PS4 Rumors?
- Thread starter AzBat
- Start date
Anybody hear any rumors on PS4 yet? I've heard through second hand source that Sony is already hard at work on PS4 here in the US(Montana I believe). Any truth or are they smoking something hallucinogenic?
Tommy McClain
I wouldn't be surprised
"Second hand sources"?
Good question.
I am pretty sure the evolutions of Cell over the next couple years is already pretty firmly in hand and STI have a general idea of their options/desires for the 2010-2012 timeframe.
Likewise the contract with NV was said to be future looking, so it is not too outlandish to conjecture the PS4 will likely have a NV GPU of some sorts. I am sure NV has some internal R&D looking at the G100+ series, but my guess would be G90 (i.e. new gen after G80 and G80 refreshes/reworks) is their main focus right now. Similarly planning in regards to memory (more Rambus?) and storage medium (multi-layered BluRay, holographic media, online distribution, etc) are being scrutinized. In many ways Sony can subtlely direct the movement of tech research they wish to exploit in some areas.
Overall my guess is Sony is already deep into silicon R&D and research, but actual chip targets, let alone system specs, are still years off. Kind of like how Sony announced initial research into what became Cell in 1999 before the PS2 launched.
One thing I think we can bet on: The PS4 will be an evolution of the PS3, specifcally Cell. Cell is a platform. Obviously there will be changed; e.g. larger LS would be a no brainer, another is some synergy in SPE (e.g. it was noted that a more scalar oriented ALU could be constructed and mixed and matched). STI invested a lot in Cell and I think Sony will be behind it for a long time. Of course some of the bigger questions for the PS4 is how large can they make it due to the limitations of silicon process reductions in the 2012-2015 timeframe and how much will other technologies come into play. Kind of early to be dead set on any thing at this point, but I think history says they are already looking into it.
I am pretty sure the evolutions of Cell over the next couple years is already pretty firmly in hand and STI have a general idea of their options/desires for the 2010-2012 timeframe.
Likewise the contract with NV was said to be future looking, so it is not too outlandish to conjecture the PS4 will likely have a NV GPU of some sorts. I am sure NV has some internal R&D looking at the G100+ series, but my guess would be G90 (i.e. new gen after G80 and G80 refreshes/reworks) is their main focus right now. Similarly planning in regards to memory (more Rambus?) and storage medium (multi-layered BluRay, holographic media, online distribution, etc) are being scrutinized. In many ways Sony can subtlely direct the movement of tech research they wish to exploit in some areas.
Overall my guess is Sony is already deep into silicon R&D and research, but actual chip targets, let alone system specs, are still years off. Kind of like how Sony announced initial research into what became Cell in 1999 before the PS2 launched.
One thing I think we can bet on: The PS4 will be an evolution of the PS3, specifcally Cell. Cell is a platform. Obviously there will be changed; e.g. larger LS would be a no brainer, another is some synergy in SPE (e.g. it was noted that a more scalar oriented ALU could be constructed and mixed and matched). STI invested a lot in Cell and I think Sony will be behind it for a long time. Of course some of the bigger questions for the PS4 is how large can they make it due to the limitations of silicon process reductions in the 2012-2015 timeframe and how much will other technologies come into play. Kind of early to be dead set on any thing at this point, but I think history says they are already looking into it.
Acert93, nice speculation. I agree on most points. The major question I have is should we already be expecting to see patent applications? Also, does Sony really have a Sony office in Montana? Though it's possible I've mis-remembered it and it's actually another northwestern state like Wyoming. I specificially remember it was in small town where they knew they would not be bothered and also control info from reaching the public. Looks like it didn't work too well. LOL
Tommy McClain
Tommy McClain
Accuracy depends on what that Sony employee does.
PS4?! 
Jesu H. Christo. On a crutch. PS4?! I have an overwhelming urge to lock this, but I'm going to write it off to the third Sapphire & Tonic on a Saturday night.
