Steve_Tk
Senior HTF Member
- Joined
- Apr 30, 2002
- Messages
- 2,833
I'll never get permission to buy a new iPad every year 
Home Theater
Entertainment & Streaming Content
Physical Media
Home Theater Equipment and Hardware
Other Diversions
Bargains and Classifieds
Home Theater Forum
WSJ: iPad 3 with Retina Display early 2012 (1 Viewer)
- Thread starter Ronald Epstein
- Start date
Kudos to them if they pull this off. The issue won't be about just how dense the screen it is; it's the processing power to move the pixels fluidly.
I'm wondering about cost. Will this be the new product standard, a la iPhone 4? Or is it such a leap in manufacturing difficulty that it will be a higher-priced option, with the rest of the line using the current screens? An iPad 3 Pro alongside the iPad 3?
It will definitely be a coup; there are $350 video cards on the PC & Mac side that can't get full framerate (30+) consistently accelerated. The idea of cramming a 2k accelerated tablet.. I root for them to pull it off, it would be an amazing accomplishment. The Macbook Air, with a much larger screen is capped at a much lower resolution (in fact, all of Apple's laptops have a lower screen resolution with more horsepower)
Interesting.
Interesting.
My four year old work laptop is driving 3360 x 1050 (dual 20" monitors) right now without a problem on a venerable WinXP system. Isn't 2D acceleration old hat, which is the bulk of iOS use? It's just about having enough RAM onboard.
It's 3D games that will be a problem. Apple might cheat, giving 3D GPU acceleration at the current resolution with pixel doubling to get it on screen. Or if the next system gives another 7x increase over the current graphics system, there may be more than enough power for a 4x increase in pixel count.
It seems like the hardest part is screen manufacturing, not GPU.
It's 3D games that will be a problem. Apple might cheat, giving 3D GPU acceleration at the current resolution with pixel doubling to get it on screen. Or if the next system gives another 7x increase over the current graphics system, there may be more than enough power for a 4x increase in pixel count.
It seems like the hardest part is screen manufacturing, not GPU.
It's 3D and Video acceleration that will be the kick in the ass. Accelerating 2D is nothing because much of the content is stagnant.
/t/314060/wsj-ipad-3-with-retina-display-early-2012#post_3875013
What would the ramifications to battery life be with that kind of processing power?
Here's the issue.
An iPad2 has a total of 1024x768 or 786,432 pixels to move.
An iPad3 with this resolution, would have 2048*1536 or 3,145,728 pixels to move.
So, by default, that would require four times the GPU memory space, just for starters. Add into that that all previous apps would have to upscale to new resolutions fluently or, older apps would have to downset the resolution - which is a more likely option. So, let's assume the resolution is configurable and that apps can set it high or low; in a low app mode (the traditional 1024x768) battery life would be pretty darn good; in high resolution mode..
Also, while pixel density is a great factor, for many users it's also a non-factor; open large text is the grounds of Apple's fastest growing audience.. not kids, the elderly. Those in the 65+ are adopting iPad like gangbusters thanks to simplicity.. but higher resolution is not as beneficial as larger and more fluid writing.
And while 2048x1536 sounds hella impressive, we also have to think about real world application usage. Apple is not going to suddenly start selling 2k resolution videos from studio; at their current weighted bandwidth (which is primarily 720P anyway) you'd be talking about moving more then 180Mb/s of raw data.. that would tax even a macpro on the I/O component, and the storage space for content would be immense.
The concept of it is very cool - and maybe they will do it, but the logistics of doing it, as well as the real benefits will be interesting to see how they do it and maintain a cost and a return that are worthwhile. People make the comparison to the iPhone4.. well, realize, the iPhone4 at retina on a 3.7" display is 600k pixels (roughly) which is less then the iPad2 NOW. It would be almost 5X less pixels then a retina ipad. That's a monumental leap. And those devices when they first came out were heavily subsidized.. a WiFi iPad3 with a retina display would not be subsidized. And unlike Amazon, Apple isn't losing money on a device to sell it.
