Predict: The Next Generation Console Tech

Status
Not open for further replies.
No. 55 million consoles, 35 million live accounts

Not having an optical disc does not automatically mean the game must be downloaded from the Internet.
I've suggested alternative disc-less distribution years ago - from buy-on-flash, to bring-your-own-flash to kiosks and so on.
Steam kind of does this already. A Steam game "disc" is just bunch of files that populate the Steam cache "faster".
 
2 Gigabytes of super fast GDDR5 memory would be pretty cool.
Would GDDR5 be suitable in a UMA setup or would it have to be standard DDR type memory? In which case we are only now beginning to see the development and ratification of DDR4.

2GB is too little. This fall/winter we will have first cards with 4GB GDDR5. If you think of the 2013/14 timeframe, then 8GB GDDR5 should not be a stretch.

For what it's worth, when X360 launched in 2005 the highest-end graphics cards also had 512MB GDDR3 (X1800 XT, 7800 GTX 512MB version).

If they continue with this trend, we might get 8GB still.
 
PS3 has 512 MB. 4GB would be an 8x increase. By comparison, PS1 to PS2 saw a 10x increase; PS2 > PS3 saw an 16x increase; XB > 360 was an 8x increase. N64 to GC was 10x. Of course, choice of RAM doesn't rely on some formula based on previous increases, and the choice will be made for economies. If RAM has dropped substantially in price, it could be they decide to go with more, or if 4GBs is deemed an excess, or if the design uses eDRAM, they may go with less. Still, 2GBs would be unprecedentedly small increase of only 4x, contrary to the explosion in storage capacity we're seeing, while 8GBs would be a 16x increase.
 
I expect 4, with an outside chance of 2 or 8gb.

However I'm far more concerned about the storage system. Having a tonne of memory is great, but it you can only read at 20MB/s then beyond ~ 3GB is not going to be especially useful.
 
I am curious why is MS still using 64MB chips for RAM (no change from 2005)? Sony started with 8x32MB DDR3 + 4x64MB XDR and now they are using 256MB chips for both DDR,XDR.

BTW it would be crazy if sony goes 12x256MB XDR2 not splitted memory pool?
 
I am curious why is MS still using 64MB chips for RAM (no change from 2005)? Sony started with 8x32MB DDR3 + 4x64MB XDR and now they are using 256MB chips for both DDR,XDR.

BTW it would be crazy if sony goes 12x256MB XDR2 not splitted memory pool?



Good question.

But maybe Sony does not follow the path to put back 12 chips because of space, TDP / wattage and if do this you may come right away (less than 12 months) and another smaller model in the same way that migrated from 60GB to the 40GB.

XDR2* would be very interesting and since sony already used by two generations Rambus they could compete even with GDDR5!

* http://www.rambus.com/us/technology/solutions/xdr2/index.html

"The XDR™2 memory architecture is the world's fastest memory system solution capable of providing more than twice the peak bandwidth per device when compared to a GDDR5-based system. Further, the XDR 2 memory architecture delivers this performance at 30% lower power than GDDR5 at equivalent bandwidth.

Designed for scalability, power efficiency and manufacturability, the XDR 2 architecture is a complete memory solution ideally suited for high-performance gaming, graphics and multi-core compute applications. Each XDR 2 DRAM can deliver up to 80GB/s of peak bandwidth from a single, 4-byte-wide, 20Gbps XDR 2 DRAM device. With this capability, systems can achieve memory bandwidth of over 500GB/s on a single SoC...."


Edit:

"File Name: XDR™2 Memory Architecture
Today's powerful graphics and multi-core processors require significantly higher memory performance when compared to traditional single-core processors. Without adequate data bandwidth, memory becomes the limiting factor in delivering the performance desired in next-generation gaming, graphics and computing systems. As an example, the current generation of gaming systems require up to 50GB/s of memory bandwidth. Current graphics processors need as much as 128GB/s. Over the course of the next five years, gaming, graphics, and multi-core computing applications are forecasted to push memory bandwidth requirements to 500GB/s and beyond."

http://www.rambus.com/us/downloads/document_abstracts/products/xdr2_memory_architecture.html

I agree here...about text...for who will ensure they do not arise in the next 5 years processes that do not require more bandwidth than 50/70GB/sec?
Frostbite 2 engine that gives us a glimpse of what to expect in the coming years already use (forgive if I'm wrong) with 158MB VRAM for 1080P at high bandwidth(100+ GB/sec) requirements quite high(and will have to see results benchmarks when after release date...)*.

