Predict: The Next Generation Console Tech

Status
Not open for further replies.
Ok the latest alleged Durango rumors per sources :p suggest a Cape Verde GPU, as much as 8GB of DDR3 (!).

These same imply PS4 looking at a better GPU but less RAM.

Salt salt salt but interesting. They make me turn over Mark Rein's comments all the more. Perhaps he wasn't just referring to Sony when trying to pump next gen specs up as I originally assumed, maybe the XB3 GPU isn't great shakes either.

I am not sure I trust these "sources" but I love a good detective hunt and all. For me the gold standard source is still lherre and he seems pretty bullish on next gen hardware.

Cape Verde also makes some sense in light of the 6670 allegations, since it's not that far from CV and perhaps MS recently upgraded.

Cape Verde at least looks good compared to 6670.

The other thing I keep coming back to is Epic's slide about 1.1 TF needed for UE4 at 720P. All along I've wondered why they would mention that info if perhaps not trying to tell us something? Could be nothing of course. It doesn't change the fact next gen games at 720P would be damn disappointing.

As I said though, I'm skeptical of all the above, but it's interesting to discuss anyway.

Coming back to it though, Cape Verde would still be highly disappointing. It barely trumps the exceedingly old 4890 from 2009 housed in my PC.
 
If I were in SonyHQ's shoes, I'd have been planning to have a nextgen box designed for 28nm with a target $400 BOM and sell the machine as close to BOM as possible with a launch as soon as the silicon was ready (q4/2012).
What will they be doing in 2014, 2015, and 2016? If the Bom can't be dropped more than $100 over 5 years thanks to technoligcal limits, do you really think it's best to have the console up at $300 minimum price. How is that getting the box into users hands to sell content?

The whole content strategy necessitates getting mass adoption, which necessitates either an amazing product (iPad) or a low price. Technological limits on production suggests the price drops cannot be relied, meaning a low pricepoint in 2015+ requires a lower starting point. Ergo, the need for a cheaper launch box. We already have a history of high-end consoles that shows they don't sell massive numbers until the price comes down, so unless the console companies are happy to sit at 50-100 million users while content services on PCs and iThings are exposed to hundreds of millions of people, they need to have an eye on the low, mass-adoption price.
 
Ok the latest alleged Durango rumors per sources :p suggest a Cape Verde GPU, as much as 8GB of DDR3 (!).
DDR3 makes sense - it's supposed to be an APU after all. And those come with DDR3.

That being said, we're probably looking at specs and performance along the lines of 2xTrinity. Not extraordinary - but nothing to sniff at.
 
Ok the latest alleged Durango rumors per sources :p suggest a Cape Verde GPU, as much as 8GB of DDR3 (!).

Coming back to it though, Cape Verde would still be highly disappointing. It barely trumps the exceedingly old 4890 from 2009 housed in my PC.

For a console, I think a 7770 level GPU would be a huge improvement. It would be interesting to see if we would see a APU + 7770 GPU though.

I think the 7770 gets really good performance with only a 72 GB/s of bandwidth. I think to get those kind of bandwidths out of DDR3, you'd need a pretty fat bus. 8 chips combined for a 256-bit bus at 2.8 Gb/s (fastest DDR3 I've seen) would net you ~89 GB/s. Sufficient for the Cape Verde GPU, but good enough for both CPU+CPU? I don't know.

If that's sufficient and you're going with 8 chips, then the question is whether 8 Gb or 4 Gb chips are available, cheap enough, and in volume for next year. Since DDR3 is such a high volume product, it may be possible to get 8 GB of DDR3.
 
Ehh, on AMD, I just disagree with the inclusion of EDRAM in the 360. That's probably debatable but I think on balance it harmed the system. That was an AMD idea.

But doesn't it depend on the trade offs?
What is the trade off is 4Gb DDR3 and some larger than 360 amount of EDRAM or 2Gb DDR5?
I know which I'd rather have.
Of course the EDRAM cost could offset the difference in memory cost, I'm just pointing out that it's not a simple EDRAM bad thing.
If you polled developers about the EDRAM probably the only thing they disliked about it is being forced to tile because it wasn't big enough.
 
