• Competitor rules

    Please remember that any mention of competitors, hinting at competitors or offering to provide details of competitors will result in an account suspension. The full rules can be found under the 'Terms and Rules' link in the bottom right corner of your screen. Just don't mention competitors in any way, shape or form and you'll be OK.

next gen consoles

Seanspeed said:
Variable hardware is not cost effective. And the last thing it does is make it easier for devs to work on.

There's also some really big questions about a continued upgrade cycle rather than a 'base and hot' model. Continued upgrade cycle means that you're stuck with requiring backwards compatibility which means you're in a perpetual cross-gen development philosophy(which is never ideal) PLUS it means that you're forever held back by outdated tech. And not just in horsepower terms, but architectural and hardware features, which are a HUGE part of the graphical advances we see.

I think it makes more sense to do a base model and then a hot model. But then RESTART after that. Basically, have a PS4 and then a PS4 Neo. But then do a PS5 and start fresh. Do not do a Neo 2.

We, as PC gamers, would also benefit from a restart cycle rather than a constant evolution. As we always have. So keep that in mind before you root for regularly upgraded consoles.
Click to expand...

it's cost effective for the price of the chip, it's cheaper to make 2x300mm² or 3x200mm² chip than a 600mm² especilay when moving into smaller nodes, for console makers that is a game changer, because they will be able to deliver quite a punch, they had this problem they couldn't solve, PC moving too fast for consoles, it's not microsoft's lack of trying to slow it, but this is a game changer for them, and to me from the leaks everything indicate that already went this route.
it's not complicated for Devs because they do not change architecture, it's still the same hardware, just need a different way to communicate with 1,2, or 3, this can be solved on the software side, Devs would need to adjust but that doesnt mean it will give them much more work like crossplatform ports.
and ofc it wont be perpetual, for exemple ps4 and ps4 neo, equiped with 1 pitcairn and x2 pitcairn respectively, when PS5 shows up, it will another chip like polaris for exmeple.
this seem so logical to me, i dont get why some ppl sees it like far fetched!
and about thermals, that you seem to focus on, a simple liquid cooled box would allow the dissipation of 500tdp, while being quiet, and accommodated in small sized console.
 
Last edited:
This "multiple dies on an interposer" is just pie in the sky for now. People started talking about it lately like it's definitely happening, based on nothing. They get themselves all hyped up, it's funny to observe lol.
 
Orangey said:
This "multiple dies on an interposer" is just pie in the sky for now. People started talking about it lately like it's definitely happening, based on nothing. They get themselves all hyped up, it's funny to observe lol.
Click to expand...

why ? is it like technically impossible ? or financially impossible ?
or something i missed to consider ?
 
AlamoX said:
why ? is it like technically impossible ? or financially impossible ?
or something i missed to consider ?
Click to expand...

Don't worry, he thinks gpus can't and won't be made on 14nm lpp. Fury X is an example of multiple dies on an interposer and the whole industry is gearing up for it. There isn't a huge amount of reason to do it for multiple gpu dies on one interposer right now. There are huge reasons to do it with APU/SOCs sooner for being able to optimise the various types of unit on the chip. Make the gpu on one process, cpu on another and put them together on an interposer. Lower leakage for a higher clock speed cpu and higher density on a lower clock speed gpu rather than sub optimal density for the gpu and lower clock speed on cpu as they are both made on the same process.

Apparently it's based on nothing because in his vast knowledge of not paying any attention to the industry, there is a huge amount of research being done by all the big players into the area, the push towards such chips is apparent from industry talks over the past 4-5 years in the same way interposer/on package memory talks were being made before that.
 
AlamoX said:
why ? is it like technically impossible ? or financially impossible ?
or something i missed to consider ?
Click to expand...

More like

-Raja says something vaguely related in an interview
-people start talking about what he means
-assumptions from that discussion become fact
-people start making new assumptions based on these "facts"
-the cycle continues
 
BBC seems to believe Sony will output games at 4k.
Due to their source.

"Sony revealed to the Financial Times on Friday that it is working on an upgraded PS4 codenamed Neo. It said it would be able to play games in 4K resolution - but would not be ready to unveil at E3."


BBC article
 
I daresay a GTX1080 will not only wipe the floor with the PS4.5 and whatever it is Microsoft have up their sleeves but even the PS5 and next Xbox will probably be trailing the inevitable 1080ti when they are released in five years time.
 
There is no way on earth a PS4 will run games natively at 4k. If you're expecting to play games in 4k on a console, I have a feeling you might be disappointed.

1) Most people still have 1080p TVs
2) The hardware needed to run 4k natively would need to fit in the console form factor and consume as little power as possible so a compact power supply can cope and to avoid potential overheating
3) Such hardware is expensive
4) The current consoles, which were released just a few years ago, struggle with 1080p. 4k gaming would be a massive jump to essentially be apart of the same generation

I personally think the new console will be something along the lines of 1080p/60FPS, possible 4k bluray video player and better support/performance for Playstation VR.
 
