• Ever wanted an RSS feed of all your favorite gaming news sites? Go check out our new Gaming Headlines feed! Read more about it here.
  • We have made minor adjustments to how the search bar works on ResetEra. You can read about the changes here.

Which team are you on?

  • Double Team (1997)

  • Team Walnut

  • The A-Team

  • Team "No One Can Stop Mr. Domino"

  • Sports Team

  • "I'm a loner, Dottie. A rebel."

  • Team Margarita


Results are only viewable after voting.
Status
Not open for further replies.

uncleniccius

Member
Nov 3, 2017
1,082
Wired and wireless !!!!

Mcbazel Brook X One Adapter für Xbox One Controller auf PS4 Nintendo Switch mit Remap Turbo Wireless Connection und Wiederaufladbarem Akku mit Gam3Gear Schlüsselanhänger: Amazon.de: Games

Mcbazel Brook X One Adapter für Xbox One Controller auf PS4 Nintendo Switch mit Remap Turbo Wireless Connection und Wiederaufladbarem Akku mit Gam3Gear Schlüsselanhänger - Kostenloser Versand ab 29€. Jetzt bei Amazon.de bestellen!

Brook X ONE Adapter Wireless Converter and Rechargeable Battery making your XBOX ONE and Elite Controller Compatible with PS4, Nintendo Switch, XBOX One Console and PC: Amazon.co.uk: Computers & Accessories

Brook X ONE Adapter Wireless Converter and Rechargeable Battery making your XBOX ONE and Elite Controller Compatible with PS4, Nintendo Switch, XBOX One Console and PC: Amazon.co.uk: Computers & Accessories
Well , fuck. If next gen wasn't so close I'd be buying that - thanks for mentioning it I had no idea!
 

riotous

Member
Oct 25, 2017
11,347
Seattle
You guess they would have to know the amount of power they are ultimately designing the games for, no matter if the devkits version change and so does the current power they have at disposal, but maybe they just rely on the devkit they have and with each version they restart the designing. Must be that...
What? Where did I suggest any of this?
 

DrKeo

Banned
Mar 3, 2019
2,600
Israel
I don't see why this has to be true, or why so many here seem to think disabling WGPs, SAs or SE is something that has to be done in "real-time" at the application level.

We're talking about a BC mode. It's a driver level change, forcing the system into a specific mode for BC. I wonder whether it can be done through the shader compiler, to make it subdivide workgroups into wavefronts that address only a subset of the available resources; so simply not dispatching work to certain WGPs, rather than actively switching them off.

When CUs/WGPs are disabled in the Fab, a similar change must be implemented to ensure wavefront workloads aren't dispatched to the disabled/fused WGP/CU. The control logic to enable it should be in-built in the design, as the nature of disabling defective CUs/WGPs means that every WGP/CU needs to include the control logic to disable/enable them.
Doesn't the command processor dispatch directly to the dual compute units level? It would be as simple as changing the microcode inside the command processor (or even just some sort of BIOS) during manufacturing.
AFAIK in RDNA the lowest component the Command Processor can assign commands to is the SA. The SAs are independent, the CP distributes and dispatches workload between the SAs and they independently distribute them between their WGPs. It's one of the reasons RDNA is supposed to be flexible. If you turn off a WGP, the SA is the one that needs to handle it, the CP doesn't even need to know that a WGP was disabled. If an RDNA GPU wants to go into GCN BC mode (which exists in every RDNA based card), the WGP knows how to split the resources between the two CUs as if they were two separate CUs.

That's at least my understanding of RDNA.

This last part is me guessing, but IMO if you want to turn off just a few specific CUs inside a SA in rea-time, it will have to be a hardware level change, just like the WGP -> 2 CUs in BC mode is. What I mean by that is IMO, if Sony wants the ability to do that, it will be part of their "special sauce".

cc DukeBlueBall Colbert
 
Last edited:

Gdourado

Member
Oct 1, 2018
139
Lisbon, Portugal
If it's 9 vs 12 it will be no different than what you have in current gen. One with higher resolution and a better framerate in some games, that's about it.

that Is my point.
I have a One X and a PS4 Pro.
And the best looking games for me are God of War, Uncharted 4, lost legacy and Horizon.
So as much as I would like a 12 tflop ps5, I think Sony first party studios will probably do great things with 9...
 
Feb 10, 2018
17,534
I kinda wanna see an under-powered PS5 purely for the reactions.

And I love their first party output

Reactions from both sides are going to be awful and depressing.