PS4?! At least when we engage in this kind of insanity on the PC side it is typically only 6 mos to one year away.
Jesu H. Christo. On a crutch. PS4?! I have an overwhelming urge to lock this, but I'm going to write it off to the third Sapphire & Tonic on a Saturday night.PS4?! At least when we engage in this kind of insanity on the PC side it is typically only 6 mos to one year away.
PS4?!
PS4?! At least when we engage in this kind of insanity on the PC side it is typically only 6 mos to one year away.
LOL, thanks man. You should know my track record I'm not one to post baseless claims or rumors. Heard about this a couple of months ago and have been sitting on it. Been trying to figure out if Sony in fact has an office in Montana(or wherever). Figured might as well put it out there to see if somebody else could confirm. Anyway, I about crapped my pants when I heard it. Came from somebody that I never knew was connected to the game business.
Tommy McClain
PS4?!
PS4?! At least when we engage in this kind of insanity on the PC side it is typically only 6 mos to one year away.
To be fair Geo, Sony announced work on what would become Cell BEFORE the PS2 shipped in the US.
As for the 6 month GPU thing... HAHA! G80 isn't out and there has been rumblings about NV's next-gen design for a long time; likewise loooooong before R520 came out we were not only into deep speculation on R580 but had a decent leg up on R600. So while we were all using R420 class GPUs we were already bored with the unreleased R520, had a good idea on the R580, and had a general barometer on R600! That is like us skipping the PS4 (old hat!) and revving up the PS5 chatter!
Don't give us any flack on the PS4 kind sir!
This much is true.
This thread benefits now from the Vysez Circle of Protectionâ„¢, speculation can now roam free in this thread!
Actually, this forum is way more interesting and/or better when we speculate about incoming products based on the latest news some of us heard from the grapevine or on patents.
The forum is perfect for discussing, quietly, old architectures too.
Discussing current stuff, especially the newly released products, is almost a vain endeavor seeing that one would have to say nothing but positive things, about the architecture on topic, because of some members who tend to be emotionally attached to their consumer electronic products to the point of believing that they're perfects and that competition products are in every way inferior no matter what.
Despite how early it is, there's always work being done on each successive hardware generation, most of the time, even before the next immediate iteration has yet to launch - So it's only natural that they're working on PS4 as we speak, and that MS is working on concepts for the next iteration of Xbox, and Nintendo is working on the successor to the Wii... What will we call that... Wii-ii?
That above, is a given, what is not a given is what form these hardwares will take on.
In another five or six years where will we be, what will realistically even be possible? Right now, we're seeing a shift just a little way out to 65nm, and I can say with a pretty high degree of certainty that we'll be coming out of the 45nm era when these things start to go into production - Maybe these machines will include 32nm parts. How many processor cores can you place on a 32 nano chip the size of CELL or the 360's CPU?
Will another five or six years see CPU-GPU convergence? Will we see homogeneity among code types, a type of ubiquity among command types and accepted programming practices between CPUs and GPUs? Will they both be capable of switching "hats", so to speak? Will my GPU do CPU work if my CPU becomes bogged down with too much, and vice-versa?
Will the next hardware iteration be little more than a cluster of extremely flexible GPUs? Will a cluster of shader units do all the work for the whole system? Will they process all the physics, and all the AI, and draw all the pictures?
This is all hard to say.
What is not hard to say is this - The partnerships formed right now, are not likely to go anywhere for the forseeable future - SONY is likely to stick with Nvidia, MS is likely to stick with AMD-ATI, which adds a new dynamic to the whole thing... Nvidia is likely to try to match the flexibility of their next-next-next-gen GPUs to whatever hybredised monster AMD-ATI come up with, and my own theory goes that PS4 and Xbox 360 will have either Hybredised CPUs (CPUs first and foremost with hardware accelerated 3D cores built directly onto their silicon), Hybredised GPUs (GPUs first and foremost with SPE-like multi-purpose cores built onto their silicon), or both... maybe even multiples... Perhaps a two by two hybred CPU hybred GPU grouping.
