SplunkyMunkey
Member
Instead of zen6c cores sony should of went with full fat zen6 cores and 2 lp cores instead of zen c cores. I'm sure sony likes to make things weaker. That bus speed is gonna hit the console hard
[MLiD] AMD Magnus APU Full Leak: RDNA 5, Zen 6, 110 TOPS NPU = XBOX Next-Gen Console!
- Thread starter Lunatic_Gamer
- Start date
- News Rumor Hardware
winjer
Member
That would eat into the power budget and the die space of the SoC, leaving less for the GPU and NPU.
HerjansEagleFeeder
Member
I wish we could go back to modest consoles, affordable prices and platform holders that don't tolerate missing QA, bad performance or other technical issues on their store front. If we hadn't god awful framerates, stutters and so on who would care if a console is not on par with a high end PC?
Syphon Filter
Member
they are never going to be on par with a high end pc.
Baemono
Member
When was that
Negotiator101
Member
Apparently all plans are getting scrapped if you listen to the insiders.
HeisenbergFX4
Member
Gaiff
SBI’s Resident Gaslighter
I don't see how they were wrong. It's going out faster than the PS5.
HerjansEagleFeeder
Member
Everything before PS3/360, but I'm sure a real smart person like you will find tons of pseudo reasons why that doesn't count right?
SplunkyMunkey
Member
Make the apu bigger like Microsoft has done with a bigger gpu at 58cu and a 192bit bus and with the technology that cerny was applying would be similar to the likes of series x and ps5. I'm not saying they would but if ms made magnus atleast 150 dollars more compared to ps6 then sony knew that they know they messed up
HJuggernaut
Member
If MS is going with a chiplet design for nexgen I dont see why they are sticking with AMD, Nvidia+Intel would have been much better partnership in terms of everything really CPU, GPU, AI.
Zathalus
Member
Intel would be worse in terms of the CPU, and the cost for going with two different vendors will be rather high.
Gaiff
SBI’s Resident Gaslighter
MGS3 was running at 15-25fps, so was OOT. The standards have increased massively when it comes to performance expectations.
HJuggernaut
Member
not really you use binned intel cpus which is what MS is going to do with their 11 core cpus, and NVIDIA smokes anything AMD does in terms of Raster, RT and AI. Unified mem pool fast interlink between cpu+gpu. Must better solution than sticking with AMD thats been shafting MS since day one.
Last edited:
Xbox 360 slim
Member
Fanboy BTFO
kevboard
Member
not since day 1. the Xbox 360's AMD (ATi) GPU shat on the PS3's Nvidia GPU for example, even tho it released a full year earlier.
and for the Xbox One generation, AMD was also still absolutely competitive... there Microsoft just made a stupid decision with their idiotic memory setup, that ballooned their APU size and price.
it really only was this generation that AMD was indeed an issue. because AMD was completely behind the times (around 5 years behind) in the only real advancement this generation gave us.
twilo99
Member
GTA being the boat in question
Zathalus
Member
And yet cost will be much much higher. I'm also not sure if Intel binned chips would be better than what AMD currently offers. Nvidia probably would be, assuming Rubin isn't as disappointing uplift like Blackwell is.
twilo99
Member
I'd watch it
Baemono
Member
I wasn't trying to be the smartbum, it was a geniuine question.
But now you say before PS3/360 I think the problem is that people didn't really care about framerate or image quality as they've been taught to care since the PS360 generation, which isn't for the best
winjer
Member
sncvsrtoip
Member
Lower gpu frequency for smaller soc ? Sony realy going cheap way nextgen.
HJuggernaut
Member
its already 1000 bucks
winjer
Member
Is it though?
kevboard
Member
Kepler: "no way any game would run worse on Magnus!"
Ace Combat developers: "hold my F-16C"
(context: Ace Combat 6 ran worse on Xbox One X than on PS4 Pro, even tho they ran at 100% matched settings
)
Last edited:
All this just sounds like a 5080/9070xt difference to me. Maybe 25% at most theoretically if spit balling, more interested in pricing. If it's a $400 difference or even $300 I have to imagine a already tough hardware sales history will just be even tougher next gen. Definitely a more niche audience than the whole Pro audience imo.
vaibhavpisal
Member
What is the use of NPU?
winjer
Member
You need to ask?
vaibhavpisal
Member
Serious question. I was under impression GPU was better at these tasks.
winjer
Member
Depends on how Sony and AMD wants to implement it.