I'm not saying this is impossible - it may happen, and it would be incredible and I would snag one. But the potential drawbacks and realities of doing it are huge.
An iPad2 has a total of 1024x768 or 786,432 pixels to move.
An iPad3 with this resolution, would have 2048*1536 or 3,145,728 pixels to move.
So, by default, that would require four times the GPU memory space, just for starters. Add into that that all previous apps would have to upscale to new resolutions fluently or, older apps would have to downset the resolution - which is a more likely option. So, let's assume the resolution is configurable and that apps can set it high or low; in a low app mode (the traditional 1024x768) battery life would be pretty darn good; in high resolution mode..
Also, while pixel density is a great factor, for many users it's also a non-factor; open large text is the grounds of Apple's fastest growing audience.. not kids, the elderly. Those in the 65+ are adopting iPad like gangbusters thanks to simplicity.. but higher resolution is not as beneficial as larger and more fluid writing.
And while 2048x1536 sounds hella impressive, we also have to think about real world application usage. Apple is not going to suddenly start selling 2k resolution videos from studio; at their current weighted bandwidth (which is primarily 720P anyway) you'd be talking about moving more then 180Mb/s of raw data.. that would tax even a macpro on the I/O component, and the storage space for content would be immense.
The concept of it is very cool - and maybe they will do it, but the logistics of doing it, as well as the real benefits will be interesting to see how they do it and maintain a cost and a return that are worthwhile. People make the comparison to the iPhone4.. well, realize, the iPhone4 at retina on a 3.7" display is 600k pixels (roughly) which is less then the iPad2 NOW. It would be almost 5X less pixels then a retina ipad. That's a monumental leap. And those devices when they first came out were heavily subsidized.. a WiFi iPad3 with a retina display would not be subsidized. And unlike Amazon, Apple isn't losing money on a device to sell it.
I'm not saying this is impossible - it may happen, and it would be incredible and I would snag one. But the potential drawbacks and realities of doing it are huge.
3D games is the last thing I think about regarding a new Retina Display (since I'm not playing them). After using the iPhone 4 for six months, the iPad's screen was noticeably inferior. Compared to the Kindle, the iPad is, if not worse, not better for readability. As on the iPhone, a Retina Display will improve every single app with text: Safari, Calendar, Contacts, books, magazines, Safari, the Calendar. And also games. While it might not be noticeable to those in their 60s, it will be a clear improvement to everyone else.
It will also be a *huge* marketing advantage over Android tablets for the next six to 18 months. (They're only now catching up with the iPhone 4's screen, 18 months later.)
Certainly there are technical challenges, not least of which is simply manufacturing a panel that large with that pixel density in the volumes necessary.
Apple's going to do a Retina Display for the iPad. It's only a question of when. iPad 3 in 2012 is looking better and better for it to happen.
It will also be a *huge* marketing advantage over Android tablets for the next six to 18 months. (They're only now catching up with the iPhone 4's screen, 18 months later.)
Certainly there are technical challenges, not least of which is simply manufacturing a panel that large with that pixel density in the volumes necessary.
Apple's going to do a Retina Display for the iPad. It's only a question of when. iPad 3 in 2012 is looking better and better for it to happen.
Like I said, it's easy when you talk about the Kindle, where E-Ink has a totally different means of viewing it so the optics are different. But in the case of an pixel-dense screen here, the jump to double resolution is a monumental one. We'll see. We have to remember most iPad apps are written around 1024x768. The move up will either mean upscaling (significant performance hit) OR it will mean that the screen flips into a 1024x768 resolution mode as I noted above (a non-native resolution). That has mixed impact also.
One of the bigger things here is that even 64Gb will seem paltry to sustain the resolution and cache rate. You'll need to look at 64/96/128Gb as preferred to handle the data requirements here.
I'm not saying I'm not rooting for them to pull this off, but there are a lot of real whips and arrows in this jump that don't exist in doubling a 3.7" screen's resolution. Then again, I'm reminded that up until the day of, MacRumors and elsewhere were posting "stats" on an iPhone5 with "sourcing" that turned out bogus (reshaped bezel! new home button! bigger screen size!)