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


On a side note, even though I don't believe random things like this, as literally anyone can post them. However just for the fun of it if these were the specs what would the difference actually be in games?

playstation-5-xbox-series-x-specs.jpg
 

Fezan

Member
Oct 26, 2017
3,274
Just read the article it was whole bunch of nothing. I expected at least something closer to real specs or confirmation or denial of things but nothing.
 

Deleted member 12635

User requested account closure
Banned
Oct 27, 2017
6,198
Germany
AFAIK in RDNA the lowest component the Command Processor can assign commands to is the SA. The SAs are independent, the CP distributes and dispatches workload between the SAs and they independently distribute them between their WGPs. It's one of the reasons RDNA is supposed to be flexible. If you turn off a WGP, the SA is the one that needs to handle it, the CP doesn't even need to know that a WGP was disabled. If an RDNA wants to go into GCN BC mode (which exists in every RDNA based card), the WGP knows how to split the resources between the two CUs as if they were two separate CUs.

That's at least my understanding of RDNA.

This last part is me guessing, but IMO if you want to turn off just a few specific CUs inside a SA in rea-time, it will have to be a hardware level change, just like the WGP -> 2 CUs in BC mode is. What I mean by that is IMO, if Sony wants the ability to do that, it will be part of their "special sauce".

cc DukeBlueBall Colbert
Maybe I misinterpreted the white paper. WGP indeed would be uncommon.
 

Thorrgal

Member
Oct 26, 2017
12,349
People question the GitHub info as relevant now in February 2020 not only because it's old, and not only because Matt has already told us to move past it, but because it is nonsensical.

The PS5 GPU won't be one with 36CU at 2Ghz. It will either have more CU, lower clocks, or most likely both.
 

Psyrgery

Member
Nov 7, 2017
1,745
if the $450 BOM is true then Sony had an extra $69 to play with, with $66 of that going into keeping the die size the same, im not sure how they could afford to pay for extra cooling, ssd and gddr6.

This sadly points to a smaller die. maybe 300mm2 which should cost them around $120 and allow them to invest the rest into extra costs for cooling and the higher prices for ssd and vram.

Serious question: how do we know if the $450 figure is believable and not made up or just someone's theory/estimation.

It looks to me as if everybody is taking this number as gospel whereas everything else is questioned and discussed to death
 

DrKeo

Banned
Mar 3, 2019
2,600
Israel
Serious question: how do we know if the $450 figure is believable and not made up or just someone's theory/estimation.

It looks to me as if everybody is taking this number as gospel whereas everything else is questioned and discussed to death
It was also confirmed by ZhugeEX on twitter, that's why everyone believes the number. But these numbers are an estimate, ZhugeEX's estimate for XSX is 460$ - 520$, a pretty big range. So you shouldn't take 450$ as a final number, it's just a ballpark estimate, just like 490$ is the XSX's ballpark if we average out the 460$ - 520$ figure.
 

褲蓋Calo

Alt-Account
Banned
Jan 1, 2020
781
Shenzhen, China
On a side note, even though I don't believe random things like this, as literally anyone can post them. However just for the fun of it if these were the specs what would the difference actually be in games?

playstation-5-xbox-series-x-specs.jpg
It will be huge. The small difference will only be exacerbated by the fact that PS5 progammers hand code in GPU assembly whereas on Xbox they have to rely on bloated API, locking even more performance behind the door.
/jk
 

FF Seraphim

Member
Oct 26, 2017
13,737
Tokyo
Reactions from both sides are going to be awful and depressing.


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


On a side note, even though I don't believe random things like this, as literally anyone can post them. However just for the fun of it if these were the specs what would the difference actually be in games?

playstation-5-xbox-series-x-specs.jpg

For one thing with that power we might be able to play basketball in the next Spider-man game.
 

DrKeo

Banned
Mar 3, 2019
2,600
Israel
Reactions from both sides are going to be awful and depressing.


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


On a side note, even though I don't believe random things like this, as literally anyone can post them. However just for the fun of it if these were the specs what would the difference actually be in games?

playstation-5-xbox-series-x-specs.jpg
There will be zero difference, this meaningless difference will be swallowed by the small performance buffer that exists in order for games to run at locked framerates. It will be the closest two consoles have ever been.
 

disco_potato

Member
Nov 16, 2017
3,145
I could see 550 for the XSX and 499 for the PS5. BUT the existence of Lockhart means that Microsoft could eat the losses of the XSX via gamepass, software sales, and money made from Lockhart assuming they sell that console for profit.
A premium product is designed to make a profit not to be sold as a loss. That's what lockhart would be for.
 

