Predict: The Next Generation Console Tech

[Cheezdoodles]

I think it's extremely unreasonable to expect 8 GB ram, probably even 4 GB, for next-gen consoles. That money for this excessive amount of memory really is spent better elsewhere (or not at all).

4-8gb in 2010-11 is inline with the same memory budgets as X360\PS3 had. Its not exessive, its the same jump that was done last generation.

Especially when there are technologies (streaming, pixel shader and procedural textures, compression, tesselation) that let you work around memory limitations.

None of these technologies lets you completely work around memory limitations, and many cost a lot of resources, thus worse performance thus worse graphcis. 4gb is not excessive, i would say its the absolute minimum for a PS4\Xb3 that has good graphics in 2011.

 

[Gubbi]

4-8gb in 2010-11 is inline with the same memory budgets as X360\PS3 had. Its not exessive, its the same jump that was done last generation.

I would assign the following likelyhoods for memory capacity for a 2011 console as:
2GB: 50%
4GB: 40%
8GB: 10%

Moore's law is slowing down so that we currently see doubling of capacity every two years instead of every 18 months. The 360 launched in 2005 with 512MB so I'd naively expect to see 512*2^(6/2) = 4GB for a 2011 release console. If you look at the (global) launch of the PS3 we're closer to 2GB.

There will be a lot of focus on costs next gen. Sony will not make the mistake again and bleed multi-billion dollars on subsidizing sales initially. Microsoft went with the cheap and cheerful mass market GDDR3 while Sony partially went with the higher bandwidth XDR, but with a price primium.

Next gen both will need something boutique (faster XDR or some sort of GDDR-x ) to satisfy the bandwidth needs for what looks to be many-cored beasts. This will increase per bit cost and push the console further towards 2GB.

All IMO.

Cheers

 

[Shifty Geezer]

There will be a lot of focus on costs next gen. Sony will not make the mistake again and bleed multi-billion dollars on subsidizing sales initially...

I still say this statement is premature. Until we see the life-time earnings of these machines, we don't know if the investments of lossy hardware pay off or not. It may yet prove to have been a very profitable choice if the platform becomes establish as a centre for media consumption, spurring on sales of games, BRDs and content all filling Sony's coffers. If that happens, it becomes another valid option to take an early hit of $billions in order to establish a premiere platform with long term profitability.

 

[Acert93]

Next gen both will need something boutique (faster XDR or some sort of GDDR-x ) to satisfy the bandwidth needs for what looks to be many-cored beasts. This will increase per bit cost and push the console further towards 2GB.

Hey Gubbi,

I think in this regards I think Sony and MS engineers are going to evaluate the bottlenecks and best "bang for buck" for component investment and ... I wouldn't count shifting budgets toward more (and faster) memory out. You mentioned the need for faster memory, which is absolutely true. Having dozens of cores is fruitless if you don't have the memory to feed them.

My guess is that with optical media being... so... slow... that many of the stream techniques used (and strained) today will be even less useful. I think HDD caching may shore this up some, but I think a combination of procedural content (to ease load times) as well as gobs of RAM may end up being a solution they at least entertain. Load times will still be long, but once things are loaded you are in a much better situation than having 2GB and trying to figure out how to stream from an optical drive. A lot of this also depends on the feedback developers give: are we looking at needing more memory for larger interactive/persistant worlds? There is always an inflexion point. Out what point is gobs of memory not contributing to the end experience; at what point does memory outweigh the perks of more processing power?

With a 2011-2012 window I think there is still a little bit of time to see how things play out on the market. In the original post I deposited this idea:

Memory.

PS1 (4MB) => PS2 (36MB) = 9x in 5 years
PS2 (36MB) => PS3 (512MB) = 14x in 6 years
N64 (4MB) => GCN (40MB) = 10x in 5 years
Xbox (64MB) => Xbox 360 (522MB) = 8x in 4 years

As costly as memory is, 4GB of memory (8x the Xbox 360/PS3) seems pretty assured, unless the focus moves towards bandwidth.

Of course, as I noted back then, and as you point out, Moore's Law is not only slowing but any console design will also have to consider the difficulties of going from one process node to the next. If MS/Sony are stuck on 32nm for 3 years this will severely impact their designs.

Of course, we could see something radical like a "Cell Phone" style pricing model...

 

[Shifty Geezer]

Load times will still be long, but once things are loaded you are in a much better situation than having 2GB and trying to figure out how to stream from an optical drive.

In this regard, how's about some flash storage as a data cache? Perhaps 4/8 GBs of Flash just for streaming purposes? Load your level data into there, and stream off what you need to render the current scene, as it were, bypassing direct optical access. Would a half-and-half fast working RAM, fast cached storage be a viable system?

 

[Acert93]

In this regard, how's about some flash storage as a data cache? Perhaps 4/8 GBs of Flash just for streaming purposes? Load your level data into there, and stream off what you need to render the current scene, as it were, bypassing direct optical access. Would a half-and-half fast working RAM, fast cached storage be a viable system?

