Predict: The Next Generation Console Tech

Status
Not open for further replies.
1080p may become the 'standard' for next-gen consoles.

Although I have no doubt that some games will be rendered @ 720p because some developers will choose the lower-end of HD so that they can gain performance/speed.

Just like in this current-gen, some developers use resolutions *below* 720p for the purpose of better performance.

720p requires half the fillrate & less bandwidth compared to 1080p. With that said, there's no doubt that a larger percentage of games will be rendered @ 1080p next-gen.



So were will PC's poor all that extra power? We're only going to need double the power for 60fps/1080p in "next gen" console games so thats achieved 1 year after their release. What do even more powerful GPU's do after that?

Only doubling the power/performance of X360 & PS3 would absolutely -not- automatically allow for 60fps/1080p in all current-gen games. For some games it would, yeah, but far from across the board in all games.

With that said, 60fps/1080p is just the framerate & resolution aspect of game visual performance. What about actual graphics? Are you saying you would be happy with X360/PS3 level of geometry detail, complexity, lighting, texturing, pixel shader/ing, animation, etc, if only it ran at 60fps/1080p?


You asked:
"What do even more powerful GPU's do after that?"

More powerful GPUs, whether they're two, three , five or ten times as powerful, would allow developers to create more complex graphics per frame.
More geometry detail. More complex pixel shaders. Higher quality textures and more of them. Better animation. Better, more complex, more dynamic lighting, and so on.

Next-gen consoles such as XB3/PS4 will launch sometime between 2011 and 2013. They will then have to last for most of, if not all of the decade. Do you think that the level of graphics detail we have now will be "good enough in 2015, 2016, 2017, 2018, 2019, even at 60fps/1080p? I say no. I won't hesitate for a second to say that graphics are still and will still be important. Graphics aren't everything, but gaming is still very much a visual medium.

Also, if you think real-time game graphics cannot get much better than 360/PS3 , you'd be wrong. Game graphics can also go well beyond Crysis. We are nowhere near the limit of what even the average untrained human eye can perceive. I mean even people who don't understand graphics, they will still be able to see a huge difference if visuals make a big enough leap. I also don't believe that game companies are gonna be crushed by huge costs in development. Developers will overcome the challenges. Better tools will come about. Programmers will make the transition from multicore CPUs (current-gen) to manycore CPUs (next-gen). The way the art pipeline is 'done' will change & adapt to allow games with far better visuals to be acomplished.


I'm not saying it will be easy. I'm not saying there won't be some companies that fail, or some game projects that tank. That happens every gen. Some developers and industry people feared PS2 specs in 1999 saying that only 4 or 5 of the top game companies would ever push PS2. The Ps2 was made to shine by dozens of developers. That was a difficult architecture. PS3 is a difficult architecture, but its getting pushed. Xbox 360 is an easier architecture, its getting pushed.

Next-gen XB3,PS4 with manycore CPUs, GPUs that are beyond RV770 and GTX 280 (in many ways) and GBs of RAM will provide game environments, physics and graphics not possible today. Or, perhaps I am wrong. Maybe the videogame industry will stop advancing visuals for the most part. Maybe Microsoft & Sony will do their own Wii' with a 50% to 2x increase in CPU & GPU performance. I doubt it though ;)
 
Last edited by a moderator:
Thanks for the response Megadrive but I think you completely misunderstood what I was saying ;)

When I said "We're only going to need double the power for 60fps/1080p in "next gen" console games" I was referring to the XB3/PS4 and the assumption was that those consoles would be running the games at 1080p/30fps. Hence PC's only need double their power to get to 60fps at 1080p. Obviously though if the consoles ran at less than 1080p or 30fps then PC's would need relatively more power to achieve the higher settings in the same game.

All the above is also assuming a large increase in the actual graphics side of things at the same time which is of course, the most improtant aspect.

The problem is that aside from a few cases today, PC graphics are tracking console graphics pretty closely with the main improvements being resolution and framerate. i.e. to run a modern console game like say Bioshock at 1080p and steady 60fps you need a good 3-4x the power of RSX/Xenos and so GPU's like the 9800GTX still have a worth while job to do simply attaining that holy grail of image quality/framerate. Generally PC games come with minor graphical improvements as well which will pull the framerate down a bit compared to the console versions but nothing major.