But an NPU is a dedicated unit, just for AI. While on a GPU, it's spread out across the shaders.
thicc_girls_are_teh_best
Member
I was on B3D earlier today and thought about something: for other reasons, I've been doing a lot of research on older game systems, and recently one of those is the Bandai Pippin. If you're not aware, it was basically a Macintosh clone developed between Apple & Bandai released in 1996, focused as a multimedia computer for the living room with gaming support.
It ran on a version of System 7 OS, specifically 7.5.2, and Apple intended it to be seen as a format standard similar to VHS (or what would eventually become of DVD). In that way, their mentality wasn't too far removed from The 3DO Company, and the business model was similar. Apple and Bandai just leveraged technologies and a framework already present with Macintosh computers of the time though, whereas 3DO had to build all that up from scratch.
Anyway, the interesting thing about Pippin from a technology & business model POV, is that while it shared features with Macintosh systems, it didn't just outright run Macintosh software out-of-the-box. Developers had to alter their Macintosh software to fit a Pippin profile, because the way Pippin accessed programs to run in the file structure differed from Macintosh computers. That's in part because it had no installed OS; each disc came with an OS image to run the application off of instead.
This is because Apple viewed the Pippin as an appliance. Nonetheless, it had very similar hardware upgrading functionality to a computer, albeit (initially) proprietary RAM memory cards for example, until they made a switch later to use standard DIMM sticks. Apple also intended to license the Pippin "format" out to other companies beyond Bandai, but since the Pippin was such a massive failure, it didn't get that far. Plus Steve Jobs killed all Macintosh clone projects in 1997 after he rejoined Apple anyway.
Well, the reason I'm bringing this up is because I am definitely thinking this is the approach Microsoft are going to take with the next Xbox when it comes to Windows compatibility. Yes it's easy to say "3DO", but they didn't have a PC OS platform to co-manage the way Microsoft does, either. So I think the Apple/Bandai Pippin is actually a better point of reference. Some of the rumors have already said MS are phasing the Win32 dev environment over to Xbox, so on the backend side of things that's definitely happening. However, I'm strongly doubting native Windows apps are going to "just run" on the thing. MS are probably going to use some custom version of Windows for the next Xbox's OS that still has portions of Xbox OS for BC & legacy reasons.
I think they're probably going to focus on making the porting work for Windows app developers to the next Xbox minimal, maybe by just changing a few lines of code, some metadata, and pointing to some specific API libraries or whatever. They may have to test for a few things, and get QA verification, but I guess that can be relatively quick & seamless. Maybe MS even sets up a program to offload that work for app developers for a very small fee, and they only collect a small percentage cut on sales of commercial applications, not free ones. What versions get "ported" to the Xbox, still count as a different instance vs. Windows, though. It'd be equivalent to what Apple & Bandai set up for the Pippin, or Amiga for the CD32 (or Fujitsu for the FM Towns Marty).
Like, we've already seen this in action decades ago, but the tech was way less advanced and magnitudes more decentralized/heterogenous, so there were tons of compatibility issues that just couldn't be overcome without hardcore reprogramming huge chunks of the application for a different device, even if the two devices shared a similar CPU & GPU, or a similar OS, even if the software was programmed in a high-level language like C (high-level compared to assembly. C itself is pretty low-level compared to stuff like C++ or C#, let alone Python). I think it also makes the most economical sense for MS to take this approach, because it still allows for messaging and marketing of the Xbox devices to be somewhat easy. As for things like Steam support, I'm of the belief that it's NOT going to pertain to the storefront itself. Steam has a launcher where you can access installed games and run them, but the launcher itself isn't the storefront. I guess it's kind of similar to GOG Galaxy, in that way.
Microsoft are probably going to add the necessary Windows functions to enable a Steam launcher on the Xbox, and let people transfer Steam games they've purchased elsewhere to their Xbox (either entire library or just whatever game they want to play at that moment) via the launcher support, but there'll probably be some limitations and maybe some compatibility issues with certain games expecting either hardware and/or software/OS features not present on the next Xbox. But, that probably won't be an issue for most games, and the actual technical specifications of the next Xbox would be the potential limiting factor. Specific games, if they wish, can get a "port" to be more fully compatible with the next Xbox, same way as said for the general apps.