I had thought of that (I think I mentioned that in this thread) but I am not sure the flash Sony/MS would bundle in a console would be much faster, if at all, than a HDD. I think storage will be an important factor next gen, so flash my find itself redundant.

Although a "Core" sort of SKU with 16GB of Flash memory and "Pro" style consoles with HDDs may be viable. I guess it depends on the speed, cost, and lifespan of the FLASH and the data needs. I think there is more and more benefit, and money, to including a large mass storage device on a console. Why go with a Flash drive when you can get a much larger (faster?) HDD? Size and complexity, true. But with companies angling on selling music, movies, DLC, maybe even mod support, etc a small storage console may not be a great idea, especially if you have 50GB optical drives and need some way to cache that data to get it quickly to memory.

Oh well, hard to predict these things this far out. Who knows, maybe Sony will submarine MS and go a year early and turn the table. Process technology is pointing at 2011/2012, but maybe Sony jumps the gun with a strong offering in late 2010 or early 2011 and snags a ton of defacto exclusives?

 

[Shifty Geezer]

The fastest flash tech has much better seek times than HDD, so on paper I think there's a valid theory behind a Flash cache, as a sort of slow-RAM that's not actually too slow to be any use, but doesn't cost as much as providing multi-GB of fastest RAM. My wonder at the moment is how useful it would be for developers, as if it's not very effective, no point worrying about costs!

 

[wco81]

Oh well, hard to predict these things this far out. Who knows, maybe Sony will submarine MS and go a year early and turn the table. Process technology is pointing at 2011/2012, but maybe Sony jumps the gun with a strong offering in late 2010 or early 2011 and snags a ton of defacto exclusives?

I'd like to see someone come out early, with true HD support (no need to upscale).

But probably waiting a year or two increases the likelihood of more RAM.

Maybe if the video downloads sell better, they'd spend more money for more storage. Or maybe more RAM to support multimedia features.

 

[Cheezdoodles]

I had thought of that (I think I mentioned that in this thread) but I am not sure the flash Sony/MS would bundle in a console would be much faster, if at all, than a HDD. I think storage will be an important factor next gen, so flash my find itself redundant.

Your average SSD today is 10x faster than your average HDD today.

I guess it depends on the speed, cost, and lifespan of the FLASH and the data needs. I think there is more and more benefit, and money, to including a large mass storage device on a console. Why go with a Flash drive when you can get a much larger (faster?) HDD?

Samsungs 256gb SSD for the mobile laptop market has reported speeds of 200MB/sec read and 160MB/sec write, flash also have extremely low seek times compared to HDD's.

Right such disc are to expensive, but in a few years? No HDD can compete with flash memory in terms of performance either way.

 

[Acert93]

Maybe I am misunderstanding you, but do you think in a "couple years" a comparably sized SSD drive will be within shouting distance, in regards to cost, to a standard HDD? We can talk about SSDs when a 256GB drive with superior performance lands under $100. Maybe for the Xbox 4

While we are at it, lets bring up halodisks

 

[RobertR1]

A SSD would be silent compared to HDD. I think the advancement in tech will come from a few factors though.

Speaking for MS which will be the first ones out. They'll work with developers on trying to find a balance between increasing budget of games and pushing performance further. Development cost for games this gen has sky rocketed and I don't think the studios want to risk extinction from make a low selling game or two. That might ultimately determine how far the tech will advance. MS would obviously be smart to keep their hardware and toolset PC like so it's less of a financial strain on the studios to adapt to the new hardware.

The other factor being the Wii effect. If the Wii slows down it's steamroll a bit and MS starts making good profit and holding onto a decent marketshare (this is for them to determine not fanboys and anti MS people on the net) they might be more inclined to push towards higher tech vs spending r&d money on adding Wii components to their system.

One thing MS and Sony can certaily learn from nintendo is that it's ok to ship a game with their new consoles. This goes a long way to establishing consumer mindset. Wiisports showed the public that the Wii will be a cool toy that anyone can pick up and have fun on. Shipping it with Zelda or Resident Evil could have let to a very different mass consumer mindset.

 

[Cheezdoodles]

Maybe I am misunderstanding you, but do you think in a "couple years" a comparably sized SSD drive will be within shouting distance, in regards to cost, to a standard HDD? We can talk about SSDs when a 256GB drive with superior performance lands under $100. Maybe for the Xbox 4

I was just responding to your statement about flash performance vs HDD. Flash is faster. HDD's cannot compete vs SSD's on peformance.

Going by samsungs statements, SSD's could very well competee with HDD's on a large scale by 2011-12. They will not be cheaper than HDD's (eventually it will) per gigabyte, but they will be cheap enough for your average joe to buy one.

Considering how tremedously fast flash memory has gone down in price, (and we are not nowhere near the limits of cost reducting flash memory), it doesn't have to be to expensive in 4 years.