One of the bigger things here is that even 64Gb will seem paltry to sustain the resolution and cache rate. You'll need to look at 64/96/128Gb as preferred to handle the data requirements here.
I'm not saying I'm not rooting for them to pull this off, but there are a lot of real whips and arrows in this jump that don't exist in doubling a 3.7" screen's resolution. Then again, I'm reminded that up until the day of, MacRumors and elsewhere were posting "stats" on an iPhone5 with "sourcing" that turned out bogus (reshaped bezel! new home button! bigger screen size!)
This is (expected to be) a double res screen. It's a reasonable guess it will work just like the iPhone: apps will be pixel-doubled if not operating at native res. That's not computationally expensive. Raw pixel count aside, these are the same concerns as with the iPhone: larger art assets, more memory, etc. But it's not been a problem for the past 18 months. Look at your apps: aside from 3D games and GPS, they're small. Ignoring TomTom, my 7th largest app, and the first non-game in the list, is Numbers at 160 MB, a universal app. Most apps are about 50 MB -- do you they are primarily iPad art and will quadruple in size?
...But it seem like it will require an interim fork of ios 5/6 before the usual big release at the next iPhone. A very interesting product release.
I still think the earlier iPhone leaks had a some ground in reality. It just wasn't a prototype of this year's, but next's. Will be interesting to see what comes around then. Matt has provided good reason's for the caution against getting up hope for a doubled iPad screen rez (remember it's doubled on both dimensions so 4x the pixels!). This confuses people easily, remember that a jump from a 4mp to an 8mp camera isn't doubling the resolution. it's only going a quarter more. I think Apple can get there, but it's going to seriously strain a lot of stuff, 2012 might simply be too early yet. And dave, 3d performance might not matter to you but I assure you it is to Apple. The games performance of first two gens of iPad were big selling points for it, both 2d and3d and there is no way you can just drop that in the next release. It's likely that the next iPad will compete directly with Kal-El which will be used in Android tablets. It might be useful for you to see the performance that it is aimed at: http://blogs.nvidia.com/2011/05/project-kal-el-demo-previews-future-of-mobile-gaming/ http://www.engadget.com/2011/05/29/nvidias-quad-core-kal-el-used-to-demo-next-gen-mobile-graphics/ They are demoing on 1280x800 which is not all that big a leap.
Except it's not really pixel doubled. We get two trimmed up because we think "double res" it's actually quadrupled if they make this shift. While this seems like "oh, that's not that big" find anyone who renders in x264 video (think Apple MP4, HD, etc.) and then go ahead and quadruple the video stream output and maintain framerates. It's not nearly as easy as you'd think. The issue is far less on a 3.7" screen, which remember, has the same processor as the Ipad2 now. You can't expect to quadruple pixel movement or upscale factors and have this be "without processing cost". The processing cost would be significant, as Nvidia has already pointed out; their Tegra3 is one of the first that actually accelerates 1080P at 29.97fps (Video)
http://www.anandtech.com/show/5072/nvidias-tegra-3-launched-architecture-revealed/2
But most vendors are trimming it to 1280x800 for battery purposes, etc. (and also note, at 1080P, you're talking 40Mb/s..) So, wait, if the Tegra3 can handle 1080P full flow, couldn't A6 do 2048x1536? Ok, here's the difference:
1024x768 = current res (786432 pixels)
1920x1080=2,073,600 (nearly 3 times more)
2048x1536=3,145,728 (4 times more)
So, a 50% increase in pixels being moved.. also realize NO vendor is actually proposing anything with a 1080P flow to an inbuilt screen at this moment because the cost factor is not only significant, but the battery and capacity use factor is significant.
But wait, why isn't this the same as moving to a Retina on an iPhone4?
Ok, here goes:
iPhone4: 960x480 = 230,400
iPhone3GS: 480x240 = 115,200
230k pixels is meager on the same processor that for a pad: 1024x768 is 786k pixels. This is where the confusion factor hits. People say "well, that doubled! so did this so why.." Ok, here's the thing: when the iPhone doubled, it went from 480x240.. to 960x480.. notice something? One of the variables (480) stayed the same.