DrKeo

Banned
Mar 3, 2019
2,600
Israel
A premium product is designed to make a profit not to be sold as a loss. That's what lockhart would be for.
Thing is, XSX isn't a premium product, it's the direct competitor to the PS5. XSX is the "norm", XSS is the unique product here. I'm not sure XSX exists in order to make more money on hardware, that's what a 2TB SKU is for, it's MS's main console. We've never seen a product like that, the rules have changed.
 

Psyrgery

Member
Nov 7, 2017
1,745
It was also confirmed by ZhugeEX on twitter, that's why everyone believes the number. But these numbers are an estimate, ZhugeEX's estimate for XSX is 460$ - 520$, a pretty big range. So you shouldn't take 450$ as a final number, it's just a ballpark estimate, just like 490$ is the XSX's ballpark if we average out the 460$ - 520$ figure.

Noted, thanks for the heads up!
 

thuway

Member
Oct 27, 2017
5,168
Reactions from both sides are going to be awful and depressing.


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


On a side note, even though I don't believe random things like this, as literally anyone can post them. However just for the fun of it if these were the specs what would the difference actually be in games?

playstation-5-xbox-series-x-specs.jpg
This is just about as much nonsense as the Github leak.
 

AegonSnake

Banned
Oct 25, 2017
9,566
Going back to the 66% markup Richard talked about for APU die, i wanted to look at what that means for lockhart because all of a sudden $300 doesnt seem too plausible.

Some assumptions:
  • PS4 28nm 350mm2 die = $100 = $10 per 35mm2
  • 7nm 60% markup = $16 per 35mm2
  • RX5500xt = 158mm2.
  • Zen 2 = 40mm2
  • Misc = 20 mm2
  • Total Lockhart APU Size = 218mm2
  • Lockhart fab cost = 218mm2/35mm2 * $16 = $99.65
We are basically at the same die cost as the PS4. The only place where they will save some money is the $30 on the disc drive. That puts them at $370, but the nand and dram costs increase will affect their prices just as much as the ps5 and xsex. regardless, even at $370 they are only $11 less than the PS4 which was sold at a loss and needed a first party game sale per console to break even.

I really dont see how that is going to be a $299 console unless MS takes a $150 loss on it.
 

AegonSnake

Banned
Oct 25, 2017
9,566
Serious question: how do we know if the $450 figure is believable and not made up or just someone's theory/estimation.

It looks to me as if everybody is taking this number as gospel whereas everything else is questioned and discussed to death
i did say IF IT WAS TRUE.

all we are doing here is questioning and discussing it to death like we do everything else. it will run its course. DF just made a new article on the bloomberg article which is why people are discussing it. no one is taking it as gospel.
 

III-V

Member
Oct 25, 2017
18,827
In a completely custom, co-designed chip it seems logical you might be able to get an extra command from the CP for this specific purpose. I don't see why not. It's not a command that would normally be required for desktop GPU so the fact it doesn't exist in the existing RDNA instruction set isn't much of a shocker. Just my 2c.

It is nonsensical to imagine we are locked into a 36 CU (or multiples thereof 😂) for the future generations of PS5. What are they going to do? PS6 at 4.3 GHz.

Let's Do Better than try our best to fit this old narrative.
 
Last edited:

El-Pistolero

Banned
Jan 4, 2018
1,308
Just wanted to say that consoles are not plastic boxes. They are a wonderful pinnacle of human capabilities! Amazing technology to the service of artists that give people so much joy in their lives. Just because some childish console warrior out there is using them to somehow validate his insecurities, do not despise consoles calling them plastic boxes. Instead, you should call some people meat boxes! 😂

This isn't serious, is it? oO
 

DukeBlueBall

Banned
Oct 27, 2017
9,059
Seattle, WA
Going back to the 66% markup Richard talked about for APU die, i wanted to look at what that means for lockhart because all of a sudden $300 doesnt seem too plausible.

Some assumptions:
  • PS4 28nm 350mm2 die = $100 = $10 per 35mm2
  • 7nm 60% markup = $16 per 35mm2
  • RX5500xt = 158mm2.
  • Zen 2 = 40mm2
  • Misc = 20 mm2
  • Total Lockhart APU Size = 218mm2
  • Lockhart fab cost = 218mm2/35mm2 * $16 = $99.65
We are basically at the same die cost as the PS4. The only place where they will save some money is the $30 on the disc drive. That puts them at $370, but the nand and dram costs increase will affect their prices just as much as the ps5 and xsex. regardless, even at $370 they are only $11 less than the PS4 which was sold at a loss and needed a first party game sale per console to break even.

I really dont see how that is going to be a $299 console unless MS takes a $150 loss on it.

One correction, die size and die cost isn't linear. 350mm2 die is $10 per 35 mm2. 200m2 die might be only $8 per 35mm2.
 