* http://www.videogamer.com/pc/battlefield_3/news/battlefield_3_pc_system_specs_revealed.html

" Recommended System Requirements:

OS: Windows 7 64-bit
Processor: Quad-core Intel or AMD CPU
RAM: 4GB
Graphics card: DirectX 11 Nvidia or AMD ATI card, GeForce GTX 460, Radeon HD 6850
Graphics card memory: 1 GB
Hard drive: 15 GB for disc version or 10 GB for digital version "
 
Last edited by a moderator:
As far as I know, XDR was commercially used only in PS3. I would very much like to see an example of XDR2 being used in consumer electronics, because I have seen none so far.

There must be some intrinsic reason why it's not used. It might be cost, or something deeper, like the minimum datasets that XDR2 fetches being too large for most applications (that was the going theory in overclock.net's Future Memory thread).
 
I expect 4, with an outside chance of 2 or 8gb.

However I'm far more concerned about the storage system. Having a tonne of memory is great, but it you can only read at 20MB/s then beyond ~ 3GB is not going to be especially useful.

How fast are the fastest commercially viable/cheap Blu Ray drives now?

Edit: checking Newegg, lowest price Blu Ray PC drive is a 59.99 Samsung model at 12x. That seems to be 54 MB/s recording speed, I assume read speed is the same? Read speed I cant seem to find.

It seems to be 6X the speed of 2X Blu Ray in PS3. So relatively you could have 3GB of memory with similar loading?

I think a cache of flash might help a lot there too as we've often discussed.
 
Last edited by a moderator:
It's a certainty that next gen consoles will ship with some form of extensive local storage, either in the form of a HDD or, more probably, a SSD.

Optical media, if used at all, will only be used for install, due to the benefits of orders-of-magnitude faster read-speeds/load-speeds that HDD/SSD provide.

Discussing optical media as a viable method for data streaming (besides fmv-s) in 2014 should in belong to the realm of discussing which phone modem gives the most stable throughput in 2005. ;)
 
How fast are the fastest commercially viable/cheap Blu Ray drives now?

Edit: checking Newegg, lowest price Blu Ray PC drive is a 59.99 Samsung model at 12x. That seems to be 54 MB/s recording speed, I assume read speed is the same? Read speed I cant seem to find.

It seems to be 6X the speed of 2X Blu Ray in PS3. So relatively you could have 3GB of memory with similar loading?

I think a cache of flash might help a lot there too as we've often discussed.

What are you trying to say? Facts are 1x bd-rom is 36MBit/s. PS3 has 2x drive with 72MBit/s. PS3 drive is CALV(same speed through whole disc). Those 12x drives are not constant speed, top speed is only achieved on outer rim.

It might be possible to upgrade from ps3 level(2x) drive to around 6x speed drive next gen achieving theoretically three times faster sequential read performance than ps3 has. There are pretty good arguments against game binary sizes growing directly in proportion to main memory size.
 
As far as I know, XDR was commercially used only in PS3. I would very much like to see an example of XDR2 being used in consumer electronics, because I have seen none so far.

There must be some intrinsic reason why it's not used. It might be cost, or something deeper, like the minimum datasets that XDR2 fetches being too large for most applications (that was the going theory in overclock.net's Future Memory thread).
Well, the place for XDR wouldn't be CE devices that don't need it, but computing, and that's already perfectly served with DDR and GDDR. The PC isn't configured for a single pool of shared RAM, so doesn't need large quantities of super-fast RAM, and all the chipsets and suppliers are all geared towards the existing DRAM format. Without a particular impetus to swap over to a whole new memory architecture, requiring new processors, motherboards, and RAM incompatible with other PCs, there's no reason for PCs to switch, which severely limits the market. XDR is better suited to supercomputing, although there may be similar cost issues there too. Howver, if it's as power efficient as claims say, it'll have operating cost savings.

On paper a unified XDR2 memory pool makes considerable sense in a console, and I'd be pleased to see 4GBs of XDR2 in PS4.
 
SSD would be way too expensive.

The idea of a hard drive less unit imo is still a very good one for cost reasons. It certainly helped Microsoft a lot this gen.

