tablets to replace PCs?
- Thread starter smr
- Start date
I was responding to the gentleman who stated "in our lifetime". I think we will get mobile GPUs half as powerful as desktops within the next 10 years, simply because they will progress at a much faster rate than current PC technology.
New tablets displays in 2013 and 2014 will be 1080p as standard... it won't be as far away as you think.
Again, as I said, the lower cost is due to the far greater amount they will sell. If 10 million buy it at £10, thats better than 2 million at £30.
Indeed, but it's unecessary imo (used one, wasn't bowled over), as well as being a 4:3 resolution. I'll be happy for Apple to stick to that kind of thing, and for the Android market to output high quality 1080p LED and OLED screens.
Something along the lines of "But 2048x1536 is widescreen you dolt", I imagine. Before you remembered it wasn't.
Something along the lines of "But 2048x1536 is widescreen you dolt", I imagine. Before you remembered it wasn't.
Not quite. In fact nothing to do with you being wrong.
Was basically going to say most win8 tablets are going to be full HD as well. As they have to be at least 1366 wide, then realized that's a fair bit below 1080p, although details so far suggest most are going to be full hd anyway.
I was only pulling your leg mate.
Yeah, Win 8 tablets will be 720p minimum I think. Not interested in them personally though.
They might slightly narrow the gap, and may reach 50% of todays GPU's in 10 years, but they won't be at 50% of desktop GPU's, not even close, doubt they'll even hit 25%.
Yes they'll be 1080p, but there's a big difference between rendering a game at resolution x and upscaling to actually rendering at 1080p, which is what PC's do now (as well as some console games) and by the next couple of years all console games. Mobile GPU's simply aren't powerful enough to do complex graphics at 1080p, it will either be massively simplified and/or lower res and upscaling...
Possibly, but we'll see, I'm not sure if the publishers after putting in large sums of money to generate a console quality game (mostly the graphics seems to be the time/cost consuming thing so possibly a moot point) releasing that cheaply in the hope of large sales, they might just stick with what they know and sell for a premium.
And the games, even though it has by far the most powerful GPU currently available (~39gflops compared to the Tegra 3's 7 or so...), run at a much lower res and are upscaled, I believe most if not all still run at the 1024x768 res of the iPad 2, and will do long term, the GPU isn't capable of good quality graphics at that res, some pc graphics cards struggle at that res and they're easily into the 1000+ gflops now...
There are loads of games that run full retina. And not just angry birds.
But not overly well if they are complex 3D, as that article he linked shows.
'Loads', not loads of graphically demanding/impressive games by any stretch, some exceptions to the rule, although when comparing to the consoles/pc's they lack in the visuals department (whilst still being very impressive for such a limited device)
And yes, I realise that the situation is pretty much the same with consoles, it's rare there to find games that run at full 1080p, which considering even they are many times more powerful than the mobile chips just underlines the point. Note that I was specifically mentioning next-gen consoles when talking about their ability to do proper 1080p.
Nvidia seem to disagree with you about the potential of mobile GPU's:
Mobile GPUs to be on-par with consoles in a couple of years? NVIDIA says it’s possible: http://www.androidauthority.com/mobile-gpu-performance-evolution-77655/
Processing power of current consoles. And that's only in 2 years.
Aren't Nvidia the ones who, you know, make graphics cards on both PC and mobiles and stuff?
I think 50% as fast as PC's is a definite possibility within 10 years or less. And again, I think the PC platform is very overpowered and sadly underutilized... you don't need as much power as a top PC to make a game look good if you have a great engine and optimise the code well enough.
Last edited:
I think Steve Jobs said it best in one of the AllThingsD talks where he said tablets could be thought of as cars and PCs more like trucks. Tablets will be for everyday use with PCs reduced to specialist tasks. I think the next 10 years or so will so that he's bang on the money with this one.
No, there really are fps type games, very graphic intense that run at Regina nativ resolution. As well as several racing sims.
As the graph above shows. Mobile CPU/Gfx has come a long way. Consoles has stopped due to their nature and even pcs arent increasing like th user to be.
As for that one article, incredibly one sided.
Mobile GPu power in the iPad has basically been doubling with each new release, if that trend continues, 2015 iPad will be somewhere bang in the middle between ps3 and xbox360. So the ipad4 will easily handle the retina display.
http://www.industrialtoys.com/mobile-vs-console-the-war-heats-up/
Last edited:
Nvidia seem to disagree with you about the potential of mobile GPU's:
Mobile GPUs to be on-par with consoles in a couple of years? NVIDIA says it’s possible: http://www.androidauthority.com/mobile-gpu-performance-evolution-77655/
Processing power of current consoles. And that's only in 2 years.