In order for Apple to pull this off, they will need a chipset that will move 50% more pixels and decode then Nvidia's Tegra3 (an accomplishment they have at no point found in the past). So, you get through that fact, and now you have to realize you're going to go through a major shift in HTML5 Video.. which will upconvert to match a page (see: Youtube). That stage will be at the GPU/processor level.
I will root for Apple to pull this off, because it would be a stunning, stunning accomplishment. The move from A5 to A6 (their first quad core) would be a HUGE jump in data pathway performance and performance in power usage and video power. It would be a leap like Apple has never at any point accomplished.
To give you an idea, the data flow to support the display would be significantly (and I mean SIGNIFICANTLY) more then the MacbookAir, who's high res 1366x768 (or 1 Million pixels) pushes a third of the pixels, and uses an outboard i7 HD controller for video acceleration.
For them to inhouse all of this in an A6 wouldn't just be a slash at other tablets; they would instantly become, almost overnight, the key support of the second strongest pitched CPU on the market, pushing Intel and making me wonder why they would use Intel in desktops/etc.
I say this seriously.. the IO transfers that it would have to support as well to do this would be such a major jump that for most common tasks, it would be foolish for them to not switch everything but MacPros over...
I don't know, the more I think about it, the more I think it would be a stunning, incredible accomplishment.
http://www.anandtech.com/show/5072/nvidias-tegra-3-launched-architecture-revealed/2
But most vendors are trimming it to 1280x800 for battery purposes, etc. (and also note, at 1080P, you're talking 40Mb/s..) So, wait, if the Tegra3 can handle 1080P full flow, couldn't A6 do 2048x1536? Ok, here's the difference:
1024x768 = current res (786432 pixels)
1920x1080=2,073,600 (nearly 3 times more)
2048x1536=3,145,728 (4 times more)
So, a 50% increase in pixels being moved.. also realize NO vendor is actually proposing anything with a 1080P flow to an inbuilt screen at this moment because the cost factor is not only significant, but the battery and capacity use factor is significant.
But wait, why isn't this the same as moving to a Retina on an iPhone4?
Ok, here goes:
iPhone4: 960x480 = 230,400
iPhone3GS: 480x240 = 115,200
230k pixels is meager on the same processor that for a pad: 1024x768 is 786k pixels. This is where the confusion factor hits. People say "well, that doubled! so did this so why.." Ok, here's the thing: when the iPhone doubled, it went from 480x240.. to 960x480.. notice something? One of the variables (480) stayed the same.
In order for Apple to pull this off, they will need a chipset that will move 50% more pixels and decode then Nvidia's Tegra3 (an accomplishment they have at no point found in the past). So, you get through that fact, and now you have to realize you're going to go through a major shift in HTML5 Video.. which will upconvert to match a page (see: Youtube). That stage will be at the GPU/processor level.
I will root for Apple to pull this off, because it would be a stunning, stunning accomplishment. The move from A5 to A6 (their first quad core) would be a HUGE jump in data pathway performance and performance in power usage and video power. It would be a leap like Apple has never at any point accomplished.
To give you an idea, the data flow to support the display would be significantly (and I mean SIGNIFICANTLY) more then the MacbookAir, who's high res 1366x768 (or 1 Million pixels) pushes a third of the pixels, and uses an outboard i7 HD controller for video acceleration.
For them to inhouse all of this in an A6 wouldn't just be a slash at other tablets; they would instantly become, almost overnight, the key support of the second strongest pitched CPU on the market, pushing Intel and making me wonder why they would use Intel in desktops/etc.
I say this seriously.. the IO transfers that it would have to support as well to do this would be such a major jump that for most common tasks, it would be foolish for them to not switch everything but MacPros over...
I don't know, the more I think about it, the more I think it would be a stunning, incredible accomplishment.
/t/314060/wsj-ipad-3-with-retina-display-early-2012/30#post_3875133
Top posting for the lose. I thought you were replying to me.