My point is that this trend looks set to continue and even go further than it is now. But rather than needing 4x the power to reach the top image quality/framerate settings like we do today we will only need 2x next gen. So were you could argue that a 9800GTX is all you would ever need today because it can run all console games at 1080p/60fps, next gen the equivilent could be only needing something like an 8800GTS 320MB! Without major improvements to the core graphics, more powerful GPU's than that could be worthless until you start moving beyond 1080p (which won't become mainstream for many years if ever).

At present the PC does still have a few titles which need the extra power, e.g. Crysis, but it seems next gen even these might be gone. Hence the question, where will we pour all the extra power - because the assumption is that devs won't be increasing the core graphics all that much for the PC versions of games.

What i'm hoping is that that trend will reverse and devs will start making larger increases to the core graphics of PC versions because of the larger number of machines that will be capable of handling then (relative to this generation).
 
I hope my next question is appropriate under this thread title.

How comparable are the xbox360 unified shaders versus Radeon HD4850 unified shaders? Or are those called stream processors today or what is the difference? Xbox360 has 48 Unified Shader Pipelines and HD4850 has 800 stream processors, but this can not really mean HD4850 has 800/48 = 16x more shader power than Xenos, can it?

Where I am getting at is, that if we knew approximately how 1 Xenos shader compares to 1 AMD HD48xx shader and extrapolating how many shader processors are there in mid range GFX card in 2010, we could have an idea of shader power excepted from XB3/PS3 compared to Xbox360/PS3.

So... yeah! :D
 
I'm not sure how comparable shader ALUs are b/w Xenos and R600+, but I suppose you could (very) broadly say that Xenos has 48 shader units and R600/RV670 have 64 more advanced shader units. By that (il)logic, RV770 has 160 shader units (=800/5).

Or I could try to follow aaronspink's lead and say Xenos has three shader cores ("arrays," as Dave calls them), R600/RV670 four ("SPU clusters," per Rys), and RV770 ten--again, keeping in mind R600+'s cores are more capable than Xenos', and also ignoring how texturing and other parts of the pipeline are tied into each "core."

As to how comparable Xenos' shaders are to RV770's, that's for someone smarter than me to answer, but I think it's fair to say the difference is nowhere near 800/48 on average, and maybe not even in a corner case (the worst I can uncomprehendingly imagine is an RV770 shader unit can work on 5 scalars per clock while Xenos' can work on 2, so 160*5=800 vs 48*2=96). I'm not sure how Larrabee or DX11 will change how GPUs will look in the next-gen console timeframe, or how midrange GPUs relate (on 360's release, Xenos was closer to R580--a high-end GPU--in terms of shader compute power than a midrange card). This paragraph is tiny b/c I'm on even shakier ground than my previous two.

As to whether this post is appropriate, that's for a mod smarter than me to decide. =)
 
From reading (again) Dave's article, by today marketing terms xenos could be viewed as 240 stream processors.
Not sure it's relevant to perfs anyway.In fact in the article says:
Each one of the 48 ALU's can co-issue a vector (Vec4) and a scalar instruction simultaneously, essentially allowing a "5D" operation per cycle

Then it should be interesting to have further informations on how this is made out.
But I feel like it should be close (thus less advanced) than what we find in actual ATI gpu.

Xenos should benifit form both ILP and TLP.
 
Last edited by a moderator:
PC also have the opportunity to go above 60fps, ever since the Q2/Q3/UT days. for now only weirdos/sensible persons that stick to their CRT can enjoy it but soon real 120Hz LCD monitors will be introduced and eventually we'll have OLED and other techs.

I prefer to have 60fps as a minimum. 100fps average seems the sweet spot for really smooth motion, and 150fps average with min rate in the 80s the perfection for some quake 3 playing. Vsync off is very reasonable at 100 or 120Hz. vsync introduces gaps in the framerate and even with triple buffering it looks noticeable to me (except if you're playing an old game locked or almost at the max)

I would be very curious to see far cry very high at 100fps average, 4x AA. so there's still GPU power to burn. In the future games targetted at consoles would look good enough on PC I'd say. Considering game artists working on high res textures, normal maps etc. and downsizing them as needed to make them fit on consoles. Content and data management will be more and more the main thing.
 