In fact I just realized, a DD only console would thus require a HDD and probably a pretty big one. You're almost better off with an optical drive, which is dirt cheap and in turn would give you the freedom to do HDD less SKU. Never mind that an optical is required for all the markets without great broadband of course.

Now, I still think an HDD less SKU may require some flash cache next gen.

due to the benefits of orders-of-magnitude faster read-speeds/load-speeds that HDD/SSD provide

It's mostly faster seek times right? If anything the streaming speeds should close in favor of the optical media, since 8X-12X Blu Ray is much faster than this gen's optical, but HDD's will not increase much in speed from this gen.

Anyways the difference to optical media is not as great as you make it out. I believe I've seen cases at compare sites like lens where HDD-less 360 version of multiplatform game loads faster than PS3 one with a HDD install. Of course done correctly this should not be the case, but casting the HDD less 360 as "unplayable" or the like is disingenuous.

Here's a more recent example, Deus Ex HR, the difference is real but not unplayable :http://www.youtube.com/watch?v=Jb1aUN1vZcU
 
SSD would be way too expensive.

The idea of a hard drive less unit imo is still a very good one for cost reasons. It certainly helped Microsoft a lot this gen.

This topic has been discussed to death in this thread http://forum.beyond3d.com/showthread.php?t=48244

I think the consensus is optical is not going to be fully gone in any scenario because there are markets where optical still is only viable distribution method. DD only SKU's might exist. It makes sense to make a cheap base level sku that can be made even cheaper in future(think about small flash soldered to motherboard or memory card to store only saves and temporary data. 2.5" drives cannot be made cheaper due to labour and parts and only scaling is you get more space for similar price. Not good for making a 99$ unit at some point). Power users need a LOT space where 2.5" or 3.5" optional HD makes sense.

Pick your combination and it all has been discussed to death in the thread above. Also you can add download kiosks to the picture where user takes his mass storage and downloads the game in a shop.

I'm in the camp where optical+some magic amount(32GB?) of fairly fast flash memory makes sense. This will give a good stable baseline for developers to do streaming/caching and flash price should scale down nicely during time, especially so as the performance doesn't need to increase as the cycle matures. Have an optional 2.5" or 3.5" hdd for those who are into downloads and media consumption(more expensive sku). Anyway we are still 2+ years from next gen starting which puts the flash prices and io controller prices to fairly reasonable range, especially so if the performance is reasonable not the fastest possible.
 
Last edited by a moderator:
Yeah, my post was meant to be more directed that I think a HDD-less SKU next gen is still probably a good idea.

Heh, especially if you buy into the "small upgrade" theory so many in this thread seem too. Then costs become an even bigger concern as any modest machines would also need to sport modest prices from the beginning.

But even if they are big powerful 399+ monsters, the same pattern that always occurs will again soon enough, the chips will fall in price, the HDD wont, becoming an increasingly difficult portion of the price.
 
Personally the more super fast RAM there is the better.
As Graham said earlier 2GB and 8GB are outliers but if I had a choice of either 4GB of average DDR3 or 2GB of very fast GDDR5 I would pick the latter due to its speed advantage. I don;t believe we will see 256bit buses but maybe a 192bit one...

As to games requiring oodles of RAM space for code, I still think most of it (80%) will be consumed by textures and geometry rather than code. What is the ratio these days?

It seems most are in consensus, a UMA system is going to be simpler from a programming perspective (see JC) and long term should be cheaper as well.

So guys have we given up on EDRAM yet? Imagine a 4x increase in capacity for the EDRAM to feed the GPU. *Drools*
 
eDRAM is necessary for bandwidth. If the RAM is fast enough next gen, eDRAM will be an unnecessary complication. And TBH I hope the RAM is fast enough and consider that more likely than not.
 
I believe the eDRAM currently is severely overrated; in this current gen it looked okay, but only because a competing console was bandwidth starved.
It does not help with the fact that AA methods which could benefit from eDRAM are not applicable to current/ modern or future (deferred for example) renderers.
so IMO eDRAM does not have a place in 2011 graphics, assuming 2011 hardware will not have really low bandwidth :)
 
Other than TSV and 3D ICs there is nothing on the horizon which would make external RAM substantially faster Shifty ... so I don't quite see why you would expect external RAM to be fast enough.
 
Status
Not open for further replies.
Back
Top