Predict: The Next Generation Console Tech

Status
Not open for further replies.
I wonder if the role of hardware scalers increases for next gen consoles.

Xbox has a good hardware resolution scaler, but I wonder if there are any new technological developments and advances made in the last years which could be important for a gaming machine? For instance be prepared for even higher than 1080p displays which may be important in the next couple of years.

Besides hardware resolution scalers, I wonder if it is possible and if it would make sense to include a hardware frame interpolation scaler? We already saw interesting frame interpolation results based on software implementation from the Force Unleashed devs.
 
By the way. Nowadays which custom parts would make a 1,5 tflops GPU in a closed system behaves similar to a 3-4 tflops desktop GPU ?. The bandwith to VRAM being huge is something already made and we all are discounting for future systems. Some kind of fixed lighting hardware?. MS made the perfect custom decisions for X360 ( unified shaders, unified RAM and good chunk of EDRAM ) but i am failling in seing how they could make it this time to compete to high end PCs graphically...

So, what could make a custom GPU not to be the procession power so important?. ( i am wondering this after reading Bkillian´s post about tflops not being the main characteristics to look at ).
 
Last edited by a moderator:

I think the leaks (we had plenty of them so far) are not the problem but their credibility is. I am sure that in the time around PS Destination some new or old (confirming what others said) rumours will surface but except guessing and arguing about their plausiblity we won't get much more information out of them. If someone now calls a new and unknown AMD mobile GPU by its name and has the specifications and leaks that this is going to be in the PS4 and AMD presents that GPU weeks after that leak that would be a step forward. However I fear that most rumours will be very, very general because they are (educated) guesses.
 
By the way. Nowadays which custom parts would make a 1,5 tflops GPU in a closed system behaves similar to a 3-4 tflops desktop GPU ?. The bandwith to VRAM being huge is something already made and we all are discounting for future systems. Some kind of fixed lighting hardware?. MS made the perfect custom decisions for X360 ( unified shaders, unified RAM and good chunk of EDRAM ) but i am failling in seing how they could make it this time to compete to high end PCs graphically...

So, what could make a custom GPU not to be the procession power so important?. ( i am wondering this after reading Bkillian´s post about tflops not being the main characteristics to look at ).

On that note lherre made a positive post now (after I was getting a bit gloomy)

http://www.neogaf.com/forum/showpost.php?p=46145528&postcount=5459

Only one thing, you can't compare current retail gpu's with next gen console gpu's per se ... there is a lot of customs things inside them that make them incomparable as you are trying to make.

Not sure I totally buy that any customizations are going to completely make up for a lack of brute force, but I'll take it.
 
If the 720 is the rumored low level of power, than i doubt PS4 will be stronger to any significant margin. Sony is bleeding money as it is, they would not need to go far to stay competitive now that blu ray is no longer super expensive and they aren't using cell.

But i'm not on that level yet of thinking that they are both weak. So we have plenty of statements saying that the customizations will factor a lot into how the console performance. So what does that mean? Can we atleast confirm GCN2 architecture(assuming that those 8000 cards to come out this year)? That would probably boost the efficiency anyway.
 
By the way. Nowadays which custom parts would make a 1,5 tflops GPU in a closed system behaves similar to a 3-4 tflops desktop GPU ?. The bandwith to VRAM being huge is something already made and we all are discounting for future systems. Some kind of fixed lighting hardware?. MS made the perfect custom decisions for X360 ( unified shaders, unified RAM and good chunk of EDRAM ) but i am failling in seing how they could make it this time to compete to high end PCs graphically...

So, what could make a custom GPU not to be the procession power so important?. ( i am wondering this after reading Bkillian´s post about tflops not being the main characteristics to look at ).
From the articles on GPU API hinderance (PC environment), just dropping it in a closed box without DX; should give about that performance boost.
 
From the articles on GPU API hinderance (PC environment), just dropping it in a closed box without DX; should give about that performance boost.

That already happens with every console. Maybe you could say a 1 tflop gpu would perform like a 1,5-1,8 one for only being programmable to the metal. But it won´t be enough this time as you will have soon 5 tflops gpus in the PC space. If Steambox is a reality it will have also that advantage and probably it will launch with a top-end desktop like GPU or a SLI config. And don´t forget Power vr 6-Rogue in next Apple and Android tablets...TBDR make tiny gpus behave like bigger ones and once they reach 14nm 1 tflop in an Ipad wouldn´t be far from reality.
 