Aren't Nvidia the ones who, you know, make graphics cards on both PC and mobiles and stuff?
I think 50% as fast as PC's is a definite possibility within 10 years or less. And again, I think the PC platform is very overpowered and sadly underutilized... you don't need as much power as a top PC to make a game look good if you have a great engine and optimise the code well enough.
That graph has an amusingly stupid scale, it also effectively states that in 2014 mobile chips will be as powerful as consoles, which tend not to use top-end hardware even at release, which were released in 2005. Wow, isn't that impressive, or not...
It might also be worth noting that they failed to gain any contracts for the new consoles I believe, so they may be ever so slightly biased towards a market they're actually competing in.
By the time mobile chips, apparently, reach the lofty levels of mid-high-end 2005 the next gen consoles are likely to have been out for 6-12 months and will be many times more powerful again.
PC graphics cards are huge chips that cost $50-100 per chip and can use 150W+ of power. Consoles are cheaper per chip, and more limited due to designs but probably in the region of 60-80W. Then come mobile devices where the entire SoC needs to run <10W and cost $10 or less. At any point if you compare current mobile to consoles and pc's it will lose out drastically, closest it will come is right at the end of the long and static console lifespan where the hardware becomes seriously outdated.
Mobile chips are becoming ever more efficient, and will continue to ramp up, but as long as PC gaming exists (and it will not die imo) then PC's will be many many times more powerful.
Of course even their own graph kinda proves my point, current consoles can't create highly detailed graphics at 1080p, yet we're 2 years away from mobile chips reaching even that point...
I agree with you on PC's being underutilized, having 2 major graphics manufacturers, multiple OS's and other peripherals definitely hits the performance that's actually available to games, and this is why Consoles traditionally provide a better experience at the start of their lifetime even with seemingly sub-par hardware, but that then tails off as hardware for PC's continues to evolve and surpass that of the consoles, even taking into account the inefficiencies.
And will mobile devices really be great for that? Android isn't going to win any efficiency awards, iOS is a bit better perhaps but still not at the level of consoles, I'd guesstimate that Windows 8 is going to sit somewhere between the two. iOS has the benefit of small hardware variation (currently all PowerVR 5 series so optimising will be very easy), Android is obviously fragmented in that respect so optimisation will be very hard, and windows 8 is again fairly open isn't it so another problem. Will they even manage to be optimised better than PC code?
You mean this galaxy on fire 2 game that recently removed support for full retina resolution in favour of 1440x1080:
http://forum.fishlabs.net/viewtopic.php?f=29&t=3003
oops...
Aha, nice edit
Reading the article now, brb
EDIT:
Read it,
Not sure I agree with him, he essentially basis his argument on a continued doubling of performance of the iPad every year, which would be done using the traditional method of throwing more transistors at the problem, which makes sense especially as the PowerVR chips are already very efficient so although there are some gains to be made there the easiest way will be more cores as always.
He also claims this can be done via increasing die size, which is again understandable as that is the method Apple have used to date:
A4 - 45nm - 53.3mm2
A5 - 45nm - 122.2mm2
A5X - 45nm - 165mm2
It's also worth nothing that there is a 32nm version of the A5 and that die shrink saved 40% down to 69.6mm2.
Now the A6 will almost certainly be on Samsung's 32nm process, although there is also the possibility of TSMC at 28nm, it's unlikely they can use Intel who are the current leaders on 22nm. So assume 32nm, which gives us a 40% reduction of die space and that the easiest (read the cpu cores and memory controller being identical) doubling still cost them ~35% we can see that they can double the performance again at 32nm keeping within the 165mm2 of the A5X, great news.
The next one will likely be on the same process, and hence be another larger die, more cost, more power, more heat, how long can that last?
The iPad 3 itself did suffer from both weight and battery life due to the larger die, how much more can it take, and what about the increase in die size due to improved cpu architectures, hardware decoders, image processors and memory controllers?
But then the problem starts, it takes Intel who are easily the best at moving on the lithography processes ~4 years to get volume production on a new process, Apple can double performance easily every process change, but the rest are tricky, and although they have some scope for catching up in process terms that's not going to last long.
I don't see it happening beyond the next 2-3 iterations, it will improve year-on-year but a full doubling will be a much rarer occurrence.
Last edited: