News AMD to design processor for Xbox Next: Team Red extends long-standing Microsoft partnership

[-Fran-]

Mostly it's about AMD being the only one willing to work for pittance.

But if Intel would have been announced instead, you'd be mocking AMD for not being chosen, right? 😀

Come on, Terry.

They can but the issue is that Apple is so far ahead of everyone with ARM it’s scary.

Biggest win here is the cross store thing they’re doing which is a big issue for Sony

First* point is a non-point: Apple Silicon only serves Apple. Let me know when Apple starts designing hardware to run in other platforms intentionally.

Your second point has merit to be mentioned: "XBox everywhere" as an approach is not a bad idea, but undermines the hardware division a bit, which this news is all about. The console now is going to be, at best, an entry experience to the XBox ecosystem and games instead of being "the product". A bit of semantics and all, but I think the point is not to hard to get. I agree with what you said here though. Sony relying on exclusivity is going to put them on an island, next to Nintendo, with some titles. This being said, Sony has made it clear they are looking to port 1st party titles to PC in an orderly manner, so there's that. Nintendo is the only one still in that island without the will to build bridges.

Regards.

 

[TerryLaze]

But if Intel would have been announced instead, you'd be mocking AMD for not being chosen, right? 😀

Come on, Terry.

Sure I would laugh at AMD but I would also say that it would be a terrible deal for intel and that they only did it to hurt AMD ,taking a loss to do so.

 

[-Fran-]

Sure I would laugh at AMD but I would also say that it would be a terrible deal for intel and that they only did it to hurt AMD ,taking a loss to do so.

While a reasonable assumption, given past Intel and MS dealings, I wouldn't be so sure it's not convenient for AMD from the income side. It's a, somewhat, steady income source and it's definitely good to have. As I said before: it'll depend on what MS really wants to do. If they want to cheap out, then AMD can just give them an "off the shelf* " design they probably have in a bin, somewhere. Think about what MS (Xbox) is looking for: a cheap, very cost-effective, APU to just be capable to stream and play games on the go and, guess what, that already exists so AMD can just give them any APU design and ask MS if they want to tweak any IP in it.

For AMD, specifically, it may be very low cost to get into this venture with MS compared to any company which have not been building itself as a "design house". From memory, I don't recall many others doing that. The biggest one being nVidia, but we all know how dealing with them is like. Qualcomm? LOL, no. Apple? LOL x2, no. Intel? They could, but much like @bit_user said: they want higher margin parts and that would require a bigger learning curve for them as they'd want to use their own fabs?

Which is another interesting bit*: the fabs and what MS+AMD can do there. I'm guessing TSMC in the USA as an initial thought, so that tells you what process they'll be looking to use and similar designs AMD already has out there in the wild for them. Otherwise, outside of the USA, it can be any process where, to me, N3X would make the more sense.

Regards.

 

[LordVile]

But if Intel would have been announced instead, you'd be mocking AMD for not being chosen, right? 😀

Come on, Terry.


First* point is a non-point: Apple Silicon only serves Apple. Let me know when Apple starts designing hardware to run in other platforms intentionally.

Your second point has merit to be mentioned: "XBox everywhere" as an approach is not a bad idea, but undermines the hardware division a bit, which this news is all about. The console now is going to be, at best, an entry experience to the XBox ecosystem and games instead of being "the product". A bit of semantics and all, but I think the point is not to hard to get. I agree with what you said here though. Sony relying on exclusivity is going to put them on an island, next to Nintendo, with some titles. This being said, Sony has made it clear they are looking to port 1st party titles to PC in an orderly manner, so there's that. Nintendo is the only one still in that island without the will to build bridges.

Regards.

Point was that the only ARM chips that perform well and are worth a damn at the moment for these applications are from Apple so building an ARM console on chips that are subpar at best is pointless.

Little bit different, the fact you can just access steam from an Xbox will be a game changer rather than being trapped in Sonys store. It also positions Xbox to be the steam machine valve wanted but could never actually accomplish or maybe they just straight up allow you to install the OS on a PC down the line. PC gaming made couch friendly and all that.