Play from the HDD, or, Testing & Conditioning for 3D Gen 4

Synopsis: "Play from the HDD" is as much about future consoles as it is for the here and now. Is the right/wrong approach for the 4th gen 3D consoles? Can you think of a better solution for "next-gen" memory design?

General Thoughts: MS announced they are allowing gamers to now copy their games to the HDD which, in turn, will allow for quiter gameplay (less DVD whirl) and faster load times (HDD > DVD in seek time, transfer rate). The press release read something like this:

Play from hard drive. Copy your games from the game disc and play directly from the hard drive. Not only will the drive not spin, but load times are quicker, as well. Of course, you will still need the disc in the tray to prove you own the game.

While a nice "perk" to those with the HDD space (or a HDD at all! zing!) I started this thread because I believe there is something more subtle to this move: testing for the next console generation as well as consumer "conditioning."

On the testing front MS has the ability to track how many consumers use the feature, the impact on HDD space, benefit/detriment to the gaming experience, and so forth. They are also now in a position to survey consumers on their thoughts on the implimentation, what demerits consumers find with it and what they like, etc with an eye toward future console development.

The conditioning element, though, is what I find interesting. The amount of "positive" response I have seen here and at Gamersyde is surprising, yet it appears a lot of gamers are willing to trade off HDD space for less noise and faster load times. Having endured 3 years of non-HDD standard performance limitations it appears the HDD options is now a "positive point" which can be conditioned into a consumer selling point down the road (instead of taken for granted).

My take on this is that this is a strategic move for MS that plays into the Xbox 360 as well as the Xbox 3.

The Now
* Give consumers less noise now.
* Give consumers faster load times now.
* Resolve disk spanning complaints (ugh, I have to swap disks?!).
* Resolve some disk space issues (dev needs 20GB of space? Require the HDD).
* Performance tracking.
* Consumer feedback.

The Future
* 4th Gen consoles are facing storage issues.
+ Games will continue to grow in memory footprint requirements.
- As worlds become larger and more detailed there becomes a need to populate memory quickly for game access.
+ Bandwidth needs will increase for storage and immediate access.

+ Solid state technologies remain very expensive, but...
- Offer improved transfer rates and seek times, in many cases, over optical media

+ Optical media is cheap, but...
- Optical media is not fast; transfering 2GB-4GB of data (extrapolated memory footprint for new consoles) at 20MB-30MB/sec. is a much lower ration than even this generation and won't suffice on new consoles.
- Optical media is not quick; seek rates are pathetic.
- Optical drives can be loud (especially faster ones), take up significant space, increase failure rate, design complexity, etc

* Distribution: Digitial distribution from online networks is be a major factor
+ May not displace retailers (shelf space is important for mindshare, retailer exposure) but will continue to suppliment.
+ Other media, like music and movies, are encourching the console consumer market.
+ The community/social network concept is slowly trickling into the console space.
+ Steam has demonstrated that pre-caching content is viable (game goes "gold," consumers with intent to purchase cache large chunks of the game so when the game is officially released they can play sooner than later).


I am not sure how the new consoles will tier their memory systems but there are significant troubles ahead. While there isn't a single motive behind the Xbox 360 game caching, I believe it has been positioned for market research and testing for their next console. What I find interesting is that it is a low-tech "PC" approach. IMO the PC is often a ripe market for product testing: if something is viable and affordable it can gain some headway in the PC market. At this point the approach of "installing" games to a HDD appears to be a viable middle ground for the 4th gen 3D consoles:

* HDD space per GB is cheaper than SSD.
* HDD's offer superior seek times compared to optical media.
* HDD's offer superior transfer speeds compared to optical media.
* HDD is necessary for continued Digital Distribution development.
* HDD installation allows companies to retain Optical Media for cheap media distribution but leveraging the HDD for superior performance.

While not a sexy solution, and HDD costs don't diminish much over the lifecycle of a console, I think the new consoles are at a point where a HDD will finally offer enough "selling points" (digital distribution, perpetual social networks, pre-game release caching, "Avalache" style p2p networks, semi-resolution to game load times and perpetual world issues, game modifications and customization, demos and trailers, and so forth) to justify their inclusion again from MS. And not just as a content save area, but in a PC-style "installation" approach. Buy a game, install a game (play a minigame, watch a intro video, etc to viel the installation), play the game.

