despite the fact that they know that the PS4 doesn't provide adequate power for the PSVR.

What Sony wants is for PS4 users to buy the PSVR for $399 and think that "this is kinda cool, only if it was a bit better.. frame rates, etc", and then 6 months later reveal the NEO that will improve all of that. What is your choice as a PS4/PSVR user at that point? You've already invested $399x2. Do you invest another $399 into NEO or do you give up everything and switch platforms?

Sony is trying to make sure that PS4 owners have too much "skin in the game" after buying a PSVR to switch platforms.

But Sony didn't expect that MS would just jump an entire generation and be ready to go Holiday 2017, so now they are in quite a pickle. They can't announce the NEO before the PSVR has actually lauched because it will alienate all their entire current user base. The closer the PSVR and the NEO launch together the more people realize it's not just $399 to buy the PSVR, it's $399 plus the $399 for the NEO. Because that's what you need if you actually want the thing to work.

And, of course, MS has said at E3 with their Developer interviews that the 4.4 Tf the NEO supplies still isn't enough.

I don't see how anybody doesn't see the corner that Sony painted themselves into, but the reason that MS isn't in that same corner is because MS doesn't have a VR solution.

 

[temesgen]

A couple of things here, first of which is that Sony had no idea MS was going to do a generational leap that they did with the Scorpio. That's the first bit.

The second bit is that Sony doesn't have PSVR ready yet. There's no release date on that. This is a problem for Sony because they want to release and sell the PSVR to their current user base (PS4 owners), despite the fact that they know that the PS4 doesn't provide adequate power for the PSVR.

What Sony wants is for PS4 users to buy the PSVR for $399 and think that "this is kinda cool, only if it was a bit better.. frame rates, etc", and then 6 months later reveal the NEO that will improve all of that. What is your choice as a PS4/PSVR user at that point? You've already invested $399x2. Do you invest another $399 into NEO or do you give up everything and switch platforms?

Sony is trying to make sure that PS4 owners have too much "skin in the game" after buying a PSVR to switch platforms.

But Sony didn't expect that MS would just jump an entire generation and be ready to go Holiday 2017, so now they are in quite a pickle. They can't announce the NEO before the PSVR has actually lauched because it will alienate all their entire current user base. The closer the PSVR and the NEO launch together the more people realize it's not just $399 to buy the PSVR, it's $399 plus the $399 for the NEO. Because that's what you need if you actually want the thing to work.

And, of course, MS has said at E3 with their Developer interviews that the 4.4 Tf the NEO supplies still isn't enough.

I don't see how anybody doesn't see the corner that Sony painted themselves into, but the reason that MS isn't in that same corner is because MS doesn't have a VR solution.

My point is NEO doesn't have a purpose really outside of PSVR, ideally you'd have both launching around the same time would make the most sense at least in my opinion.... Regarding 4.4Tf being enough, I think that kinda depends on your solution it may not be enough for MS but I'm inclined to believe it is enough for Sony otherwise why target that spec?

 

[BRiT]

I'm inclined to believe it is enough for Sony otherwise why target that spec?

It's the cheaper cost to Sony option.

 

[wco81]

My point is NEO doesn't have a purpose really outside of PSVR, ideally you'd have both launching around the same time would make the most sense at least in my opinion.... Regarding 4.4Tf being enough, I think that kinda depends on your solution it may not be enough for MS but I'm inclined to believe it is enough for Sony otherwise why target that spec?

Well maybe 4K video.

Maybe Sony has some hopes for the UHD Blu-Ray format, since it's still selling TVs.

 

[temesgen]

It's the cheaper cost to Sony option.

Ok then why 4.4 instead of 4 or 3.4? I'm not saying your wrong bc I have know way of knowing but it seems to me that 4.4Tf is targeted to improve PSVR frame rate.

Well maybe 4K video.

Maybe Sony has some hopes for the UHD Blu-Ray format, since it's still selling TVs.

why would you need 4.4Tf for 4k video? Amazon Fire can do 4k video and its using a cell phone processor

 

[wco81]

You don't need that power but I'm saying the product brings 4K video in addition to better performance for VR.

I think more people are likely to use 4K video than pay hundreds for VR gear on top of what they pay for the console.