Nintendo has always been a bit different because they haven’t relied on 3rd party titles for a while. All they need to do for a console to be successful is Mario kart, Smash, 3D Mario, one or two Zelda’s and a handful of second string franchises like Kirby and some remasters.

 

[TerryLaze]

While a reasonable assumption, given past Intel and MS dealings, I wouldn't be so sure it's not convenient for AMD from the income side. It's a, somewhat, steady income source and it's definitely good to have. As I said before: it'll depend on what MS really wants to do. If they want to cheap out, then AMD can just give them an "off the shelf* " design they probably have in a bin, somewhere. Think about what MS (Xbox) is looking for: a cheap, very cost-effective, APU to just be capable to stream and play games on the go and, guess what, that already exists so AMD can just give them any APU design and ask MS if they want to tweak any IP in it.

You are missing the point here.
Anything AMD produces that is not a server CPU right now is losing them money, binding up a big portion of wafers for low cost console APUs is costing them money.

It used to be an indispensable somewhat steady income source for amd back when they had only jaguar and excavator that nobody wanted to buy.

For AMD, specifically, it may be very low cost to get into this venture with MS compared to any company which have not been building itself as a "design house". From memory, I don't recall many others doing that. The biggest one being nVidia, but we all know how dealing with them is like. Qualcomm? LOL, no. Apple? LOL x2, no. Intel? They could, but much like @bit_user said: they want higher margin parts and that would require a bigger learning curve for them as they'd want to use their own fabs?

For AMD the amount of R&D that MS and sony pays for is why they keep up working with them, at least that's my opinion.
All the other companies can pay their own R&D and put their resources in higher margin products.
Why learn to make something cheap to sell it at very low margins?!What's the point?

Which is another interesting bit*: the fabs and what MS+AMD can do there. I'm guessing TSMC in the USA as an initial thought, so that tells you what process they'll be looking to use and similar designs AMD already has out there in the wild for them. Otherwise, outside of the USA, it can be any process where, to me, N3X would make the more sense.

Yeah, MS and Sony might even force AMD to use intel foundry as a second supplier, or depending on how taxes work if they get tax cuts for using an US company they might only use intel.

Although new nodes are way too expensive for an console so they are not going to use any of the new US FABs from either company.

 

[bit_user]

Anything AMD produces that is not a server CPU right now is losing them money,

That's not accurate.

binding up a big portion of wafers for low cost console APUs is costing them money.

It probably would not be AMD that's buying the wafers, but rather Microsoft. AMD does these designs as a contractor, but MS owns the resulting IP and probably negotiates all of the manufacturing contracts. MS has its own cloud CPUs, so they're already doing wafer buys and managing the whole chip product process.

https://newsroom.arm.com/blog/arm-powered-microsoft-azure-cobalt-100-vms
​

Although new nodes are way too expensive for an console so they are not going to use any of the new US FABs from either company.

I expect the new consoles to use a N3-family node, so that they get an "optical shrink" for free. This is similar to how Sony used N7 for the PS5 and then later did a respin on N6. If N3 is too expensive, then I guess it'd have to be one of the N4 nodes.

 

[TerryLaze]

That's not accurate.

Ok, it's making them less money than they could be making.
That difference between the larger and smaller amount of money they could make is not a loss....it's just less money they make but that's not a loss that's just them making less money than they could........
tl;dr
they lose money

It probably would not be AMD that's buying the wafers, but rather Microsoft. AMD does these designs as a contractor, but MS owns the resulting IP and probably negotiates all of the manufacturing contracts. MS has its own cloud CPUs, so they're already doing wafer buys and managing the whole chip product process.

​

https://newsroom.arm.com/blog/arm-powered-microsoft-azure-cobalt-100-vms​

That would need intels approval due to the cross licensing.

 

[-Fran-]

Point was that the only ARM chips that perform well and are worth a damn at the moment for these applications are from Apple so building an ARM console on chips that are subpar at best is pointless.

I know where you're coming from, but again it is a non-point as much as saying "if pigs fly, I'd be flying next to them". If you can convince me (or anyone here) that MS (or Sony, et al) can convince Apple to make Hardware for them while keeping their IP, I'll take your point as a valid one. From another perspective: there's a lot of graphical IP that Apple does not have which they'd need to either license or build from the ground up. Not even considering MS would have to use Metal-compliant software, as I'm 300% sure Apple wouldn't want to support DirectX or any special version of it.

Little bit different, the fact you can just access steam from an Xbox will be a game changer rather than being trapped in Sonys store. It also positions Xbox to be the steam machine valve wanted but could never actually accomplish or maybe they just straight up allow you to install the OS on a PC down the line. PC gaming made couch friendly and all that.

Nintendo has always been a bit different because they haven’t relied on 3rd party titles for a while. All they need to do for a console to be successful is Mario kart, Smash, 3D Mario, one or two Zelda’s and a handful of second string franchises like Kirby and some remasters.

Agreed.

You are missing the point here.
Anything AMD produces that is not a server CPU right now is losing them money, binding up a big portion of wafers for low cost console APUs is costing them money.

It used to be an indispensable somewhat steady income source for amd back when they had only jaguar and excavator that nobody wanted to buy.

For AMD the amount of R&D that MS and sony pays for is why they keep up working with them, at least that's my opinion.
All the other companies can pay their own R&D and put their resources in higher margin products.
Why learn to make something cheap to sell it at very low margins?!What's the point?

I somehow don't think that is the case and @bit_user answered on my behalf. There's a lot of details we're not privvy to in the MS (or Sony) relationship with AMD. Like I said, it could very well be very close to "cost zero" for AMD to do this for all we know and "that pathetic amount we're assuming is low margin" is actually all margin. Who knows 🤷‍♂️

Yeah, MS and Sony might even force AMD to use intel foundry as a second supplier, or depending on how taxes work if they get tax cuts for using an US company they might only use intel.

Although new nodes are way too expensive for an console so they are not going to use any of the new US FABs from either company.

Yes, I can see that happening.

Regards.

 

[LordVile]

I know where you're coming from, but again it is a non-point as much as saying "if pigs fly, I'd be flying next to them". If you can convince me (or anyone here) that MS (or Sony, et al) can convince Apple to make Hardware for them while keeping their IP, I'll take your point as a valid one. From another perspective: there's a lot of graphical IP that Apple does not have which they'd need to either license or build from the ground up. Not even considering MS would have to use Metal-compliant software, as I'm 300% sure Apple wouldn't want to support DirectX or any special version of it.

No. My point was they won't be using arm because the only chips worth a damn are AS therefore they won't be using ARM. Please read before you respond.

 

[-Fran-]

No. My point was they won't be using arm because the only chips worth a damn are AS therefore they won't be using ARM. Please read before you respond.

So Qualcomm shouldn't make Snapdragon because M1 exists? So Android should migrate to X86 or RISC-V? Your statement implies no one should try to make any ARM-based designs because the M1 was created, which is just weird to say?

Sorry, but I know what you mean. It's just you're coming from the worst possible angle here and makes zero sense in the context.

Microsoft, if they want to make an ARM design and not X86, or even RISC-V for their next handheld/console/whatever, they can with AMD, because AMD is not tied to any particular CPU ISA (EDIT: I forgot they have X86-64 extensions, LOL) and have a very strong graphical portfolio to go with. Like I said, the closest one in that regard is nVidia. Apple, I'm 200% sure, is not even part of the conversation for all the things I've said already.

I'm not even being pedantic here, I hope? It's just so weird you'd try to argue that Microsoft can't make an ARM-based design because Apple's is so much better. Specially in isolation, since, you know, Microsoft actually writes their own OS, like Apple does. So they have equal expertise in there, I'd say?

If you still think I'm wrong on where I'm coming from, please do explain as I'm not trying to be obtuse on purpose.

Regards.

 

[LordVile]

So Qualcomm shouldn't make Snapdragon because M1 exists? So Android should migrate to X86 or RISC-V? Your statement implies no one should try to make any ARM-based designs because the M1 was created, which is just weird to say?

Sorry, but I know what you mean. It's just you're coming from the worst possible angle here and makes zero sense in the context.

Microsoft, if they want to make an ARM design and not X86, or even RISC-V for their next handheld/console/whatever, they can with AMD, because AMD is not tied to any particular CPU ISA (EDIT: I forgot they have X86-64 extensions, LOL) and have a very strong graphical portfolio to go with. Like I said, the closest one in that regard is nVidia. Apple, I'm 200% sure, is not even part of the conversation for all the things I've said already.

I'm not even being pedantic here, I hope? It's just so weird you'd try to argue that Microsoft can't make an ARM-based design because Apple's is so much better. Specially in isolation, since, you know, Microsoft actually writes their own OS, like Apple does. So they have equal expertise in there, I'd say?

If you still think I'm wrong on where I'm coming from, please do explain as I'm not trying to be obtuse on purpose.

Regards.

I’m saying they wouldn’t use ARM because chips like snapdragons aren’t good enough for the purpose. They’re fine for smartphones and at a push low power laptops. They’re not at the level of AS or even close to it at this point in time. AMD are further behind. It would take years of development and pushing lower end chips into portable devices before they could launch a higher end SKU. It took Apple a decade of development, billions of devices and vertical integration at a scale no one else on the industry has to get where they are. Qualcomm have had around the same amount of time but without the integration have to effectively lie to make a comparison. Add in to the fact you’re asking game devs to write for ARM now too because as great as proton and wine are they’re not perfect enough for a console demographic.


Microsoft don’t make their own chips and they don’t make technology specifically for their own chips. They have to outsource it to a 3rd party and use existing technologies and even then Microsoft for ARM runs better through a VM and a translation layer on Apple silicon than it runs on Microsoft’s own dev kit.

 

[TerryLaze]

Like I said, it could very well be very close to "cost zero" for AMD to do this for all we know and "that pathetic amount we're assuming is low margin" is actually all margin. Who knows 🤷‍♂️

You only have to look at their official numbers to know that it's a very small amount of money, all margin or next to no margin.
It's in their gaming division together with GPUs , they made a total of 647mil in revenue in q1 2025 so including the GPUs...
And that turned into 496mil operating income BUT including client so desktop CPUs...
A total of 2,941 turned into 496, what is that something like 15% for consoles GPUs and CPUs all together ?

As I said, anything other than server is barely making any sense for them right now.

 

[-Fran-]

I’m saying they wouldn’t use ARM because chips like snapdragons aren’t good enough for the purpose. They’re fine for smartphones and at a push low power laptops. They’re not at the level of AS or even close to it at this point in time. AMD are further behind. It would take years of development and pushing lower end chips into portable devices before they could launch a higher end SKU. It took Apple a decade of development, billions of devices and vertical integration at a scale no one else on the industry has to get where they are. Qualcomm have had around the same amount of time but without the integration have to effectively lie to make a comparison. Add in to the fact you’re asking game devs to write for ARM now too because as great as proton and wine are they’re not perfect enough for a console demographic.


Microsoft don’t make their own chips and they don’t make technology specifically for their own chips. They have to outsource it to a 3rd party and use existing technologies and even then Microsoft for ARM runs better through a VM and a translation layer on Apple silicon than it runs on Microsoft’s own dev kit.

Ok, that's fair to point out and I agree.

So the question I have for you is: do they really need an M1 competitor for a portable device or even a console? (performance and power level characteristics)

The complexity of the M1 (EDIT: M-line, I should say) and size, makes it rather expensive to produce and, like you said, a lot of R&D behind it, but when you look at it from a "building blocks" perspective, it's not a real revolution on what it is; or at least, to me it is not. Hindsight 20/20 as always, but it's not like you can't get close to its performance with the right building blocks. Or, said in another way, the M1 was designed with a specific type of workload in mind, so they just got rid of any baggage that they don't want. From what I understand, not everything is rosey in the Apple world from the software side due to this, as you also mention, there's still a lot of tooling which hasn't been properly migrated over, but should finally be ported correctly with MacOS 27 when they completely drop X86 support.

