Predict: Next gen console tech (9th iteration and 10th iteration edition) [2014 - 2017]

[AlNom]

Hopefully not a rebrand of the 470.

If you're looking at the highly scientific timing of 3 years, there was the top end in 2012 was Tahiti, which became the highest mid-range part in 2015 as Tonga. Of course, there was no fab node change.

AMD's HBM offerings are a whole other class, but that level of HW (64CUs/64ROPs) would probably be mid-range with the shift to 7nmFF. It'd be interesting to see how much more cache they'd throw on there given the density improvement.

 

[Prophecy2k]

You guys are looking at cutting the wrong drive for a PS4 slim. Cut the optical drive.

Two drives have been handicapping console BOMs for a while now, but we will get back to one drive...just the hard drive, in the long term.

If one of the key drivers of a PS4 Super Slim, is extending the PS4's life by reaching much deeper penetration in price sensitive markets (i.e. emerging markets) then cutting the optical drive (i.e. the only realistic content delivery system for some territories) is not an option.

If all the PS4 super slim is achieve is the more price-conscious consumer in the three major markets (i.e. N.America, Eurozone and Japan) then sure, an optical drive-less machine makes sense.

 

[Kaotik]

If one of the key drivers of a PS4 Super Slim, is extending the PS4's life by reaching much deeper penetration in price sensitive markets (i.e. emerging markets) then cutting the optical drive (i.e. the only realistic content delivery system for some territories) is not an option.

If all the PS4 super slim is achieve is the more price-conscious consumer in the three major markets (i.e. N.America, Eurozone and Japan) then sure, an optical drive-less machine makes sense.

Also they could finally make it UHD-BD-compatible

 

[Technewsreader]

I feel like this gen got 2 major factors that made it everything but easy to show off a huge advancement in graphics early on in the cycle. To me personally difference between PS3 and PS4 can be huge in some titles like Uncharted 4 but in others its an "ok" to big difference.Maybe people with excellent 4K HDR screens think otherwise?
1. Most of the AAA going open world
2. Jump to pseudo or native 4K.

What I do expect is to have TheOrder or Uncharted 4 levels of detail (or better) in future Open world games. I wouldn't expect much more to be honest. Lets say if 1080P would've been the target resolution for 20 years Starting in 2006 and we would just have much improved HD screens instead of 4K then things would look a bit different next-gen.

 

[Prophecy2k]

I feel like this gen got 2 major factors that made it everything but easy to show off a huge advancement in graphics early on in the cycle. To me personally difference between PS3 and PS4 can be huge in some titles like Uncharted 4 but in others its an "ok" to big difference.Maybe people with excellent 4K HDR screens think otherwise?
1. Most of the AAA going open world
2. Jump to pseudo or native 4K.

What I do expect is to have TheOrder or Uncharted 4 levels of detail (or better) in future Open world games. I wouldn't expect much more to be honest. Lets say if 1080P would've been the target resolution for 20 years Starting in 2006 and we would just have much improved HD screens instead of 4K then things would look a bit different next-gen.

Horizon Zero Dawn is one of, if not, the best looking console game to date and its open world. Also, 4k didn't arrive until 3 yrs in, so wouldn't have impacted dev's ability to show off next-gen visuals "early in the gen.".

There's always variation in visual fidelity across games in each generation. Last gen, few games looked as good as God Of War 3 or Halo 4.

Game budgets and developer skill are the main factors that contribute towards game being able to fully take advantage of the hardware and show off top-tier visuals. And if you're talking about early in the gen, then target platforms must be taken into account, since many games released over the first couple of years were cross-gen (thus had to run on systems with 512MB of RAM).

 

[Technewsreader]

Horizon Zero Dawn is one of, if not, the best looking console game to date and its open world. Also, 4k didn't arrive until 3 yrs in, so wouldn't have impacted dev's ability to show off next-gen visuals "early in the gen.".

There's always variation in visual fidelity across games in each generation. Last gen, few games looked as good as God Of War 3 or Halo 4.

Game budgets and developer skill are the main factors that contribute towards game being able to fully take advantage of the hardware and show off top-tier visuals. And if you're talking about early in the gen, then target platforms must be taken into account, since many games released over the first couple of years were cross-gen (thus had to run on systems with 512MB of RAM).

Yes 4K introduction wasn't that early but still in the first half of the gen and you have a point with horizon but the reason i haven't mentioned it is because I heard that it is not a very interactive Open World so I don't expect it to have much dynamic physics. It is no Watch Dogs 2 which btw looks pretty good to me after playing the demo. I'm not that much of a "graphics nerd" to begin with but I have nothing against seeing results on Final hardware. It does excite me and know i'm not alone.

 

[Tkumpathenurpahl]

Was that a threat?

 

[mpg1]

https://www.globalfoundries.com/sites/default/files/product-briefs/7lp-product-brief.pdf

• >40% performance improvement at iso power (vs. 14nm)
• >60% power reduction at iso frequency (vs. 14nm)
• Up to 30% lower die cost (vs. 14nm)

..................................................well then.

/deletes everything

Their website makes it sound like it's either or when it comes to the perf and power reduction:
https://www.globalfoundries.com/technology-solutions/cmos/performance/7nm-finfet

"This technology provides world-class performance, power, area and cost advantages from 7nm scaling. Based on 3D FinFET transistor architecture and optical lithography with EUV compatibility at key levels, 7LP technology delivers more than twice the logic and SRAM density, and either >40% performance boost or >60% total power reduction, compared to 14nm foundry FinFET offerings."

 

[AlNom]

Right. My issue was that it seemed like a relatively small improvement when you're looking at TSMC's jump to 10nm and then 10nm to 7nm.

Of course they could have bungled the language:

GF 30% lower die cost ->7nm chip costs 70% of 14nm fabricated chip.

TSMC -> 2x density (16 -> 10) & 1.6x (10 -> 7) = 3.2x density -> ~30% of chip area (16nm -> 7nm)

Unless TSMC's much smaller chip size has an inflated cost compared to GF (both fabs are comparing different things, indirectly), GF's marketing isn't quite doing a great job.

i.e. GF is comparing chip cost vs TSMC pimping a substantial improvement to density (which still may end up costing comparatively to GF's offering at this time, but who knows, and how does that even make sense given that costs are always changing according to volume & maturity).

------------

Power-wise, TSMC claims >40% for each step to 7nm. (1.4*1.4)

------------

And yeah, usually you pick either-or just , but usually we're on the same page on how they compare the stats. TSMC website blurb does indeed discuss the mobile process & the high perf. process for 7nm, so I imagine it's a similar pick & choose characteristic on there.

 

[iroboto]

Where's a ~mid-range AMD gpu going to be in 2 years? Will it even be at gtx 1080 level of perf?

I don't believe so. There's a lot of assumptions that would have to come true for AMD to deliver so perfectly in the next 2 years. I suspect a 1070 level would be a more conservative target to achieve.

Assume linear console price drop of $50 per year.
In 2 years, Scorpio would be $399 8c Jaguar/6TF w/ 12GB GDDR5.

That should put things into perspective imo.

Any new licensing deals for new technologies would increase costs imo, at least over the existing licenses that they own with Jaguar today.

 

[mpg1]

I don't believe so. There's a lot of assumptions that would have to come true for AMD to deliver so perfectly in the next 2 years. I suspect a 1070 level would be a more conservative target to achieve.

Assume linear console price drop of $50 per year.
In 2 years, Scorpio would be $399 8c Jaguar/6TF w/ 12GB GDDR5.

That should put things into perspective imo.

Any new licensing deals for new technologies would increase costs imo, at least over the existing licenses that they own with Jaguar today.

As I said before I don't know if Scorpio is a great comparison when it comes to price. It's a low volume "premium product" and Microsoft did a little extra to achieve those clocks than what would be seen on a more mass market console. I don't expect Sony to do things like vapor chamber cooling and Hovis method.

Sony will target a price point of $399 and will rely on gains from the chip rather than "overclocking".

 

[MrFox]

As I said before I don't know if Scorpio is a great comparison when it comes to price. It's a low volume "premium product" and Microsoft did a little extra to achieve those clocks than what would be seen on a more mass market console. I don't expect Sony to do things like vapor chamber cooling and Hovis method.

Sony will target a price point of $399 and will rely on gains from the chip rather than "overclocking".

Hovis seems to be microsoft rebranding AMD's AVFS. Vapor chamber is just a flat heat pipe. It's necessity is based on power density, which will be higher at 7nm and above. So I expect both next gen regardless of the brand.

The Xb1x price must be more related to the uhd bluray and the additional ram chips, along with lower yield caused by the higher number of I/O, slightly larger soc, and higher clock. Also the low airflow to the power supply means they needed a more expensive one (higher efficiency). It's easy to reach 100 or so with these points alone.

It's the year delay which puzzles me. Maybe AVFS requires a lot more time to design sensing "loops" all around the chip. Maybe yield improved this year.

 

[BRiT]

The Xb1x price must be more related to the uhd bluray and the additional ram chips, along with lower yield caused by the higher number of I/O, slightly larger soc, and higher clock.

IIRC, The UHD drive was at most $24 when the Xbox One S was released a year ago if a 2TB drive costs $0 more.

EDIT: It was said to be $15.50 more, at $33.50 total compared to $18.

FROM: http://www.businesswire.com/news/home/20160805005103/en/IHS-Markit-Teardown-Analysis-Microsoft’s-Xbox-Brings

The combined hardware and manufacturing cost of the new Xbox One S game console from Microsoft, the new analysis says, amounts to $324. That’s just $24 more than the most recent Xbox One teardown analysis conducted in March of this year, primarily due to upgrades in the fabrication of its central processing unit (CPU), the inclusion of a two terabyte (2TB) hard drive and an ultra-high-definition (UHD) BD-ROM player in place of the previous conventional Blu-ray drive.

The new console also includes a UHD BD-ROM player, with a bill of materials (BOM) cost of $33.50, versus the previous $18 conventional Blu-Ray drive.