Last edited:
Orangey said:
If you want to bait someone into an argument at least do them the common courtesy of not deliberately misrepresenting their position.
Click to expand...

You didn't repeatedly call 14nm lpp unsuitable for high power or gpus? Oh wait, you did, over and over, and when questioned on a single source or why you think a 200mm^2 gpu or 400mm^2 gpu makes absolutely any difference at all you never have an answer.

Tell me why, if on a 232mm^2 the transistors in the middle can work next to each other, but if you make a 400mm^2 chip the transistors in the middle suddenly can't work together because the chip size is bigger? There is literally no difference in how a process produces big or small chips. A 232mm^2 using 150W and a 464mm^2 chip using 300W are fundamentally no different in how they are produced. If a wafer has a defect rate that means say 10 of 250 of the smaller chips would fail, then 10 of the 125 bigger chips would fail, but there won't be more defects on the wafer, that isn't how it works, the size of the chip only effects the overall yield, nothing else. There is no type of chip that can be made small but not large on any process, only a performance that transistors can achieve on any size of chip.

The issue as you always portrayed it was you can make low power chips but not high power chips then you conflated incorrectly that this meant small or big chips. You simply didn't know what you were talking about but your stance was always it's not suitable for high performance. If a process can't make a 400mm^2 gpu because of the amount of power it uses with the transistor/process design that process also absolutely could not make a 232mm^2 gpu.

This has always been the problem you think there is some fundamental difference in process technology that lets you make small but not big chips and you're entirely wrong about that.
 
Orangey said:
More like

-Raja says something vaguely related in an interview
-people start talking about what he means
-assumptions from that discussion become fact
-people start making new assumptions based on these "facts"
-the cycle continues
Click to expand...

lol, complains about making assumptions, by making assumptions about the lack of facts this is based on. Raja didn't say something vague, the guys who give talks to the industry about interposers and how to use them(look them up, there are dozens) have been making the same talks to the industry about putting different chips together on an interposer for years. AMD has gone into a lot of detail on this subject in various technical talks and so have dozens of other industry players.

This isn't vague, AMD have said they WILL do this, the entire industry has said they WILL do this.

Oh, and Xilinx what... oh right, they already have chips out that combine up to 4 dies on an interposer to have better yields than making one giant die. So it already happens in the industry and everyone says they will utilise it... but Raja vaguely mentioned it so it's BS everyone jumped to an assumption over.

The only potential assumption is that it will happen for higher volume gpu parts soon. It's very likely we'll get an X2 type card done this way but as a lower volume niche part, it won't be used for a $200 gpu in the next couple of generations. With 10nm looking to be potentially disastrous for cost, it's very possible we'll see things using more than one die on an interposer(excluding memory, which we've already seen) before we get to 7nm to work around 10nm.
 
no way lol has no one seen the new integrated motherboards with 1070gtx in them ;)

remember they could also put in a 1080gtx if they wanted. the price of doing so is irrelevant as they lose on making the consoles anyway . they always have. they make the money on the games themselves. also a long production.

my guess is they will go with the 1070 gtx configuration
 
drunkenmaster said:
lol, complains about making assumptions, by making assumptions about the lack of facts this is based on. Raja didn't say something vague, the guys who give talks to the industry about interposers and how to use them(look them up, there are dozens) have been making the same talks to the industry about putting different chips together on an interposer for years. AMD has gone into a lot of detail on this subject in various technical talks and so have dozens of other industry players.

This isn't vague, AMD have said they WILL do this, the entire industry has said they WILL do this.

Oh, and Xilinx what... oh right, they already have chips out that combine up to 4 dies on an interposer to have better yields than making one giant die. So it already happens in the industry and everyone says they will utilise it... but Raja vaguely mentioned it so it's BS everyone jumped to an assumption over.

The only potential assumption is that it will happen for higher volume gpu parts soon. It's very likely we'll get an X2 type card done this way but as a lower volume niche part, it won't be used for a $200 gpu in the next couple of generations. With 10nm looking to be potentially disastrous for cost, it's very possible we'll see things using more than one die on an interposer(excluding memory, which we've already seen) before we get to 7nm to work around 10nm.
Click to expand...

I think 10nm will just be skipped like 20nm, especially if the 7nm looks to be on track.
 
~>Dg<~ said:
no way lol has no one seen the new integrated motherboards with 1070gtx in them ;)

remember they could also put in a 1080gtx if they wanted. the price of doing so is irrelevant as they lose on making the consoles anyway . they always have. they make the money on the games themselves. also a long production.

my guess is they will go with the 1070 gtx configuration
Click to expand...

Even if consoles did take modern hardware, top of the range at that if we are talking about the 1080, it still can't do 4K at 60 fps.
 
Muz333 said:
Even if consoles did take modern hardware, top of the range at that if we are talking about the 1080, it still can't do 4K at 60 fps.
Click to expand...

Yes it could at console level. They have much better optimisation and it will likley still have odd settings turned down graphically and wont be 60 fps but around 30 still
 
You must log in or register to reply here.
Back
Top Bottom