If asked right now what my ignorant opinion was about the biggest hurdles facing the next consoles (2010-2012 window) I would say input and storage. Per storage consumers are not going to desire enduring even longer load times than faced now. And with diminishing returns increasing graphically it is my opinion that some games becoming "bigger worlds" and more interactive (hence more memory) will be a big factor in console designs (will trading up to 4GB of memory over 2GB be a bigger boost to game design over, say, 20% more die space for GPU/CPU power?) Tiering memory for best performance/cost is going to be a focal point of new console design and, as of right now, I think the PC model which MS is introducing to the 360 seems to be a strong front runner for their new console. There may be better approaches, but this one is a known quantity at this time.

So my nugget of thought is this Xbox update is as much as testing the viability for the concept for the future. Does it work? How do consumers like it? What negatives can we resolve? Can we begin developing our pitch for this feature now and condition consumers to approach the negatives as really positives?

Feedback: I would love to hear others give feedback on realistic solutions for future consoles. Memory tends to be overlooked but it is a big part of the experience and with the growing importance of connectivity and the performance issues of optical media (as well as cost/cost reduction issues of SSDs and HDDs) there remains some significant questions about these issues. Likewise future "bottlenecks" and expected game size and needs are important factors. How much are games going to grow? What can publishers afford to fund? Will the emphasis favor fidelity per-frame or "more stuff [space] for more diversity"?

Predictions? My guess is the new consoles will use optical media (BDR), some Flash memory (cheap!), and will bite the bullet and use HDDs and MS will go back to using them, partially, like the Xbox1 and cache game content to the HDD to overcome performance issues. It appears, at this time, to be the best cost/performance tradeoff. I/O

Ps: Wow this thread has grown! I haven't read the last, like, 20 pages! Sorry if this has already been discussed (I almost started a new thread for it, but it appears to work in here). When/If I get some spare time I would like to ask some devs what are the biggest problems they are having this generation and some realistic solutions they expect to be implimented as well as their view of future HW. Lately there has been a lot of talk of everyone going cheaper HW, but one thing to consider is that Ninny's model doesn't necessarily work for everyone else. And while an uber complicated/expensive console is not reasonable, if MS/Sony undershoot they may find themselves in a poor position if the competitor brings better HW (in general) at a meager additional expense. That said, I think some budget messaging where pure performance takes a backseat to memory model and other considerations that add to the gameplay experience (and game quality) in different ways is possible. 100k poly model versus a 140k model isn't always easy to spot and making a concession at this point for, say, 50% more memory may be a solid tradeoff.

The market is interesting right now, as well, as partnerships could be of great importance. For example Intel has purchased Havok and NV has aquired Ageia. Epic, for example, is using PhysX and has a long running relationship with NV so that may be an appealing avenue. I still think that it is likely MS will go a route with more GPU silicon real-estate for the reasons that (a) higher peak FLOPs for marketing purposes (b) initial games will have some extra GPU headroom and (c) down the road the GPU resource could be leveraged as games hit the CPU wall to get more out of the system, especially for new game designs that may have a focus on tasks that can use the resources (e.g. physics). This meshes with MS's control over D3D and leveraging one of their core strengths and wouldn't put them on a straight competition (ala CPU vs. CPU) as it would change the paradigm of competition some. I wouldn't be surprised if MS went with at least some high thoroughput CPUs to accelerate development and make taming difficult to parallelize tasks quicker for smaller developers (with the moto: if you need more power, look to the vector units and/or using the shader array). The aquisition of some middleware for PC/360 development and to address the cost/ease of development issues, especially content generation, is something MS already is working on...

Well, that is all the time I have for now. I look forward to back skimming this thread!
 
I would be fine with installing games. But give you the option to do either.

So that when you get home from the store with the game, you can play it right away without having to do a long install.
 
I would be fine with installing games. But give you the option to do either.

So that when you get home from the store with the game, you can play it right away without having to do a long install.

That's exactly the option MS is giving you with the 360.
 
