Next Generation Hardware Speculation with a Technical Spin [pre E3 2019]

[AlNom]

Is there an actual link to the pastebin post?

 

[goonergaz]

Is there an actual link to the pastebin post?

I assume this one;

https://pastebin.com/PY9vaTsR

 

[AlNom]

GPU: 14TF is reasonable when you consider variants of Vega64 can get close to that on an older node

Well, we should consider how much power is being consumed.

https://www.tomshardware.com/reviews/amd-radeon-rx-vega-64,5173.html

Vega 64 to operate at a base clock rate of 1247 MHz with a rated boost rate of 1546 MHz. Obviously, AMD's peak FP32 specification of 12.66 TFLOPS is based on that best-case frequency. We typically use the guaranteed base in our calculations, though. Even then, 10.2 TFLOPS is still an almost 20% increase over Radeon R9 Fury X.

The liquid-cooled model steps those numbers up to a 1406 MHz base with boost clock rates as high as 1677 MHz. That’s an almost 13% higher base and ~8%-higher boost frequency, pushing AMD’s specified peak FP32 rate to 13.7 TFLOPS. You’ll pay more than just a $200 premium for the closed-loop liquid cooler, though. Board power rises from 295W on Radeon RX Vega 64 to the Liquid Cooled Edition’s 345W—a disproportionate 17% increase.

If we go by a rough approximation of AMD's comments on 7nm TSMC for Zen, which indicates ~50% power savings at iso-performance, then we're looking at (maybe) 150W-175W for a 7nm rendition of a Vega 64 board, which mind you, uses 8GB HBM2. For the hypothetical console, we'd have to consider the power consumption for the GDDR6 bus with 24-32GB of RAM (edit: in regards to the pastebin description). Someone may want to chime in on approximate power consumption comparison on the memory side of things, but it's probably considerable at that amount of memory.

Memory: 24GB of GDDR6 on a 384bit bus has pretty much been established as the minimum for requisite bandwidth. 32GB would be lovely, but 24 pretty much seems to be the minimum.

24GB retail then 32GB devkit would imply using 12Gbit density chips vs 16Gbit density (in clamshell) on a 256-bit bus.

256-bit
@ 2chan x 16-bit I/O = 8 chips
@ 2chan x 8-bit I/O = 16 chips
8 chips x 12Gbit = 12GB
16 chips x 12Gbit (clamshell) = 24GB
8 chips x 16Gbit = 16GB
16 chips x 16Gbit (clamshell) = 32GB

-------

The 7nm Zen presentation indicated ~2x density when switching to TSMC 7nm, although it remains to be seen what other sorts of changes would be done for a gaming edition of the GPU architecture as double-precision rate isn't important here (etc. etc.), then swapping out the HBM2 interface for GDDR6.

 

[ultragpu]

I assume this one;

https://pastebin.com/PY9vaTsR

This link says $499 price tag tho. It's one thing to believe in hardware specs but pin point down launch titles from a small third party dev is quite a stretch to me. But hell as long as the 14 TF part is true I could care less about the rest.

 

[BRiT]

More importantly, claims of GTA 6 in 2020.

 

[BRiT]

This link says $499 price tag tho. It's one thing to believe in hardware specs but pin point down launch titles from a small third party dev is quite a stretch to me. But hell as long as the 14 TF part is true I could care less about the rest.

It says price of $499 with costs to Sony of $599 (a $100 loss) per console initially. Does anyone expect the console companies to switch their business models to take losses on hardware release now?

 

[ultragpu]

@AlBran

It says price of $499 with costs to Sony of $599 (a $100 loss) per console initially. Does anyone expect the console companies to switch their business models to take losses on hardware release now?

$100 loss sounds a bit steep but is it a necessary evil?

 

[goonergaz]

This link says $499 price tag tho. It's one thing to believe in hardware specs but pin point down launch titles from a small third party dev is quite a stretch to me. But hell as long as the 14 TF part is true I could care less about the rest.

Yeah, there's some inconsistencies - maybe there's an original and someone regurgitated it?

More importantly, claims of GTA 6 in 2020.

I suppose it's possible, but I did think it was a stretch...also means no true GTA for this gen?

It says price of $499 with costs to Sony of $599 (a $100 loss) per console initially. Does anyone expect the console companies to switch their business models to take losses on hardware release now?

Sony took a small loss at the start of this gen, it's nothing completely new...not sure they'd go to $100 though.

 

[BRiT]

A $100 loss per console at 5 million first year means a hit of $500 Million to Sony.
A $100 loss per console at 10 million first year means a hit of $1 Billion to Sony.
A $100 loss per console at 20 million first year means a hit of $2 Billion to Sony.

Yet another red flag.

 

[AlNom]

Indeed. One might imagine they could be making projections based on continued popularity in sales and at a price point that worked very well for them.

Of course, it would be nice if pricing caught up with inflation not unlike what happened between the PS2 and PS3 generation, however, we saw the folly of too high a price combined with a dual SKU strategy, which - as a Framing Effect modus operandum - perhaps pushed the average selling price higher as it conditioned people to the sweeter price point over the course of the generation.

^may or may not be an accurate analysis of what happened

Spoiler

in your feeble non-mod peasant opinions :V

, but it's just some passing thoughts/impressions of what I think possibly occurred there. Feel free to point out issues for discussion.

Basically, if they expect to sell very well again, then they may be even more risk averse to losing that much on HW as they would need to compensate for those losses on software royalties and subscriptions,

Spoiler

and sales of husbando fanfic dakimakura of Kratos x Joel x Nathan x Deacon (@London-boy *ahem* ) or Aloy x Ellie. :V

Re-Capturing the digital owner audience may also be fairly important in terms of backward compatibility, and that may or may not have further implications to their ability to support BC on a hardware level for a clean end-user experience.

 

[Tkumpathenurple]

It says price of $499 with costs to Sony of $599 (a $100 loss) per console initially. Does anyone expect the console companies to switch their business models to take losses on hardware release now?

I'll try to find the quote, but I recall a Sony exec stating that the PS4 was sold at a loss which would be recouped after the purchase of 1 game and a PS+ membership.

Maybe they have enough data to suggest that, on average, people buy 2 games and a PS+ membership? Something like that, anyway.

 

[BRiT]

I'll try to find the quote, but I recall a Sony exec stating that the PS4 was sold at a loss which would be recouped after the purchase of 1 game and a PS+ membership.

Maybe they have enough data to suggest that, on average, people buy 2 games and a PS+ membership? Something like that, anyway.

I only remember profitability relating to a game purchase, did not remember it also needing the network subscription too. Under $18 loss then assuming the 30% platform fee for 1 game. That can be worked through with shipping costs or packing cost improvements. Thets a vast difference to $100 loss.

 

[Shifty Geezer]

Googled "PS4 sold at a loss" -

https://www.eurogamer.net/articles/...loss-at-launch?&source=email_rt_mc_body&app=n

PlayStation 4 hardware will make a loss at launch, but Sony expects to immediately recoup the costs when a typical user also buys a PlayStation Plus subscription and games.

Sony Japan executive Masayasu Ito made the comment to Eurogamer today in an interview at the company's Tokyo headquarters.

Eurogamer has heard from well-placed sources that Sony expects to make an approximate $60 loss per $399 unit sold. When presented with the figure, Ito denied it - but only because the company expects a typical user to buy a console with these other items.

Ito would not be drawn on whether more than one launch title would need to be bought before the whole equation resulted in Sony making money, just that the company expected to turn a profit on an average user's initial purchase.

The purchase of one launch title and a Plus subscription would indeed total close to, if not more than, the $60 mark.

 

[goonergaz]

A $100 loss per console at 5 million first year means a hit of $500 Million to Sony.
A $100 loss per console at 10 million first yeat means a hit of $1 Billion to Sony.
A $100 loss per console at 20 million first year means a hit of $2 Billion to Sony.

Yet another red flag.

Well that's assuming the loss can't shrink quickly - I seem to recall with PS4 the small loss was only for the first couple months...not that I think Sony would lose $100 per unit, I just suspect the loss might be higher than PS4 as they are in a much stronger position this time around.

 

[BRiT]

So we know the 30% platform publisher fee margin, but what do we know about the "profit" margin on Network subscription? It obviously isnt 100% as online infrastructure has costs.

 

[BRiT]

Well that's assuming the loss can't shrink quickly.

Which is why I gave different numbers because a company could easily sell 5 million in the first quarter of launch, right? If you could shrink your loss in only 3 months time, why wouldnt you delay the launch by 3 months then instead and not take the loss?

 

[goonergaz]

Which is why I gave different numbers because a company could easily sell 5 million in the first quarter of launch, right? If you could shrink your loss in only 3 months time, why wouldnt you delay the launch by 3 months then instead and not take the loss?

ah, you just confused me saying $100 loss at the start of every line. It might be (like you say) for the first 3 months/5 million, then drop to $50/the PS4 break even after that...they might be happy to 'lose' that much to gain an advantage (maybe a head-start on MS?)

 

[Shifty Geezer]

Well that's assuming the loss can't shrink quickly.

What is the factor that will shrink quickly? A die shrink isn't going to happen quickly, I don't think. RAM prices could drop. If there's an SSD, that price could drop but not by $100 worth. I highly doubt PS4's loss disappeared within a couple of months of launch, although technically that's perhaps 6 months after first manufacture when losses are incurred for the initial units.

 

[goonergaz]

What is the factor that will shrink quickly? A die shrink isn't going to happen quickly, I don't think. RAM prices could drop. If there's an SSD, that price could drop but not by $100 worth. I highly doubt PS4's loss disappeared within a couple of months of launch, although technically that's perhaps 6 months after first manufacture when losses are incurred for the initial units.

Sony were making a profit just 6 months in;

https://gameranx.com/updates/id/22112/article/sony-now-making-a-profit-on-every-ps4-sold/

I'm not suggesting the whole $100, just enough that they get to a similar scenario where PS4 was at launch (ie, they just needs the customer to buy a game/get PS+ at the same time to break even).

 

[Shifty Geezer]

6 months makes more sense - maybe 10 months after initial manufacture where RAM prices likely dropped a lot. That was something Sony absolutely knew was going to happen when they selected 8GBs of GDDR5. You'd need a similarly predictable component cost reduction for a new machine if you intended a significant loss leader. A large loss dragged out over tens of millions of consoles is not going to be viewed favourable by any of these companies.

I think at best, a "one game and a year's sub" loss could be sustained if the difference that $50 brought to hardware made the difference and would win market share. It needs to be viewed in terms of "what do we gain from this $2 billion loss of clear profits?"