Taking into consideration that, #1, MS doesn't need a performance king in neither market and just hit a performance level that "feels" next gen to them (same rationale with the current console gen) and #2, they need to hit a budget within the division (tying this to the "XBox everywhere", means the HW has now officially become secondary, if not tertiary, to their strategy), so overly complex SoC, it won't be, no matter how they source it.

You only have to look at their official numbers to know that it's a very small amount of money, all margin or next to no margin.
It's in their gaming division together with GPUs , they made a total of 647mil in revenue in q1 2025 so including the GPUs...
And that turned into 496mil operating income BUT including client so desktop CPUs...
A total of 2,941 turned into 496, what is that something like 15% for consoles GPUs and CPUs all together ?

As I said, anything other than server is barely making any sense for them right now.

Bold part mine. Yes, you're not wrong in the accounting spreadsheet we get each quarter it's not a "big number", but if it wasn't relevant to them, or see it as worthwhile, they've be spinning off whatever division/group is in charge. Who knows, maybe you're right and AMD will announce they'll get rid of that division and sell it to Microsoft or Sony or whomever wants it since it doesn't produce "big numbers" for them. The fact they've been running that for years (AMD, through ATI, has been powering consolers for a lot of years now) means they see value/benefit on having it around and keep on doing it, independent on how our external analysis turns out to be.

Again, you could be spot on and it's 0% margin with small numbers, but they're still doing it. Being "zero margin", I'd have to assume, it's not. The truth is always somewhere in the middle of the extremes, so I don't think it's worth arguing more on this specific bit as the amount of available information on margins is, I'd say, non-existant for us. The specifics on the agreements with Sony and MS are hidden under secrecy, I'm sure.

Regards.

 

[thestryker]

It probably would not be AMD that's buying the wafers, but rather Microsoft. AMD does these designs as a contractor, but MS owns the resulting IP and probably negotiates all of the manufacturing contracts.

This doesn't seem particularly likely given that AMD has sold both Xbox and Playstation silicon on the open market in CPU/APU form. I can't imagine part of their design deals with the companies would be buying back silicon that doesn't make the grade.

 

[bit_user]

Ok, it's making them less money than they could be making.
That difference between the larger and smaller amount of money they could make is not a loss....it's just less money they make but that's not a loss that's just them making less money than they could........
tl;dr
they lose money

No, making less money than you could've is not the same as losing money!

What you apparently mean to say is that it's not profit-maximizing. That's very different than losing money!

That would need intels approval due to the cross licensing.

I think you're also wrong, there.

 

[LordVile]

Ok, that's fair to point out and I agree.

So the question I have for you is: do they really need an M1 competitor for a portable device or even a console? (performance and power level characteristics)

The complexity of the M1 (EDIT: M-line, I should say) and size, makes it rather expensive to produce and, like you said, a lot of R&D behind it, but when you look at it from a "building blocks" perspective, it's not a real revolution on what it is; or at least, to me it is not. Hindsight 20/20 as always, but it's not like you can't get close to its performance with the right building blocks. Or, said in another way, the M1 was designed with a specific type of workload in mind, so they just got rid of any baggage that they don't want. From what I understand, not everything is rosey in the Apple world from the software side due to this, as you also mention, there's still a lot of tooling which hasn't been properly migrated over, but should finally be ported correctly with MacOS 27 when they completely drop X86 support.

Taking into consideration that, #1, MS doesn't need a performance king in neither market and just hit a performance level that "feels" next gen to them (same rationale with the current console gen) and #2, they need to hit a budget within the division (tying this to the "XBox everywhere", means the HW has now officially become secondary, if not tertiary, to their strategy), so overly complex SoC, it won't be, no matter how they source it.

If they were to use arm as the post I replied to suggested they would currently have to use Apple silicon to get adequate performance.

I have a portable AMD handheld and the performance isn’t exactly great. You can play some games decently, currently working my way through Doom Eternal, however the battery life of poor, the chip doesn’t perform that well being more comparable to my base M1 Pro laptop than anything more recent and you’re stuck using fairly low bandwidth RAM. It’s convenient but not a great performer. You’d also have to split the user base because ARM doesn’t make a ton of sense in a home console. It’s easier to just go x86 in both and deal with the compromises of that.