Which will be very interesting as it means PC's won't even be scaling up resolution over console games any more. Image quality should be roughly even aswell (although by then 8xMSAA may well be a high end standard for PC's.

So were will PC's poor all that extra power? We're only going to need double the power for 60fps/1080p in "next gen" console games so thats achieved 1 year after their release. What do even more powerful GPU's do after that?

Hopefully devs will do something with all that spare power, since mid range PC's will usually run at a lower res than the consoles and hence not need as much power, we may see console to PC ports becoming even more common.



By the time next-gen consoles arrive, high-end PCs will be working at 1600p (2560x1600) resolution. So some extra power will be used there.
 
I despise mandatory game installs with a passion. It's totally against the plug-in and play ethos of console gaming that goes all way back to the Atari VCS (and probably before that too).

There's also the point that the hard disk supplied with the consoles are simply not big enough to cope with a library of mandatory installations along with all the other stuff we're supposed to be using our consoles for - media, downloadable games etc.

The Bourne Conspiracy on PS3 has a 4.6GB install that takes the best part of 15 minutes to complete and the game is still sub-optimal to the 360 code which only takes a few seconds longer to load its levels - Devil May Cry 4 all over again. Same with Top Spin 3 (nine minutes!). A 40GB hard disk could conceivably be filled with 7-8 installs once factoring in some space for playable downloads etc.

Installations should be optional, not mandatory, and in this respect I think MS have got it spot on.

The future is still based on optical disc, I'm sure, and I just hope that Blu-ray drives speed up dramatically in time for the next 'gen'. I'm sure they will!
 
I despise mandatory game installs with a passion. It's totally against the plug-in and play ethos of console gaming that goes all way back to the Atari VCS (and probably before that too).

There's also the point that the hard disk supplied with the consoles are simply not big enough to cope with a library of mandatory installations along with all the other stuff we're supposed to be using our consoles for - media, downloadable games etc.

The Bourne Conspiracy on PS3 has a 4.6GB install that takes the best part of 15 minutes to complete and the game is still sub-optimal to the 360 code which only takes a few seconds longer to load its levels - Devil May Cry 4 all over again. Same with Top Spin 3 (nine minutes!). A 40GB hard disk could conceivably be filled with 7-8 installs once factoring in some space for playable downloads etc.

Installations should be optional, not mandatory, and in this respect I think MS have got it spot on.

The future is still based on optical disc, I'm sure, and I just hope that Blu-ray drives speed up dramatically in time for the next 'gen'. I'm sure they will!
I completely agrees :)

On the other side caching is fine ;) (read transparent to end user).
 
By the time next-gen consoles arrive, high-end PCs will be working at 1600p (2560x1600) resolution. So some extra power will be used there.

I'm not too sure about that. I can't see 30" screens ever becoming more than a very niche market, even in high end PC gaming. At least not for a much longer time period than the launch of the next gen consoles anyway.

I expect to see everything settle at 1920x1200 for quite a while now.
 
I'm not too sure about that. I can't see 30" screens ever becoming more than a very niche market, even in high end PC gaming. At least not for a much longer time period than the launch of the next gen consoles anyway.

I expect to see everything settle at 1920x1200 for quite a while now.
You're assuming the dpi stays the same, it'd be nice to see higher dpi screens that would allow for higher quality rendering.
 
You're assuming the dpi stays the same, it'd be nice to see higher dpi screens that would allow for higher quality rendering.

True, but I think there are barriers to adoption of large screens, the largest of which are simple ergonomics (people tend to sit quite close to their screen(s) and may feel overwhelmed sitting so close to such a large screen) and cost.
 
I think looking at the direction that the three companies are currently taking their consoles can possibly shed more light on the specifications of their next gen line-up.

Microsoft would be trying to solidify their console as the central hub of digital entertainment via online, and so far, they're the best at that.
- It would be safe to say that their next console would have a base SKU that has at least 120 gb of HD space in order to promote DLC of games, movies, and other content.
- They would also implement motion sensing into their controllers. No brainer here.
- A similar architecture to the 360 that would allow full BC and maybe even BC with the 360 detachable HD.
- I would doubt that they need an extremely powerful console to achieve their goal of living room dominance, so I don't think it would be as big of the leap from XBOX to XBOX 360. It would probably be a refinement and enhancement of the XBOX 360 with DX 11/12 capabilities, with a heavy emphasis on image quality that was loosened during this generation.
- 1GB -2 GB of memory sounds reasonable.
- I seriously doubt they would use HD media such as Blu-ray. I am thinking they might still go with DVD since games need to be less than 8GB to be viable for download from the next gen marketplace.