Xbox III with an IBM CPU (either their own design or an evolution of that inside the 360 today) and an AMD-ATI hybred GPU sounds plausible. It's doubtful AMD-ATI will allow their tech to be piggy-backed onto an IBM CPU, so maybe it'll be two AMD-ATI hybred CPU-GPUs, with one acting as the designated "CPU" while the other is the designated "GPU"... If the general performance of these plausible AMD-ATI hybreds are competitive with an IBM design around the same time, it may go this way.
PS4 with an IBM CPU (Cell-like... STI's baby) with Nvidia GPU tech built onto the silicon seems plausible, with a possible matching "C-GPU"... While AMD-ATI would have something to lose by allowing their tech to be installed in IBM designs, Nvidia would have only something to gain by it, the flexibility inherrent to a hybred design, which AMD-ATI, their only real graphics chip competitor, will soon have.
A question I have is this - Would these GPUs be be clocked 1:1 with the CPU cores?
Very interresting times lie ahead...
Maybe we'll see no hybredisation whatsoever - Maybe we'll see only forward-looking iterations of what we already have... Maybe we'll see CELL and Xenon mk IIs with four times their current processing units and twice the effective speed or efficiency, and GPUs that are equally as dense, but I don't think so. The benefits of the latter are too great to over-look, I think.
Besides the exotic stuff listed above, the standard list of perks is pretty much a given; Much more RAM, maybe as much as eight times the memory footprint, and the equivalent of GDDR6 by then, which should be inline with the internal CPU data-bus speeds we have today on CELL, only off-chip. Processor caches are likely to swell to several megabytes (ten megabytes is really not that far out, but it'll add bulk to the die) and internal bus bandwidth is likely to go into the terabytes per second range. Much of this bandwidth may be usable in preparation of data for the built-in GPU-like cores inside them.
But as I said... Very interresting times.
Dio
That above, is a given, what is not a given is what form these hardwares will take on.
In another five or six years where will we be, what will realistically even be possible? Right now, we're seeing a shift just a little way out to 65nm, and I can say with a pretty high degree of certainty that we'll be coming out of the 45nm era when these things start to go into production - Maybe these machines will include 32nm parts. How many processor cores can you place on a 32 nano chip the size of CELL or the 360's CPU?
Will another five or six years see CPU-GPU convergence? Will we see homogeneity among code types, a type of ubiquity among command types and accepted programming practices between CPUs and GPUs? Will they both be capable of switching "hats", so to speak? Will my GPU do CPU work if my CPU becomes bogged down with too much, and vice-versa?
Will the next hardware iteration be little more than a cluster of extremely flexible GPUs? Will a cluster of shader units do all the work for the whole system? Will they process all the physics, and all the AI, and draw all the pictures?
This is all hard to say.
What is not hard to say is this - The partnerships formed right now, are not likely to go anywhere for the forseeable future - SONY is likely to stick with Nvidia, MS is likely to stick with AMD-ATI, which adds a new dynamic to the whole thing... Nvidia is likely to try to match the flexibility of their next-next-next-gen GPUs to whatever hybredised monster AMD-ATI come up with, and my own theory goes that PS4 and Xbox 360 will have either Hybredised CPUs (CPUs first and foremost with hardware accelerated 3D cores built directly onto their silicon), Hybredised GPUs (GPUs first and foremost with SPE-like multi-purpose cores built onto their silicon), or both... maybe even multiples... Perhaps a two by two hybred CPU hybred GPU grouping.
Xbox III with an IBM CPU (either their own design or an evolution of that inside the 360 today) and an AMD-ATI hybred GPU sounds plausible. It's doubtful AMD-ATI will allow their tech to be piggy-backed onto an IBM CPU, so maybe it'll be two AMD-ATI hybred CPU-GPUs, with one acting as the designated "CPU" while the other is the designated "GPU"... If the general performance of these plausible AMD-ATI hybreds are competitive with an IBM design around the same time, it may go this way.