One thing I still expect to be an issue is memory stacking. The difference between a console with stacked memory and one without would basically be a generation gap -- you can easily ship an order of magnitude more bw within the same cost and power envelope, and even with half of that the difference will be very visible. So given how much money Sony has poured into the tech, I'd assume they will ship their next gen exactly when they can manufacture a chip with wide stacked memory interfaces in volume.
sadly I've few hope. I would be really suprise if Sony beats the whole embedded market to it on one side and Intel one the other side.
Rumors give Haswell with an interposer and an L4 cache (/some fast mem on the pad).
I'm sceptic sadly.
 
What will they be doing in 2014, 2015, and 2016? If the Bom can't be dropped more than $100 over 5 years thanks to technoligcal limits, do you really think it's best to have the console up at $300 minimum price. How is that getting the box into users hands to sell content?

I'll refresh your memory here... xb360p has been selling for $299 since July 2008. FOUR YEARS.

No, I don't think that pricepoint is extremely prohibitive.

Granted, it would be better to have a lower BOM and correspondingly lower MSRP, but isn't necessary for success...

Let's evaluate a few other products with mass adoption:

iPhone = $599
iPad = $499
iPod = $399
Galaxy 2S = $599
Droid = $599

If it has value, these pricepoints have proven to be acceptable.

The whole content strategy necessitates getting mass adoption, which necessitates either an amazing product (iPad) or a low price. Technological limits on production suggests the price drops cannot be relied, meaning a low pricepoint in 2015+ requires a lower starting point. Ergo, the need for a cheaper launch box. We already have a history of high-end consoles that shows they don't sell massive numbers until the price comes down, so unless the console companies are happy to sit at 50-100 million users while content services on PCs and iThings are exposed to hundreds of millions of people, they need to have an eye on the low, mass-adoption price.

This part here is interesting:

"... necessitates either an amazing product (iPad) or a low price."

I'd argue, the top priority of the two is the "amazing product". Cheap price doesn't mean much. Dreamcast had a cheap price. Jaguar had a cheap price.

Conversely, iPod was pricey. iPhone was pricey. (even ps3 was pricey)

Desire for the product must first exist before price matters. At that point, price can be considered.


Getting to the tail end of your argument, I'm not seeing where in the rulebook it says lastgen consoles must die when new ones are introduced...

The best online, media-centric, casual-friendly box to match the intended use you've outlined that I can think of is ps3/xb360.


In summary, $400 BOM for nextgen boxes can pack quite a bit of silicon, can sell on the market near breakeven, and MSRP can withstand a prolonged limited price reduction (if necessary) due to no one ever being able to figure out a smaller node. ;)
 
Last edited by a moderator:
Let's evaluate a few other products with mass adoption:

iPhone = $599
iPad = $499
iPod = $399
Galaxy 2S = $599
Droid = $599

If it has value, these pricepoints have proven to be acceptable.

Your list also implies that average level of tech performance has also proven to be acceptable.
 
I'll refresh your memory here... xb360p has been selling for $299 since July 2008. FOUR YEARS.
Yep. That suggests, if you have something new to prop your device up with like Kinect, that you can add value and maintain sales. The thing with all the devices you are listing is that they are a year old. No-one's buying 4-year old phone tech for $400+. Will the consoumer be willing to spend $400 on a 3 year old console? Of course, if MS can sell like iThing, 50 million a year for two years, then they should go for it. However, the market for core gamers wanting the best if all of...30 million? You'd have 30 million wanting a powerful console. To everyone else, you'd either need a lower price or a box with value elsewhere. Like Kinect 2, or amazing services, or something, which is wher some of us are suggesting the console companies invest their efforts more than the hardware.
 
Coming back to it though, Cape Verde would still be highly disappointing. It barely trumps the exceedingly old 4890 from 2009 housed in my PC.


For a console, I think a 7770 level GPU would be a huge improvement. It would be interesting to see if we would see a APU + 7770 GPU though.