Sony would be trying to do the same thing as Microsoft, but would be some years behind in the online space. They would be continuing their push of Blu-ray and Cell. They'll want to be profitable upon launch at the $299 price. They won't make the same mistake twice.

- The processor, without a doubt, would be cell or a successor to cell.
- DS4 controllers with 1:1 motion sensing, but DS3 is supported for their games.
- 1GB -2 GB of memory sounds reasonable.
- A minor GPU that's stripped of capabilities that the Cell can cover.
- Same hard drive system as the PS3.

Nintendo, ironically, would have the biggest upgrade in terms of raw power to their platform because they can, and have to. They would want to attract major 3rd party titles to their platforms, and would try to convince potential buyers that the Wii 2 is not just a repackaged Wii. I can just imagine the Nintendo PR: "Wii2, now with Hd cinematic graphics, and superior online experience!". Grandmothers and casuals would fall for it despite the fact that the Wii2 would be just a XBOX360+.

- Like many in this thread hinted, the CPU might be a Core 2 Duo derivative running at a low frequency.
- They'll make, FINALLY!, a step toward HD resolutions, so ram might be increased to 512mb.
- GPU might be a customized ATI gpu like the ones in the Wii and Gamecube but vastly more powerful.
- Harddrive support is a given, since Nintendo right now is trying to correct the issue of storage with the Wii.
- It would have the same disc based media as Wii.

In my opinion, the next generation of consoles would not have the traditional massive leap of computing power between generations, but more of a refinement as all 3 parties would want profitability upon or closely after launch. Coupled with the issue of diminishing returns in games and success of the Wii, I really don't see any of console manufacturers attempting to best each other in terms of graphical prowess. In other words, Nintendo is the cause of the downfall of console gaming.
 
Last edited by a moderator:
- The processor, without a doubt, would be cell or a successor to cell.
- DS4 controllers with 1:1 motion sensing, but DS3 is supported for their games.
- 1GB -2 GB of memory sounds reasonable.
- A minor GPU that's stripped of capabilities that the Cell can cover.
- Same hard drive system as the PS4.
Umh.., with a CELL 2.0 similar to what we have now I don't see Sony using a minor GPU.
And frankly I don't believe CELL 2.0 will be so different from its previous incarnation to suddenly be efficient at (let say..) heavy pixel shading.
 
Umh.., with a CELL 2.0 similar to what we have now I don't see Sony using a minor GPU.
And frankly I don't believe CELL 2.0 will be so different from its previous incarnation to suddenly be efficient at (let say..) heavy pixel shading.

So you think they'll have a stronger emphasis on the GPU this time around? I just don't see how they can have an amazing GPU, Cell processor, harddrive, and blu-ray drive and still be profitable or close to profitable at $299. One of them has to go, and I think it would be the GPU as Sony would NOT be emphasizing graphical prowess.
 
Last edited by a moderator:
So you think they'll have a stronger emphasis on the GPU this time around?.

Im curious on what numbers your basing this on. The RSX is running the show in terms of graphics. Its doing by far the most work in the graphics department.

While the cell is a superfast cpu, and excellent as far as cpu's goes for rendering graphics, its only in comparison to other cpu's that the cell is powerful when it comes to graphics.

Which makes sense because CPU's are processors designed for doing all kinds of things, while GPU's is a dedicated hardware chips designed purely for graphical fidelity.

CPU's, even the lighting fast cell, simply cannot compete with GPU's. Even a $1000 intel quad core 2 extreme cpu, cannot compete with a 4 year old nvidia NV40, hell, i doubt if it can match the visual fidelity a 6 year old Ati R300 could produce.

The RSX is much much much powerful than the cell at rendering graphics. While the Xenos that sits in the X360 is more powerful in most situations the RSX is not slow by any console standart.
 
Status
Not open for further replies.
Back
Top