 

[mpg1]

Hovis seems to be microsoft rebranding AMD's AVFS. Vapor chamber is just a flat heat pipe. It's necessity is based on power density, which will be higher at 7nm and above. So I expect both next gen regardless of the brand.

The Xb1x price must be more related to the uhd bluray and the additional ram chips, along with lower yield caused by the higher number of I/O, slightly larger soc, and higher clock. Also the low airflow to the power supply means they needed a more expensive one (higher efficiency). It's easy to reach 100 or so with these points alone.

It's the year delay which puzzles me. Maybe AVFS requires a lot more time to design sensing "loops" all around the chip. Maybe yield improved this year.

I'm not claiming that it's mind-blowing tech but my point was I'm not sure we can 1-to-1 extrapolate Xbox One X's design philosophy towards a "normal" high volume console.

It has clearly added some cost to design and manufacturing process in order to achieve those clock speeds...

 

[Globalisateur]

Hovis seems to be microsoft rebranding AMD's AVFS. Vapor chamber is just a flat heat pipe. It's necessity is based on power density, which will be higher at 7nm and above. So I expect both next gen regardless of the brand.

The Xb1x price must be more related to the uhd bluray and the additional ram chips, along with lower yield caused by the higher number of I/O, slightly larger soc, and higher clock. Also the low airflow to the power supply means they needed a more expensive one (higher efficiency). It's easy to reach 100 or so with these points alone.

It's the year delay which puzzles me. Maybe AVFS requires a lot more time to design sensing "loops" all around the chip. Maybe yield improved this year.

It doesn't puzzle me. I still think Microsoft panicked and decided to make the XBX when they first heard about PS4 Pro. Probably during the first half of 2016.

I'm ready to eat crow.

 

[MrFox]

It doesn't puzzle me. I still think Microsoft panicked and decided to make the XBX when they first heard about PS4 Pro. Probably during the first half of 2016.

I'm ready to eat crow.

I know you're joking, but it's still very weird that sony had a new console so soon when clearly they didn't need one, it took a lot of people by surprise. The "panic" was more a change of PR for MS, taking the risk of announcing scorpio that early. They had no idea of each others plans since these consoles need years to develop.

 

[iroboto]

Hovis seems to be microsoft rebranding AMD's AVFS.

i'm not sure if this is accurate, or at least, there seems to be legality odds here. Hovis method, is named after Hovis, the MS engineer to made the technique. If it was a rebrand of AMD tech, I think they be in legal deep water.

While I'm not entirely sure about what I'm writing about, a litmus test is to see if we can find a patent on it. MS usually doesn't go head first like this without their own patents either.

My buddy at MS used to do patents for them, and when you successfully make a patent for MS you get some rewards, and a nice little cube on your desk.
This guy:
https://www.linkedin.com/in/bill-hovis-12704076

Willing to bet he got one.

 

[iroboto]

I know you're joking, but it's still very weird that sony had a new console so soon when clearly they didn't need one, it took a lot of people by surprise. The "panic" was more a change of PR for MS, taking the risk of announcing scorpio that early. They had no idea of each others plans since these consoles need years to develop.

There is 1 scenario in which Global could be right, since you are correct that development of these consoles take _way_ too long. And we can see that when MS announced Scorpio at E3, the motherboard looks largely the same to the final product. Anyway the scenario:

• Both MS and Sony were both experimenting with 2 variants of their Pro models.
• Phil admits to working on two separate versions, with Sony we only got rumours of it.
• Both had a choice to release a console in 2016 or wait until 2017 for a more powerful version.
• When rumours swirled out about the Sony going to release in 2016, MS green lit the 2017 model.

That would be about as far as I can see a company 'reacting' to news on the internet. If this scenario was true, It's a pretty big risk since Sony announced the Pro after Scorpio. They could have easily eaten crow there.

anyway, I don't want to encourage this type of thought process. Too many what ifs to be a highly probable scenario. Too much out there about MS and Sony doing their own thing, that none of it seems like reacting, except the PR piece. I do recall Phil mentioning it was a cost thing which caused Scorpio to be released this year. I do wonder if it's purely manufacturing/licensing agreements. You don't know what each company signed, so that could lend to a lot of differences in what and when their products are released.

 

[mpg1]

Paraphrasing but Phil Spencer essentially said that when they looked at doing a 4K console they couldn't do it with what was available for 2016.

 

[phoenix_chipset]

I don't believe so. There's a lot of assumptions that would have to come true for AMD to deliver so perfectly in the next 2 years. I suspect a 1070 level would be a more conservative target to achieve.

Assume linear console price drop of $50 per year.
In 2 years, Scorpio would be $399 8c Jaguar/6TF w/ 12GB GDDR5.

That should put things into perspective imo.

Any new licensing deals for new technologies would increase costs imo, at least over the existing licenses that they own with Jaguar today.

1070? Xbox 1 X is already very close to 1070 in flops. Half the problem with AMD gpu's are their DX11 drivers, which wouldn't apply to a console right? Seems to me Nvidia's advantage is mostly on the software side of things in terms of performance.

On PC rx vega 64 is comparable to 1080 and vega 56 seems to be better than 1070.