And, like the Pippin, Microsoft will probably also allow for some forms of hardware upgrades, like expanding RAM. Whether that's with a proprietary format or not, is up in the air. If the APU is chiplet-based, and depending on how the chiplet setup is implemented, maybe they can even allow for GPU upgrades, via custom-format GPU expansion cards. That way users don't have to buy an entire new system for a hypothetical mid-gen upgrade.
Personally, speculating on the aspects of the OS implementation, features, streamlining Windows and Xbox integration together on the backend side (APIs, SDKs, kernel-level etc.) and business model decisions thereof, is more interesting than just talking the tech specs at this point. That's somewhat similar with PS6 as well, where some things can probably be well-guessed (i.e the handheld complementing the console), but other things are still hazy to figure out just yet, like if SIE make any significant changes to their multiplatform software strategy, or if PS6 hardware is the dev target for 10th-gen games vs. treating PS6 as a PS5 Pro++ Resolution Box.
It ran on a version of System 7 OS, specifically 7.5.2, and Apple intended it to be seen as a format standard similar to VHS (or what would eventually become of DVD). In that way, their mentality wasn't too far removed from The 3DO Company, and the business model was similar. Apple and Bandai just leveraged technologies and a framework already present with Macintosh computers of the time though, whereas 3DO had to build all that up from scratch.
Anyway, the interesting thing about Pippin from a technology & business model POV, is that while it shared features with Macintosh systems, it didn't just outright run Macintosh software out-of-the-box. Developers had to alter their Macintosh software to fit a Pippin profile, because the way Pippin accessed programs to run in the file structure differed from Macintosh computers. That's in part because it had no installed OS; each disc came with an OS image to run the application off of instead.
This is because Apple viewed the Pippin as an appliance. Nonetheless, it had very similar hardware upgrading functionality to a computer, albeit (initially) proprietary RAM memory cards for example, until they made a switch later to use standard DIMM sticks. Apple also intended to license the Pippin "format" out to other companies beyond Bandai, but since the Pippin was such a massive failure, it didn't get that far. Plus Steve Jobs killed all Macintosh clone projects in 1997 after he rejoined Apple anyway.
Well, the reason I'm bringing this up is because I am definitely thinking this is the approach Microsoft are going to take with the next Xbox when it comes to Windows compatibility. Yes it's easy to say "3DO", but they didn't have a PC OS platform to co-manage the way Microsoft does, either. So I think the Apple/Bandai Pippin is actually a better point of reference. Some of the rumors have already said MS are phasing the Win32 dev environment over to Xbox, so on the backend side of things that's definitely happening. However, I'm strongly doubting native Windows apps are going to "just run" on the thing. MS are probably going to use some custom version of Windows for the next Xbox's OS that still has portions of Xbox OS for BC & legacy reasons.
I think they're probably going to focus on making the porting work for Windows app developers to the next Xbox minimal, maybe by just changing a few lines of code, some metadata, and pointing to some specific API libraries or whatever. They may have to test for a few things, and get QA verification, but I guess that can be relatively quick & seamless. Maybe MS even sets up a program to offload that work for app developers for a very small fee, and they only collect a small percentage cut on sales of commercial applications, not free ones. What versions get "ported" to the Xbox, still count as a different instance vs. Windows, though. It'd be equivalent to what Apple & Bandai set up for the Pippin, or Amiga for the CD32 (or Fujitsu for the FM Towns Marty).
Like, we've already seen this in action decades ago, but the tech was way less advanced and magnitudes more decentralized/heterogenous, so there were tons of compatibility issues that just couldn't be overcome without hardcore reprogramming huge chunks of the application for a different device, even if the two devices shared a similar CPU & GPU, or a similar OS, even if the software was programmed in a high-level language like C (high-level compared to assembly. C itself is pretty low-level compared to stuff like C++ or C#, let alone Python). I think it also makes the most economical sense for MS to take this approach, because it still allows for messaging and marketing of the Xbox devices to be somewhat easy. As for things like Steam support, I'm of the belief that it's NOT going to pertain to the storefront itself. Steam has a launcher where you can access installed games and run them, but the launcher itself isn't the storefront. I guess it's kind of similar to GOG Galaxy, in that way.