If the 720 is the rumored low level of power, than i doubt PS4 will be stronger to any significant margin. Sony is bleeding money as it is, they would not need to go far to stay competitive now that blu ray is no longer super expensive and they aren't using cell.

But i'm not on that level yet of thinking that they are both weak. So we have plenty of statements saying that the customizations will factor a lot into how the console performance. So what does that mean? Can we atleast confirm GCN2 architecture(assuming that those 8000 cards to come out this year)? That would probably boost the efficiency anyway.

I thought that most rumors were the other way around. At least when they questioned some of the game devs. They said that the neXtBox was a bit faster than the PS4. Oh well, what it all comes down to is the same every time, we don't know anything really.
 
On that note lherre made a positive post now (after I was getting a bit gloomy)

http://www.neogaf.com/forum/showpost.php?p=46145528&postcount=5459



Not sure I totally buy that any customizations are going to completely make up for a lack of brute force, but I'll take it.

Shades of Nintendo secret sauce?

If people want to claim additional capacity thru customizations they'll need to be more specific. EDram and console API efficiency? Those have already been accounted for. HSA? Need to be more specific and with at least estimates. Audio DSP's? Please... What else is there? Support in silicon for Accelerated Data Structures and super fast sorting? Would love to see it, and its the kind of thing that could make a 1+TF gpu comparable to a 3-4TF gpu, but again need details (or at least a link to a relevant white paper).
 
Shades of Nintendo secret sauce?

If people want to claim additional capacity thru customizations they'll need to be more specific. EDram and console API efficiency? Those have already been accounted for. HSA? Need to be more specific and with at least estimates. Audio DSP's? Please... What else is there? Support in silicon for Accelerated Data Structures and super fast sorting? Would love to see it, and its the kind of thing that could make a 1+TF gpu comparable to a 3-4TF gpu, but again need details (or at least a link to a relevant white paper).

Yeah I dont know either. Only thing I can think of is ESRAM stuff.
 
That already happens with every console. Maybe you could say a 1 tflop gpu would perform like a 1,5-1,8 one for only being programmable to the metal. But it won´t be enough this time as you will have soon 5 tflops gpus in the PC space. If Steambox is a reality it will have also that advantage and probably it will launch with a top-end desktop like GPU or a SLI config. And don´t forget Power vr 6-Rogue in next Apple and Android tablets...TBDR make tiny gpus behave like bigger ones and once they reach 14nm 1 tflop in an Ipad wouldn´t be far from reality.

Price?
 
But this price is attractive to PC gamers, not to console gamers. Just look at PS3

ok. But how many of the 70 million PS3 users paid 499-599 euros for the console?. Several millions. PC gamers would pay it also as it also carries Steam... I am sure in the Xbox camp there would be people also buying it, don´t forget Valve is an american company and more valuable for gamers than MS...
 
That's a gross simplification and generalization for that matter. We don't buy things because they are American. We buy them if they are priced right and are attractive for our budget. A steambox console if its out of my price range isn't going to fly no matter who is making it. There's a reason why Sony has been third place for 99 percent of this generation while bleeding money from their gaming division.

And then we have Valve, an infinitely smaller company that runs a digital game store.

They want to market that thing as a console? That's not going to work.

Microsoft knew that they had to come in early, and they had to come in cheap, and that's what Sony and Microsoft are both going to take to heart this gen. That's why we have so much effort into customizing these parts to get performance instead of just going with an off the shelf part.
 
Does DX(or windows) overhead really cost that much in games performance for PC's?

bit-tech said:
On consoles, you can draw maybe 10,000 or 20,000 chunks of geometry in a frame, and you can do that at 30-60fps. On a PC, you can't typically draw more than 2-3,000 without getting into trouble with performance, and that's quite surprising - the PC can actually show you only a tenth of the performance if you need a separate batch for each draw call.

http://www.bit-tech.net/hardware/graphics/2011/03/16/farewell-to-directx/2

Check out the performance drop in the charts of the link below.

http://beyond3d.com/showpost.php?p=1630487&postcount=24
 
Status
Not open for further replies.
Back
Top