PS4 with an IBM CPU (Cell-like... STI's baby) with Nvidia GPU tech built onto the silicon seems plausible, with a possible matching "C-GPU"... While AMD-ATI would have something to lose by allowing their tech to be installed in IBM designs, Nvidia would have only something to gain by it, the flexibility inherrent to a hybred design, which AMD-ATI, their only real graphics chip competitor, will soon have.
A question I have is this - Would these GPUs be be clocked 1:1 with the CPU cores?
Very interresting times lie ahead...
Maybe we'll see no hybredisation whatsoever - Maybe we'll see only forward-looking iterations of what we already have... Maybe we'll see CELL and Xenon mk IIs with four times their current processing units and twice the effective speed or efficiency, and GPUs that are equally as dense, but I don't think so. The benefits of the latter are too great to over-look, I think.
Besides the exotic stuff listed above, the standard list of perks is pretty much a given; Much more RAM, maybe as much as eight times the memory footprint, and the equivalent of GDDR6 by then, which should be inline with the internal CPU data-bus speeds we have today on CELL, only off-chip. Processor caches are likely to swell to several megabytes (ten megabytes is really not that far out, but it'll add bulk to the die) and internal bus bandwidth is likely to go into the terabytes per second range. Much of this bandwidth may be usable in preparation of data for the built-in GPU-like cores inside them.
But as I said... Very interresting times.
Dio
Dio,
You bring up a interesting point with the hybridization stuff.
A C/Gpu (haha, get it) fits perfectly with the conceptual model of what the CELL is...an asymmetrical processing unit. I believe Sony's next outing (PS4) will have one chip, a CELL APU (Asymmetrical Processing Unit TM). Some cores will be unified-type shader cores, while others will be more fleshed out SPE-type cores with larger Local Stores, and one (or a few) core(s) will be the PPE-type. Who knows if we will still even use rasterization at that point. So, I'm not sure about ROPs.
You bring up a interesting point with the hybridization stuff.
A C/Gpu (haha, get it) fits perfectly with the conceptual model of what the CELL is...an asymmetrical processing unit. I believe Sony's next outing (PS4) will have one chip, a CELL APU (Asymmetrical Processing Unit TM). Some cores will be unified-type shader cores, while others will be more fleshed out SPE-type cores with larger Local Stores, and one (or a few) core(s) will be the PPE-type. Who knows if we will still even use rasterization at that point. So, I'm not sure about ROPs.
I think we'll still be using rasterisation, but I think it'll be a modification of what we currently do. Given the huge amount of shaded geometry we'll be capable of throwing at the screen, litterally tens-of-millions of polys per frame, I'd figure we'd have to begin sampling those at a sub-pixel scale, and then sampling those out, line-by-line.
Dio
Dio
First of all - awesome, one of the first real PS4 threads! Because you know when PS3 actually launches, we're going to get a little bored here in console tech.
Here's my take on what might be going on as the gen progresses...
Where before there were targeted launch years/dates, I think MS has offered the specter that they may be willing to launch updated cosoles in much more abreviated increments - and I think Sony has picked up on that. Thus it may behoove Sony to have a sort of 'rolling' development cycle for PS4, where at any given time they're ready to crystalize their PS4 plans around a certain architecture on a progressive continuum of development leading to who-knows-where.
Such that if MS pushed it to 2009, Sony would be able to freeze and form a plan around an 'in-progress' GPU architecture, and whatever the most current revision of Cell out is at the time, always ready to release a 'current' design. If it gets to the full six year average for them in 2012, then maybe we see a Cell/NVidia-GPU hybrid design, the chip they may actually hope one-day to reach as it stands now.
With the Cell architecture gaining some legs outside the console space, and NVidia now in it for the semi-long haul, I think Sony is in a good position such that whenever demanded, they could respond to a forced launch by Microsoft and not ask too much sacrifice of their developers in terms of the knowledge-base shift required in moving on to the next Cell/NVidia combo.