 

[thestryker]

If they were to use arm as the post I replied to suggested they would currently have to use Apple silicon to get adequate performance.

This is just flat out false on every level. The prior consoles used absolutely awful 8 core parts which were outclassed by Intel dual core parts of the time. The current consoles are basically slightly modified underclocked 3700Xs. Console CPUs have never come close to meeting maximum CPU performance of their time. Consoles need something that provides good enough performance without using a ton of die space and Apple isn't that.

The most likely reason Arm isn't being used is software compatibility.

 

[bit_user]

From another perspective: there's a lot of graphical IP that Apple does not have which they'd need to either license or build from the ground up. Not even considering MS would have to use Metal-compliant software, as I'm 300% sure Apple wouldn't want to support DirectX or any special version of it.

Oh, but you can already run DirectX on Metal. I have no idea how well it works, but there's a Vulkan implementation called MoltenVK that runs atop Metal. Then, you can run DXVK atop that. MoltenVK was even patched specifically to enable it!

MoltenVK modified with DXVK patches for macOS: libMoltenVK.dylib

**Disclaimer: ADVANCED USERS ONLY** - manually updating CrossOver comes at your own risk and may make it harder to get support from CodeWeavers** please see this thread on the CodeWeavers Forum. This is MoltenVK 1.1.9 modified by user NullPointerEx22 taking the most recent version of MoltenVK and...

community.pcgamingwiki.com

 

[-Fran-]

Oh, but you can already run DirectX on Metal. I have no idea how well it works, but there's a Vulkan implementation called MoltenVK that runs atop Metal. Then, you can run DXVK atop that. MoltenVK was even patched specifically to enable it!

​

MoltenVK modified with DXVK patches for macOS: libMoltenVK.dylib

**Disclaimer: ADVANCED USERS ONLY** - manually updating CrossOver comes at your own risk and may make it harder to get support from CodeWeavers** please see this thread on the CodeWeavers Forum. This is MoltenVK 1.1.9 modified by user NullPointerEx22 taking the most recent version of MoltenVK and...

community.pcgamingwiki.com

​

I mean, it comes with the same caveats of Proton: community driven initiative (ignoring Valve, for a second 😀). If we're talking about Apple doing it, do you really think they'd write a translation layer for Metal, or Microsoft for that matter? It would have to either MS supporting Metal or Apple supporting DirectX for a potential collaboration on a console/device. Plus the whole point of a console is to run as close to metal as possible, so I doubt they'd try to do either way.

Wasn't there a group trying to get Linux running in the M1-based laptops a while ago with very little success as well?

Also, I can't remember how close/different Metal was to Vulkan, but I vaguely remember it was some sort of fork, no?

Regards.

 

[bit_user]

No. My point was they won't be using arm because the only chips worth a damn are AS therefore they won't be using ARM.

I don't agree with that. Arm's X925 cores are pretty good and they're the P-cores in the Nvidia/Mediatek N1X.

https://www.tomshardware.com/pc-com...p-could-rival-intel-and-amds-laptop-offerings
​

Qualcomm's Oryon cores were decent, for a first generation product. They're also last year's news, though, and what we're talking about is next year. So, I wouldn't consider Oryon v3 out of the running.

BTW, it looks like ARM's next gen cores are already starting to surface?

https://www.gsmarena.com/leakster_a...ve_ipc_to_run_more_efficiently-news-68283.php
​

 

[bit_user]

I’m saying they wouldn’t use ARM because chips like snapdragons aren’t good enough for the purpose.

But they don't generally use the fastest cores possible. They used Puma cores, and those sucked - even back then! And they used ~3 GHz Zen 2 cores, back when Zen 3 launched at like 4.9 GHz. What matters for MS and Sony is to have enough compute power, at the right price and power/thermal profile.

Probably the only reason related to compute performance why ARM might've been ruled out is, once you factor in the emulation overhead, maybe there were some games that didn't run fast enough and they didn't want to limit compatibility with older software titles just to those developers had ported & recompiled.

