They do this every 5 years or so. Remember Surface RT ? Remember Surface pro x ? it's like clock work and just like clock work it never happens
MS leak illustrates new console development cycle
- Thread starter Shifty Geezer
- Start date
Microsoft has been known to make several billion dollar mistakes, that is right. But they have enough cash to do so anyway so in the end they have even more money because companies are forced to buy office and windows and whatnot.
Right but the whole MS is going hard into ARM is something that happens every 5 years now. That is my point.
While Surface RT may have been a mistake the Surface pro that launched a few months after wasn't. It went on to spawn a multi billion dollar market segment for MS.
cheapchips
Veteran
It's their third ARM attempt though, and aren't their version 3s the ones where Microsoft nails it? Like with Windows 3 and the Xbox O...
XB360 nailed it. XBO, not so much.
Maybe but maybe not.
It seems like this arm push is literally because of AI and qualcom having their "npu" ready first. I am sure if AMD or Intel had their npu ready they would have pushed that to the forefront.
Microsoft are doing it only because they believed that Apple ARM designs could be matched or surpassed.
I disagree with you. MS has done this multiple times in the past.
It is because they do not develop any hardware and thus are easily swayed by empty promises and fake projected benchmarks or presentations
I don't think MS are as dumb or gullible as you're making out. They do make some bizarre seeming and poor decisions, but a lack of technical ability or understanding has never been their issue.
A big plus for arm is power efficiency. Turning that into an attractive Windows product is not easy as it (until recently) meant having awful x86 compatibility or performance. MS are right to test the water from time to time with arm products, and they will go in hard if they can make devices that people want - which is to say performant devices with transparent x86 compatibility.
If MS want to share the same base architecture between a handheld and a next gen home console then arm could make sense. Back compat needs only to be good enough to sustain current gen performance on the CPU side, and die and power savings could be put towards a powerful NPU or better GPU performance. For a handheld power consumption is critical, and big batteries and short battery life are two of the things that drive up cost and limit portability of devices like the Steam deck.
MS seems to be putting more effort into Windows on Arm than ever.
I have to say at this point in time the x86 chip power consumption in a handheld is like a fart in the wind. The majority of the power consumption is the gpu. When looking at making a mobile xbox series s an 8core zen 2 at 3.6ghz on a modern process will sip power. Throw in a new generation of zen with higher ipc and you wouldn't even need to hit the same clock speeds for the same performance. The main issue I believe for portable consoles is the gpu and bandwidth to the gpu. I don't think putting an arm cpu inside of the next xbox will change that and I'd assume a next gen xbox would have much more bandwidth than the series s. But here we are 4 years later and I don't thik there is a mobile solution to provide that amount of bandwidth at a reasonable cost.
That's not the case on AMD Z1 platforms where more demanding CPU loads appear to show GPU clocks reducing, and it's not the case on PS5 where additional GPU boost in possible when the CPU is diverting part of its unused power budget there.
Games can be demanding on CPU and in a sustained way, and if next gen we're still going to be updating increasingly complex BVH structures on the CPU we're likely to want CPU performance to increase while still being within a mobile power budget.
MS have specifically focused on power and battery life as a strong point of their latest ARM based surface laptop. This doesn't mean MS will switch for a handheld console, but it does mean that there are still real world power considerations for mobile CPUs.
MS are seeing an opportunity for market growth is all. Right now these QC ARM Elite processors are gaining a huge amount of hype in the press, mainly on the unfounded idea that ARM processors are all just going to magically be as efficient as Apple's silicon. But MS doesn't care if this is true or not, they just know that it's a major new vendor coming in that will bring in more customers or at least sway some people from switching to Apple, and are all too happy to promote it as a big deal.
It does not mean they are 'betting against' x86. And it does not mean their Xbox division, which is not the same as their laptop division, will somehow be under the same spell of the magic of ARM. They will know better, and know that it's not some magic shortcut to incredible performance and power efficiency like the average Techcrunch reader or whatever might think it is.
And honestly, lord help us if MS compromise their entire next gen platform just to make a handheld system a bit more power efficient. I think that'd be a big mistake. Heck, I think an Xbox handheld could be a mistake in its own right. They aren't Nintendo. Unless Sony also went ARM, they'd just be creating even more division for themselves, and not in a good way.
Last edited:
cheapchips
Veteran
Not just Qualcomm, but Nvidia/Mediatek too. MS will have gone from two PC hardware options for Windows OEMs to four. More if Samsung and whoever else decide to join.
The early reviews point to x86 titles finally running well on Arm. Windows gets to leverage what's going on in the much bigger phone / tablet SoC market.
Whether Xbox does is obviously a question. We could 'easily' end up with an ARM portable and x86 home console.
Last edited:
The z1 has boosts built into it up to 5.1ghz so of course it will start to use more power over a fixed cpu. A home console doesn't have to concern itself over the lower power usage of the cpu portion of the apu. Its the gpu portion that drives power draw
In a mobile series s they would be targeting the fixed performance of a 3.5ghz 8 core zen 2 not the full performance of a z1. You'd also be looking at zen5/6 vs zen4 in the z1.