Unless some VR games completely redefine gaming and VR becomes a mass phenomenon.

 

[temesgen]

You don't need that power but I'm saying the product brings 4K video in addition to better performance for VR.

I think more people are likely to use 4K video than pay hundreds for VR gear on top of what they pay for the console.

Unless some VR games completely redefine gaming and VR becomes a mass phenomenon.

Fine but Brit's suggestion was cost, I would imagine it would be cheaper to simply put in a new HDMI board and Blu Ray drive than do anything with the GPU and CPU, my point is why would Sony invest money to target 4.4Tf if 4.4Tf wasn't a viable target spec? I get that there are trade-offs between power and cost but it makes zero sense to me to target a spec that doesn't meet the minimum requirement to achieve the desired result.

 

[wco81]

I'm not saying they're not interested in VR.

Obviously they're targeting higher performance for VR. They did commit to it years ago so they have to go through with it. Plus who knows, it may be the next big thing.

 

[iroboto]

Fine but Brit's suggestion was cost, I would imagine it would be cheaper to simply put in a new HDMI board and Blu Ray drive than do anything with the GPU and CPU, my point is why would Sony invest money to target 4.4Tf if 4.4Tf wasn't a viable target spec? I get that there are trade-offs between power and cost but it makes zero sense to me to target a spec that doesn't meet the minimum requirement to achieve the desired result.

A) price to the customer, what they can handle
B) cost to build
C) TDP
D) time to market

Those things together determine the flops.
It's 4.4 because it's going to come out before Scorpio. Ideally well before.


Sent from my iPhone using Tapatalk

 

[temesgen]

A) price to the customer, what they can handle
B) cost to build
C) TDP
D) time to market

Those things together determine the flops.
It's 4.4 because it's going to come out before Scorpio. Ideally well before.


Sent from my iPhone using Tapatalk

Again that doesn't address targeting a spec that won't deliver what is needed... It doesn't make sense to target 4.4 if 4.4 won't get you where you need to be. Cost TDP, time are factor into the decision but nobody would invest millions into a solution that is DOA bc its not viable, you'd wait for viable solutions and then enter the marketplace.

 

[Jubei]

1. PSVR is scheduled to release October 13th (EU) so it does have a release date, there will be demo units coming to stores in the upcoming months

2. MS saying 4.4 TF not being enough is based on what the Oculus team wanted, that doesnt mean 4.4 doesnt hit Sonys target for PSVR (assuming it was developed for that purpose)

3. PSVR doesnt work without Neo? Then how are they playing it on PS4 right now? It might not be up to Oculus standard but its also HALF the price

I dont get the impression that Neo was developed for PSVR, but rather like House said that they wanted to keep the enthusiast customers who switched to PC because they need something newer and better. And until MS says outright that Scorpio will get exclusives then they are doing the same thing. Industry is moving towards an iPhone like direction, enthusiast will always buy the latest tech but software will be targeted for the largest userbase wich is the older versions of the consoles

And until we see the price for both these consoles, we wont know wich was the right or wrong choice

What if Scorpio is $599? Vega probably wont be a 199 dollar GPU like Polaris. MS is going for their UWP strategy, wich means that even if you buy their game on PC, they still make money. That implies they wont go for a loss leader strategy with Scorpio

 

[Michellstar]

Fine but Brit's suggestion was cost, I would imagine it would be cheaper to simply put in a new HDMI board and Blu Ray drive than do anything with the GPU and CPU, my point is why would Sony invest money to target 4.4Tf if 4.4Tf wasn't a viable target spec? I get that there are trade-offs between power and cost but it makes zero sense to me to target a spec that doesn't meet the minimum requirement to achieve the desired result.

Don´t pay attention to the FLOPS, better to look at the elements of the soc.
They decided to double up the GPU cores (CUs), but they didn´t touch the 8 cores.
Orbis was about 348 nm², they could end up with a soc around the same size at 14nm (even smaller)

After that, it´s a matter of finding the combination of clocks that they are comfortable with the expected TDP of a console

I guess it´s how it went, lol

 

[davygee]

Firstly, Sony have stated that any games released from Q4 this year onwards need to have Neo compatible features, even though all PS4 titles will work on the Neo. So all the big hitters this year that come out in October/November/December will be Neo enhanced.