I wonder also on the side, since NVidia is a quite dynamic company in it's own right, what their relationship will look like in five years, or if they'll still have one. And if not, I wonder where the GPU would come from in such instance...
Here's my take on what might be going on as the gen progresses...
Where before there were targeted launch years/dates, I think MS has offered the specter that they may be willing to launch updated cosoles in much more abreviated increments - and I think Sony has picked up on that. Thus it may behoove Sony to have a sort of 'rolling' development cycle for PS4, where at any given time they're ready to crystalize their PS4 plans around a certain architecture on a progressive continuum of development leading to who-knows-where.
Such that if MS pushed it to 2009, Sony would be able to freeze and form a plan around an 'in-progress' GPU architecture, and whatever the most current revision of Cell out is at the time, always ready to release a 'current' design. If it gets to the full six year average for them in 2012, then maybe we see a Cell/NVidia-GPU hybrid design, the chip they may actually hope one-day to reach as it stands now.
With the Cell architecture gaining some legs outside the console space, and NVidia now in it for the semi-long haul, I think Sony is in a good position such that whenever demanded, they could respond to a forced launch by Microsoft and not ask too much sacrifice of their developers in terms of the knowledge-base shift required in moving on to the next Cell/NVidia combo.
I wonder also on the side, since NVidia is a quite dynamic company in it's own right, what their relationship will look like in five years, or if they'll still have one. And if not, I wonder where the GPU would come from in such instance...
Last edited by a moderator:
Actually, that's a very good point. I think everyone's going to draw up several different plans, based on different time-frames and assumptions. You can't just go down a single road. Imagine your "prototype" is based on a certain technological advance and suddenly there's a massive delay / problem with that concept. Moreover, your distinct solutions will be also based on the possible launch time-frames of your competitors.
My hunch surrounding Nvidia and the S.T.I. group at the moment is that to keep on level footing with AMD-ATI they'll begin merging technologies at some point with CELL-like CPUs. I think they saw this coming with AMD-ATI, maybe not precisely as it happened, but I think they knew that at some point GPUs and CPUs would get much closer, and wether or not it was the acquisition of ATI by AMD, or the lincensing of CPU core tech from Intel, IBM, or AMD, or someone else, they knew they had to follow suit.
With that said, I think the Nvidia-SONY relationship is only going to get stronger - That's their alliance, that's their assurance to a hybredised design, and SONY's assurance for the rights to use future Nvidia GPU designs for future hardwares. It feels like a mutual decision has probably already been made - This feel like a plan for symbiosis; Nvidia's hybred with Cell-like additions, and AMD's hybred with Radeon-esque additions. That sounds like their answer to ATI, and hybredisation in general.
Honestly has me pretty giddy thinking about it.
On the concept of serialised short-cycle system lives and launch periods, I dunno. I think both MS and SONY realise at this point that they simply cannot do that. SEGA tried this before and it failed hard for very solid reasons. I think at most, MS may try to shorten and lock launch gaps to around four years, though five seems more likely.
EDIT: I think I misread your point there.
Dio
With that said, I think the Nvidia-SONY relationship is only going to get stronger - That's their alliance, that's their assurance to a hybredised design, and SONY's assurance for the rights to use future Nvidia GPU designs for future hardwares. It feels like a mutual decision has probably already been made - This feel like a plan for symbiosis; Nvidia's hybred with Cell-like additions, and AMD's hybred with Radeon-esque additions. That sounds like their answer to ATI, and hybredisation in general.
Honestly has me pretty giddy thinking about it.
On the concept of serialised short-cycle system lives and launch periods, I dunno. I think both MS and SONY realise at this point that they simply cannot do that. SEGA tried this before and it failed hard for very solid reasons. I think at most, MS may try to shorten and lock launch gaps to around four years, though five seems more likely.
EDIT: I think I misread your point there.
Dio
Last edited by a moderator:
Similar threads
- Locked