Some proper reviews coming in show that these SD Elite processors were quite overhyped as many of us expected them to be. Not bad, just nothing particularly special, especially if you want to run anything more demanding than Youtube and some browsing or whatever.
And this is supposed to be the 'Apple competitor' custom Nuvia processors that would save the day for Android/PC. Standard ARM designs are not likely to fare any better when they do come along.
As for Xbox doing an ARM portable and x86 console, I'm not sure that really makes sense in light of this. x86 is not 'running well' on ARM, it's just...running. It can do it. But native x86 hardware will do it better, and while efficiency is important for consoles/handhelds, the performance part of the equation needs to be good, too. A handheld with already compromised mobile hardware doing lackluster x86 translation will not make for an inspiring experience.
Lunar Lake could well be more of an 'Apple moment' for laptops than these QC Nuvia ARM processors, at least from a more informed perspective. Whether media presents it like this is another thing...
cheapchips
Veteran
What will be interesting, with respect to consoles, is when we get some proper benchmark numbers with power consumption. The reviews I've seen so far have only touched on 'great battery life'.
In the PS5, and now the PS5 Pro, the power consumption of the CPU literally affects the real world performance of the GPU in real time. The Xbox doesn't and this is simpler, but not boosting and not using AMD Smarshift as Sony does mean that most of the time they were leaving performance on the table which particularly in the early years of the generation hurt MS in terms of relative performance.
This doesn't mean it's worth forgoing x86 in a next gen console, it just means that even home consoles with relatively huge power budgets have concerned themselves with this (this successful ones, from Sony).
Yeah, you'd only need at a basic level enough CPU to make sure you could run Series S games on the mobile device. Native handheld games could run in native Arm mode and get the best from perf/Watt.
All we know for sure it that MS definitely were at some point at least considering a move to Arm. They'd probably need an Arm core more exotic that anything currently around to provide enough AVX2 performance.
x86 emulation only needs to be good enough for BC, and an emulator on a Arm handheld would be build very specifically to match one specific hardware platform (Series S) running only software built and compiled specifically for that.
Platform specific games would be built natively to make best use of whatever Arm processor is in the handheld. Anything else would be odd. Arm is everywhere now and Arm compilers seem pretty good.
Performance is definitely important, and on console that's typically the performance of the GPU. On any handheld device saving power on the CPU means more power for the GPU. Saving die on the CPU means more die for GPU or NPU.
Again, MS were at least considering Arm for the next console and MS's console have proven at hotchips and with the Series consoles that they understand console design extremely well. They wouldn't have been thinking "magic" would solve their problems, and I'm pretty sure that they don't base their R&D plans on reading "Techcrunch" a few years in the future!
Benchmarks of native Arm applications, or of x86 emulated apps? One will be faster and more power efficient that the other.
No, it doesn't just need to be good enough for BC, because PS6 likely will do a lot more than this when it comes to backwards compatibility, set by expectations this gen. It will be a huge downside for Xbox when they're running BC games at last gen visuals/performance all while PS6 is doing so at improved visuals/performance. Playstation might get away with such a thing given market dominance, but Xbox cant.
And why would developers make ARM native versions of their games just for a niche handheld model of a niche console platform in Xbox? That's ridiculous. Xbox is already suffering noticeably from developers not optimizing their games well for their system given lack of market penetration. ARM would just make all this much, much worse.
Of course Microsoft might have been considering ARM, but I'm sure Sony have been as well. Evaluating the market and possibilities is expected. It does NOT mean anything at all about what they actually go with, though.
Yes that was literally my exact point. That they wont be fooled into thinking ARM was anything magical/special compared to x86. They will know better. Or at least should know better. It wouldn't be completely out of the question for some out of touch idiots at the head of Xbox to make terrible technical decisions, but if they listen to their hardware guys, they'll know ARM isn't magic.
A handheld can get away with BC performance that's sufficient for play as good as last gen home consoles, especially if it releases a bit before next gen.
So far MS has been the one to drive improved versions of last gen games with no developer input, using lots of clever interceptions of API calls from precompiled binaries and somehow also separating logic and rendering post release. It's nice, but it's not been a game changer when their two most successful competitors haven't given a shit.
I wouldn't bet on next gen home consoles being able to reliably run a single thread twice as fast as this gen either, regardless of x86 or Arm. Sony aren't exactly pushing clocks on the PS5 Pro, and they are again balancing CPU clocks and power draw against GPU clocks.
Edit: and to do what MS have done with Xbox One to Series consoles with 30 to 60 fps wizardry in (very selected) games, you'd need to bank on having your most framerate limiting thread reasonably reliably able to double up on performance.
I can make Arm versions of Unreal stuff with very little effort. I can't claim this is due to any skill on my part. It's another CPU, it's a different compiler, it's not the end of the world.
Fine tuning a game for a particular CPU architecture still goes on, but most code optimisation is not at the micro-architectural level.
It means that at the time they were considering it, the chance was greater than zero. And this was the best people in the business taking it seriously enough that they had a point where they'd have to decide.
And of course considering something doesn't mean you will actually use it.
I'm glad we've all established that no-one should think Arm is magic.