Microsoft are probably going to add the necessary Windows functions to enable a Steam launcher on the Xbox, and let people transfer Steam games they've purchased elsewhere to their Xbox (either entire library or just whatever game they want to play at that moment) via the launcher support, but there'll probably be some limitations and maybe some compatibility issues with certain games expecting either hardware and/or software/OS features not present on the next Xbox. But, that probably won't be an issue for most games, and the actual technical specifications of the next Xbox would be the potential limiting factor. Specific games, if they wish, can get a "port" to be more fully compatible with the next Xbox, same way as said for the general apps.
And, like the Pippin, Microsoft will probably also allow for some forms of hardware upgrades, like expanding RAM. Whether that's with a proprietary format or not, is up in the air. If the APU is chiplet-based, and depending on how the chiplet setup is implemented, maybe they can even allow for GPU upgrades, via custom-format GPU expansion cards. That way users don't have to buy an entire new system for a hypothetical mid-gen upgrade.
Personally, speculating on the aspects of the OS implementation, features, streamlining Windows and Xbox integration together on the backend side (APIs, SDKs, kernel-level etc.) and business model decisions thereof, is more interesting than just talking the tech specs at this point. That's somewhat similar with PS6 as well, where some things can probably be well-guessed (i.e the handheld complementing the console), but other things are still hazy to figure out just yet, like if SIE make any significant changes to their multiplatform software strategy, or if PS6 hardware is the dev target for 10th-gen games vs. treating PS6 as a PS5 Pro++ Resolution Box.
3liteDragon
Member
If you don't mind me asking, what clocks did they settle on for the PS6 CPU/GPU & Magnus?
Neo_game
Member
This gen both console PSU was rated more than 300W, PS5 in fact was 350W but from the numbers I have seen on max load it reached was 210W and SX 215W.
Insane Metal
Member
HJuggernaut
Member
68CU's 11 core cpu NPU 48 gb RAM and so on its 1000dollars at a loss for MS easy.
just the cooling here would be 200 bucks.
winjer
Member
Yes, that is one way to get a 1000$ console.
Though since it's an Xbox, it might as well be dead on arrival.
Punished Miku
Gaslighter Pro: 16.7 Fireflops
If they sell 5 million it beats Steam Deck. People should readjust their expectations for dead on arrival. Lots of niche hardware exists. This is going to be another one. No one is expecting this to sell tens of millions in any way. Have the Valve headsets sold even 1 million?
Don't think ps6 will have npu
Ah! but you forget about the second APU hidden in the powersupply that ps6 will ship with.
Btw I'd still say 'near 100%' because nothing was ever 100% winning no matter how big the advantage. Not PS4, not 1x, not even the original xbox.
Anything that isn't more than 50% higher across the board is not nearly enough to guarantee that outcome.
THE:MILKMAN
Member
That's what the radiance cores HW block basically is, isn't it?
viveks86
Member
Those are for Ray tracing. AMD's version of dedicated RT cores
Watching the cerny amd video radiance cores are hardware for rt, npu is extra ai block not sure how useful it is for graphics I'm sure it would be for Xbox copilot
HeisenbergFX4
Member
kevboard
Member
Going to be interesting to see where its priced
they could have a Series S2 that is "console priced", and a Series X2 that is premium priced
...that Series S2 could have around the same raster performance of the Series X, maybe even slightly below, but with vastly better RT perfomance offered by RDNA5 of course, and ML acceleration, so that it can easily play next gen games.
instead of having a standard and a budget model, that would be like having a budget and a pro model, with nothing in-between. cornering the PS6 on both ends. 1 console that's cheaper, and 1 console that has superior performance.
winjer
Member
An NPU or Tensor Units in the shader cores, can be used for upscaling, texture compression, ray-reconstruction, etc.
playbignbox
Member
Good specs, but it will likely be very expensive and not sell well in the end.
The only way MS could sell consoles today would be for it to be more affordable than the PlayStation (without being inferior) and to have excellent exclusive games (true exclusives, not released on PC or Play Anywhere).
Otherwise, no one should even consider getting one.
The only way MS could sell consoles today would be for it to be more affordable than the PlayStation (without being inferior) and to have excellent exclusive games (true exclusives, not released on PC or Play Anywhere).
Otherwise, no one should even consider getting one.
HeisenbergFX4
Member
It sounds like 1 true console but could be multiple SKUs of OEM PCs using this same Magnus chip