 

[MfA]

Flash has good access times, but it's still orders of magnitude below ram. That said, I hope flash SSDs are standard by next gen. If so I think we will see a lot of games installing to the SSD, with a little forethought the intro cinematics and tutorial levels should give plenty of time to get things going ... and the rest of the game can be streamed onto the SSD during normal play. For replays of savegames for games you removed from the SSD you will probably have to wait longer for (partial) installs, but I don't think that's a huge deal.

PS. flash is at the cutting edge of process technology and there isn't a lot of stretch left with MLC (two bits per cell is already making flash reliability an issue now, with smaller cells and more levels per cell things are going to get hairy). Without truly major architectural innovations (basically we need something that isn't flash) or major changes in manufacturing technology something more than say a factor 2 cost reduction (on top of reductions because of process shrinks) is not all that realistic in the near term for solid stage storage.

 

[Shifty Geezer]

Speaking for MS which will be the first ones out. They'll work with developers on trying to find a balance between increasing budget of games and pushing performance further. Development cost for games this gen has sky rocketed and I don't think the studios want to risk extinction from make a low selling game or two. That might ultimately determine how far the tech will advance.

I'm not sure the 'costs of development' argument actually holds up. The reason being, this is a big growth industry. There's loads more money coming into it, to fund bigger products. I'm seeing parallels with the movie industry. Movies now cost an order of magnitude than a generation ago, which cost an order of magnitude more than a generation earlier. People expect big, impressive, expensive effects, and Hollywood can afford it because Hollywood makes stupid amounts of money. If think the economics of games development, the balance of results, tools, optimizations etc, will balance out whatever happens going forward. An artifical cap on performance to save costs won't actually benefit the developers IMO - they can always choose to spend less after all, just as not every film made in Hollywood has to cost hundreds of millions of dollars to make.

 

[obonicus]

An artifical cap on performance to save costs won't actually benefit the developers IMO - they can always choose to spend less after all, just as not every film made in Hollywood has to cost hundreds of millions of dollars to make.

Different circumstances, though. A low-budget film will still have near-perfect resolution and be completely lifelike, and can be directly compared with far more expensive movies. That's not true for games; low-budget games will generally look (and play) far worse, to the point where you can't even have a direct comparison.

 

[RobertR1]

I'm not sure the 'costs of development' argument actually holds up. The reason being, this is a big growth industry. There's loads more money coming into it, to fund bigger products. I'm seeing parallels with the movie industry. Movies now cost an order of magnitude than a generation ago, which cost an order of magnitude more than a generation earlier. People expect big, impressive, expensive effects, and Hollywood can afford it because Hollywood makes stupid amounts of money. If think the economics of games development, the balance of results, tools, optimizations etc, will balance out whatever happens going forward. An artifical cap on performance to save costs won't actually benefit the developers IMO - they can always choose to spend less after all, just as not every film made in Hollywood has to cost hundreds of millions of dollars to make.

Remember, Hollywood studios also have a lot more movies coming out in a year than gaming studios. So if they lose on some, they can make it up on others. Gaming studios are much smaller and can only do a a title or two a year. Often 1 every two years with the development time this gen. Now if that bombs, you're hating life.

However, EA would be a great example to your Hollywood studio comparison. They're huge and push out product after product. They can take losses on some and bank on others (Madden, NCAA, etc..) but even they struggle now with games costing millions and selling only 200k copies!

 

[liolio]

RobertR1 when you say:

MS would obviously be smart to keep their hardware and toolset PC like

You think of using a SMP cpu that's it?

Because I think that Ms is unlikely to move from this design for their next system.

And in regard limiting hardware because the software couldn't keep with, I don't agree.
In fact, an improvement could be to make the cpu used in console more forgiving.
On a marketing side it would not be considered as "pushing the hardware" mostly because marketers (and uneducated persons) only understand peak figures.

You could have something that look like a twice a xenon but could be in fact could perform way better than two xenon put together.

there is a lot of room for improvement on cpu even if it doesn't land on way highter peak figures.

 

[Shifty Geezer]

Remember, Hollywood studios also have a lot more movies coming out in a year than gaming studios. So if they lose on some, they can make it up on others. Gaming studios are much smaller and can only do a a title or two a year. Often 1 every two years with the development time this gen. Now if that bombs, you're hating life.

Then change the format Supplement your big blockbuster movies with little download Arcade/PSN fillers. For individual developers that may not be an option, but it'd be publisher funding these projects. The publishers ought to have an expansive catalogue covering all budgets with prospects for earnings at multiple different levels. No-one should be aiming for nowt-but-AAA-blockbuster-Mega-sales as these things are unpredictable.

Still, this is too much of a tangent for the tech thread. IMO reigning in on the tech for cost reasons isn't a legitimate design criteria.

 

[Cheezdoodles]

I'm not sure the 'costs of development' argument actually holds up. The reason being, this is a big growth industry. There's loads more money coming into it, to fund bigger products..

Is it a big growth industry? I was under the impression that the video games market as a whole has been around 100million give or take 20 million for the last 20 years.

 

[Proelite]

Is it a big growth industry? I was under the impression that the video games market as a whole has been around 100million give or take 20 million for the last 20 years.

Try 17 billion and over for this current year.