Add in to the fact you’re asking game devs to write for ARM now too

That's nothing new, for them. Nintendo Switch is ARM, as is phones. So, if they've ever written game for Android, IOS, or Switch, they've developed on ARM. I'm sure all the big game engines support ARM.

I have a portable AMD handheld and the performance isn’t exactly great. You can play some games decently, currently working my way through Doom Eternal, however the battery life of poor, the chip doesn’t perform that well being more comparable to my base M1 Pro laptop than anything more recent

How do you even know what's the bottleneck, though? Those iGPUs are pretty tiny and IMO are much more likely to be your bottleneck.

You’d also have to split the user base because ARM doesn’t make a ton of sense in a home console.

Tell that to Nintendo.

 

[bit_user]

You only have to look at their official numbers to know that it's a very small amount of money, all margin or next to no margin.
It's in their gaming division together with GPUs , they made a total of 647mil in revenue in q1 2025 so including the GPUs...

No, the consoles are actually in a different bucket. I forget what it's called.

The fact they've been running that for years (AMD, through ATI, has been powering consolers for a lot of years now) means they see value/benefit on having it around and keep on doing it, independent on how our external analysis turns out to be.

AMD gets some intangibles out of it. One good example is Project Amethyst:

https://videocardz.com/newz/sony-un...i-enhanced-hardware-in-collaboration-with-amd
​

That's reportedly where RDNA4's huge leap in AI horsepower came from, as well as the upscaling technology in FSR4.

 

[bit_user]

This doesn't seem particularly likely given that AMD has sold both Xbox and Playstation silicon on the open market in CPU/APU form. I can't imagine part of their design deals with the companies would be buying back silicon that doesn't make the grade.

Why not? If it's useless as a console chip, due to a defective iGPU, then why wouldn't they buy it off Sony/MS and sell it as a CPU?

 

[bit_user]

I mean, it comes with the same caveats of Proton: community driven initiative (ignoring Valve, for a second 😀). If we're talking about Apple doing it, do you really think they'd write a translation layer for Metal,

Apple has officially endorsed MoltenVK. I'm not sure if they contribute to it, but they've said they'll support apps using Metal via MoltenVK.

It would have to either MS supporting Metal or Apple supporting DirectX for a potential collaboration on a console/device.

Neither. Valve is the primary maintainer of DXVK.

Just to be clear, I'm not arguing some sort of hypothetical where Apple silicon is used in a console. I'm just trying to get a few facts straight.

Plus the whole point of a console is to run as close to metal as possible, so I doubt they'd try to do either way.

Hmmm... in Microsoft's case, I'm not sure that's true. They might indeed try to keep games from using anything but Direct3D.

Also, I'm not sure it's "the whole point of a console". Consoles typically don't have much abstraction over the hardware, since a game is usually written far a specific generation, but now that Sony has gone to the point of native backward compatibility and has two generations with "Pro" consoles, they might finally be starting to see the light and I could imagine them forcing PS6 games to use supported APIs & shading languages and not to include GPU ISA dependencies. Especially now that Sony might be feeling like they don't want to be wed to RDNA2 for the indefinite future.

Wasn't there a group trying to get Linux running in the M1-based laptops a while ago with very little success as well?

The Asahi project got most of the SoC working, including the GPU. The number of Vulkan or OpenGL features supported on the GPU is fairly low, last I heard. I believe the project has now transitioned work over to Redhat.

Also, I can't remember how close/different Metal was to Vulkan, but I vaguely remember it was some sort of fork, no?

Metal launched first, though that doesn't necessarily mean much. I think Metal is a little higher-level and easier to use than Vulkan. I don't know too much else about it.

 

[thestryker]

Why not? If it's useless as a console chip, due to a defective iGPU, then why wouldn't they buy it off Sony/MS and sell it as a CPU?

Because it makes literally no financial sense?

Low volume part that isn't part of the SKU stack. Has to be soldered to a board because it isn't socketable. In the case of PS5/Series requires GDDR memory. There are so many negatives I cannot imagine why they'd touch it unless it was already theirs and the silicon was paid for.