Lets not forget that the PS4 is a considerably more powerful console compared to the XBox One. Over the last few years we have seen the difference, which generally comes in with a resolution bump (ie 900p for XBox, while PS4 gets 1080p) or framerate, or both. The GPU jump from the XBox One to the Scorpio will more or less allow for the same visuals/effects but at a higher resolution (ie 900p on XBox One to 4K on Scorpio). We can see where the figures have come from for that. Why Microsoft have chosen a 6tf GPU to power 4K.

The Neo, which is around 2.5 times the power of the PS4 won't power 4K at the same graphic fidelity. So I doubt the Neo will be pushed as a 4K gaming machine. It may be able to show higher resolutions than 1080p upscaled to 4K with better graphics etc, but not straight 4K.

Does the Neo or any other console really need to do this anyway? I don't think so. 4K is still some time off. As discussed numerous times on here. You will need a huge TV to take advantage of it anyway, unless your face is pressed up upon the screen. Mass adoption of 4K, if it happens, is still at least a good 2-3 years off.

The PSVR is releasing mid October and can run on the PS4 as is. It doesn't need the Neo to operate. But what the Neo will do is allow for better graphics/effects within those games running on VR.

The Neo, as I see it, is meant as a mid-cycle refresh of the PS4 system. It is fully backwards compatible with the PS4. All games coming out over the next few years will still work well on the PS4, but will look better or play at a higher resolution on the Neo. What the Neo does, is gives the PSVR a better chance of surviving. Before too long, the difference in graphical fidelity will be noticed between the PSVR & Oculus/VIVE. What the Neo does, is closes the gap a little.

One thing that's not been mentioned is the cost. I still fully expect the Neo to release before Xmas this year and I expect it to release at a decent price. Certainly no more than $399, but what may actually happen is that Sony may very well do a launch event to announce the new PS4 Neo launching at $349 with the original PS4 getting a price cut to $249, also may actually see two bundles for Xmas each with a PSVR and will be competitively priced.

In short, the Neo is a mid-term/cycle upgrade for the PS4. The next release, more than likely in late 2018, early 2019 will be a bigger jump in my opinion.

 

[Shifty Geezer]

The Neo, which is around 2.5 times the power of the PS4 won't power 4K at the same graphic fidelity. So I doubt the Neo will be pushed as a 4K gaming machine. It may be able to show higher resolutions than 1080p upscaled to 4K with better graphics etc, but not straight 4K.

I think that's the better choice. The vast majority of gamers will likely have 1080p sets, and would rather have 1080p games in better quality/framerate than current quality at 4K (downsampled to 1080p). If Scoprio games are XB1 games at 4K, I think interest will be muted. However, if Scorpio games have 1080p modes that render in better visuals, it'd offer far broader appeal. Basically, is it a PC with settings (even if 1080p or 4K options in games) or a 4K XB1? Regardless, it's the 1080p market that these machines probably need to appeal to, and the one that does that best (ignoring current market share etc) is probably the one that'll sell most.

 

[Deleted member 11852]

Neo only increases memory bandwidth ~24%, delta compression at best offers up to 40% bandwidth savings in optimal synthetic conditions. Pretty much every game is pushing up against the bounds of the unified GDDR5 real world capabilities. These games brought over to the Neo are going to be either bandwidth bound, and thats even before examining the cpu or gpu bound.

I was wondering about this for Scorpio. We have PS4 with 1.84Tf and 140Gb/sec bandwidth and while lots of games run solid at their intended 60 Hz (CoD, Battlefield, MGS V) we're still seeing quite a lot of games fail, or just miss, their 1080p30 target. Scorpio sounds ambitious at 4K 60Hz which is 8x PS4's 1080p30 (4x pixels, double the framerate) and in terms of raw performance Scorpio is has is 3.2x the Tf and 2.4x memory bandwidth of PS4 so was curious how this marries to 4K60.

Which generation of GCN is producing a 40% saving on saving through delta compression? Obviously, technology movies forwards and this is only going to improve.

 

[Deleted member 11852]

If Scoprio

I'll add this to my list of misspelling of Scorpio. Sony were smart to use use an anagram of One. The next PlayStation my well be the 306. or 603 or 630. Yeah, take that conspiracy theorists.