Thera

Banned
Feb 28, 2019
12,876
France
Reality is that MS as an entertainment brand is/has always been less successful than Sony and Nintendo. MS business wise does very well, don't get me wrong, but even though I think they have an awesome product and imo have pushed the gaming medium more than anyone else in the last few gens with new features, they are still not as popular as Nintendo and Sony because of perception. I think having the upper hand hardware wise will give them a boost in sales without necessarily dominating and will help balance things out a bit more, meaning more games will make their way to the Xbox/PC ecosystem.
I agree about the push on HDD and online (even if they had fee to it), but this gen ?

You more answered a "how" than a "why".
Why YOU, do want that to happened ?
 

DrKeo

Banned
Mar 3, 2019
2,600
Israel
Going back to the 66% markup Richard talked about for APU die, i wanted to look at what that means for lockhart because all of a sudden $300 doesnt seem too plausible.

Some assumptions:
  • PS4 28nm 350mm2 die = $100 = $10 per 35mm2
  • 7nm 60% markup = $16 per 35mm2
  • RX5500xt = 158mm2.
  • Zen 2 = 40mm2
  • Misc = 20 mm2
  • Total Lockhart APU Size = 218mm2
  • Lockhart fab cost = 218mm2/35mm2 * $16 = $99.65
We are basically at the same die cost as the PS4. The only place where they will save some money is the $30 on the disc drive. That puts them at $370, but the nand and dram costs increase will affect their prices just as much as the ps5 and xsex. regardless, even at $370 they are only $11 less than the PS4 which was sold at a loss and needed a first party game sale per console to break even.

I really dont see how that is going to be a $299 console unless MS takes a $150 loss on it.
An APU cost doesn't scale linearly with size. The bigger a GPU is, the lower the yields and the more silicon is wasted on the rims of the wafer. You should also include the fact that Lockheart will probably have a lower clock speed which will lower the BOM even more. It's actually the other way around IMO, if XSX's APU really is a 200$ chip, Lockheart's APU will probably cost a lot less, probably close to 100$. Add to that a smaller form factor, cheaper cooling, 4GB less RAM, slower and cheaper RAM, half the size of the SSD and no BR drive and it's BOM difference might even be higher than 150$.
 

BreakAtmo

Member
Nov 12, 2017
12,838
Australia
I wouldn't be so sure about that. You'd be surprised how many casual players I know that called me up and said "why is this telling me I don't have enough space to install this." Also, anything to reduce the install size of the game would help people in places that don't have great internet.

Oh, I wouldn't be surprised at all. I really dislike all the bloated install sizes as well, and I really appreciate games like Sekiro that keep that shit to a minimum. I just wonder what would happen if you told them "I have a way to shrink the game, but it's not going to run as well". You might be right, though.

Hell, imagine this: AI upscaling of textures after the initial data has been downloaded. It won't help with the install size, but it would help with the download size.

Fucking gimme
 

PianoBlack

Member
May 24, 2018
6,647
United States

DrKeo

Banned
Mar 3, 2019
2,600
Israel
In a completely custom, co-designed chip it seems logical you might be able to get an extra command from the CP for this specific purpose. I don't see why not. It's not a command that would normally be required for desktop GPU so the fact it doesn't exist in the existing RDNA instruction set isn't much of a shocker. Just my 2c.

It is nonsensical to imagine we are locked into a 36 CU (or multiples thereof 😂) for the future generations of PS5. What are they going to do? PS6 at 4.3 GHz.

Let's Do Better than try our best to fit this old narrative.
I totally agree, it could be a custom Sony request, part of their "secret sauce".

Regarding the PS6 and future consoles, at some point Sony will have to run these BC games on a VM, they can't fit their console hardware design to transform into every legacy X86 console they've ever made.
 
Last edited:

GING-SAMA

Banned
Jul 10, 2019
7,846
User Banned (2 Weeks): Console Warring
I don't understand why people keep speculating when we all know the Series X is going to be more powerful.

/s
 

Lady Gaia

Member
Oct 27, 2017
2,480
Seattle
This isn't serious, is it? oO

What's it to you, meat box? /s

There is something to the post in question, as it observes that the technology here isn't well represented by the reductive "plastic box" nomenclature. The amount of R&D and degree of artistic toil that goes into making a modern gaming experience possible is pretty remarkable. Where the false equivalence kicks in is that people are highly individualistic, whereas game consoles and games are mass-produced functional replicas based on a shared blueprint. You can discard the box and disc and replace them with identical equivalents trivially, which is what makes them just "plastic boxes" in a way that people aren't just "meat bags."
 
Last edited:
Status
Not open for further replies.