I think the 7770 gets really good performance with only a 72 GB/s of bandwidth. I think to get those kind of bandwidths out of DDR3, you'd need a pretty fat bus. 8 chips combined for a 256-bit bus at 2.8 Gb/s (fastest DDR3 I've seen) would net you ~89 GB/s. Sufficient for the Cape Verde GPU, but good enough for both CPU+CPU? I don't know.

If that's sufficient and you're going with 8 chips, then the question is whether 8 Gb or 4 Gb chips are available, cheap enough, and in volume for next year. Since DDR3 is such a high volume product, it may be possible to get 8 GB of DDR3.

I like this rumour very much :p, thanks to the confirmation bias (I was imagining 4GB + 4GB ddr3).
it is perfect. unless when you shrink and it becomes too much pad limited ; but new processes are coming so slow nowadays. I reckon you need at least something that can launch at 28nm, then shrink at 20nm : that lasts you for many year already. high volume 14nm or similar from TSMC is a long time ahead, considering we don't even have high volume 28nm as of yet.

I quoted the message about 4890, because like it or not many, most people don't even have a GPU that powerful. someone I know just bought a 6670 for instance (a PC with nice case, dual monitors, and SSDs to come), others may have an A8 llano or a laptop, others bought a card in 2009 but not a high end one, and then there are the many people for which the biggest GPU they own is the one in their current gen console.
 
Your list also implies that average level of tech performance has also proven to be acceptable.

There as nothing average about the performance of these devices for their form-factor...

Nobody expected a phone to be able to browse the net in a desktop like fashion and stream videos from youtube and run the plethora of apps possible ...

It's not just tech. It's tech, in what form factor, at what price, when.

Just as psp was remarkable for it's time. Not because nobody had seen subpar ps2 graphics before, but because nobody had seen them on a portable device that fits in your hands.
 
However, the market for core gamers wanting the best if all of...30 million? You'd have 30 million wanting a powerful console. To everyone else, you'd either need a lower price or a box with value elsewhere.

I agree.

Now I think Apple has clearly established the $500+ pricepoints as acceptable. Now it's just a matter of value.

There's one bit of value that MS could add which most consumers do not have a problem with spending $500+ for ... ;)
 
No, Apple has clearly established their own products to sell for $500+ price points as acceptable. Just because people are lining up in droves to by $600 ipads, doesn't mean those same people are going to be lining up in droves to pay the same for a prospective next Xbox. Or a PS3.
 
No, Apple has clearly established their own products to sell for $500+ price points as acceptable. Just because people are lining up in droves to by $600 ipads, doesn't mean those same people are going to be lining up in droves to pay the same for a prospective next Xbox. Or a PS3.

There's more than one way to sell a $500 device.

Sell it for $200 with a contract.
Sell it for $500 but throw in other valuables (Sony with BRD)
Sell it for $400 and eat the cost until profitable.

In MS' case, bundling Windows8 would be a sure fire way to get mass acceptance of a "High" price.

Bundling win8 and selling it for $200 with a two year contract would be even more appealing.

Point is, the spec of the box does not need to be in the toilet because nobody can figure out how to move a $500 BOM.
There are plenty of avenues to lure consumers other than slapping the pricetag on the box and wishing it well.


Some of you may be looking at that idea and think I'm insane .. "why would MS just give away Windows8?"

Windows generates ~$100/license
xblg generates $50/license ... every year.

The potential profits are staggering...

But ... gotta get the boxes past that pesky register first ... ;)

<insert $500 BOM Spec here>
 
Some of you may be looking at that idea and think I'm insane .. "why would MS just give away Windows8?"

I would say because including it costs nothing, it has already been developed, even porting it to PPC if needed is trivial.
now that's an idea, windows 8 on it feels desirable. it would be a revenue generator : people buying windows 8 phones and tablets (or even PC). there would be Windows 8 Metro at the least - or a combined Xbox / Metro interface, which runs Metro apps in any case.

MS can even sell the explorer.exe desktop with Word or Office. the new consoles will be powerful enough to serve as real desktops (but with draconian DRM).
PS3 was, sort of, but it was memory starved for that even at launch.
 
Status
Not open for further replies.
Back
Top