I agree it's hard, but think the difficulty lies in the display panel much more than the GPU. And the indications are they're getting the production in hand. Misc thoughts. I understand its really pixel quadrupling. But we speak in linear dimensions for convenience. A5 is 2x - 4x the Tegra 2. Why not predict A6 will likewise best Tegra 3? http://techztalk.com/techwebsite/03-19-11-apple-a5-processor-smokes-nvidia-tegra-2-chip 1080p deciding is hard. Subsequently pixel doubling (quadrupling) is computationally trivial. The hard part is the memory bandwidth. A decade ago my PIII Win98 system ran dual monitors at half or better the res were talking on. A five year old Xbox 360 does 65% the pixel count in 3D. A four year old laptop drives 115% of the pixels (dual 20" monitors) We've been hearing informed rumors on this happening for a year. Could be wrong or premature. But I'd guess Apple is prioritizing this heavily. Huge marketing win and lead if it's here in 2012. Making the display: very hard. Infinity Blade at 4x 1024x768: really hard Battery life & power consumption: probably pretty hard, but seems in reach of even the A5. But pushing the pixels for 2D apps at these pixel counts has been doable by middling PCs for years. And which is it geriatrics who don't care about resolution or games? Or gamers who care about nothing but high res 3D acceleration?
geriatrics who don't care about resolution or games? Or gamers who care about nothing but high res 3D acceleration?
Both.
Both. In regards to the "well, my XBOX360 that's years old does.." Well, there are three parts to that.. when the XBOX shipped, it used one of the better graphics on the market and sold for a loss. It's the equivelent of a Geforce4TI series.. which by todays standard is antiquated, but *shrug*.. you also have to remember that it puts out 720P/1080I, not 1080P. So, 1280x720=921k pixels; a third of what this is going to put out; and unlike an iPad, it was able to store static objects on a DVD which are pre-rendered/display content. The iPad3 could do that too - that would be easy to scale, except that requires space. A fair amount of space. And while it does video rendering, the XBOX streams at low bitrate - which is OK for 720P, would be complete crap for 2k.
You're right. pushing the pixel count for a 2D app - let's say, DrudgeReport, Engadget, Accuweather, whatever... nothing at all to it.. easy as pie. Infinity Blade? That's an issue. H264 video.. issue.
Like I said, it'd be a major accomplishment. I'm rooting for them to do it. But I'd stand by my thought that if they pull it off, a SOC A6 that can do it, they might as well divorce intel and move all of their macbooks over.
You're right. pushing the pixel count for a 2D app - let's say, DrudgeReport, Engadget, Accuweather, whatever... nothing at all to it.. easy as pie. Infinity Blade? That's an issue. H264 video.. issue.
Like I said, it'd be a major accomplishment. I'm rooting for them to do it. But I'd stand by my thought that if they pull it off, a SOC A6 that can do it, they might as well divorce intel and move all of their macbooks over.
Well they DID buy Intrinsity and PA Semi. Is that enough to propel them past the Intel juggernaut plus AMD, Nvidia, and everyone else who has their hands deep in SOCs? I'd love to believe it, but it will so fundamentally affect the rest of the production pipeline and we've heard no real rumors on what their chips are cooking up into, I'm dubious but excited.
Marco Arment noted on Build & Anyze that Instapaper has iPad 3 retina display assets to be ready day 0. 10MB universal app. The free version, 2 versions old, is 2.4 MB. One data point on possible storage creep. IPad 2 can do 1080p 3D games, from what I can find "Firemint has implemented 1080p native resolution on a mobile video game. Last week, Real Racing 2 HD was updated to support full HD video output using the new digital video HDMI dongle Apple released alongside iPad 2." http://www.eurogamer.net/articles/digitalfoundry-real-racing-2-hd-1080p-comes-to-ios The Verge video review suggests the Transformer is doing 1080p 3D output as well. It seems today's hardware can do 65% of a iPad 3 retina display, in terms of math and pixel output. Maybe it's 25%, thinking about bigger textures, graphics. It looks like we're not that far from doing 2048x1536 on mobile, 2D and 3D.
Similar threads
