That’s Windows statistics, not Surface, though I see your point. Thing is, if ARM becomes fast enough, and is way more efficient, then it doesn’t matter how many people have older x86 hardware. They will buy the shiny new thing if they find it’s snappy enough. More specifically, if the emulation performance reaches a point that ARM doesn’t get in the way and it is transparent to the user, then most people (who don’t know an ARM from an arm and a leg or x86 from the nuclear launch codes) won’t care what’s under the hood. People just want to do their office work and play their app games and they could care less about the hocus pocus going on in the black box if it does the magic well enough for their purposes.
I agree. There is little incentive on the hardware partner front currently. I am speaking to Surface and Microsoft’s immediate goals there. If ARM Surface becomes fast enough that it’s transparent to the user with the advantage of longer battery life, then that will become the latest and greatest next thing that hardware partners will want. I realize too that the biggest thing getting pushed in Brazil where many friends live is two or three generations old Pentium-based systems at Lojas Americanas. So it is going to take a while before ARM takes off in mainstream PCs worldwide and that’s again a big if depending on if NUVIA delivers in US markets first.
I’m hopeful about the Nuvia stuff now that they have been swallowed by Qualcomm, but already the hype across the board is ridiculous, even with some here. In other words, I sense major disappointment coming…
And I still haven’t crossed out Samsung’s Exynos as we’ve seen some promising WOA protypes running that chip which tells that Samsung may have cut a deal to get out from under the exclusivity Qualcomm currently has on WOA
Well, considering how underwhelming Apple’s M2 has been, even if NUVIA’s first entry is only 10-20% off from Apple’s M series, I’d still consider it a huge win in my book. I did some digging this afternoon and found some promising results here below for Surface Pro 9. It’s not like Qualcomm has not been delivering huge gains gen-over-gen already. Meanwhile, Apple is sitting on their laurels with 0% IPC improvement from M1 to M2. Here is where we sit right now with the current players. Compared to Apple, Qualcomm even without NUVIA has already been making colossal gains with approximately four times the generational gains compared to M series (11% versus 40% for single-threaded improvement gen-over-gen; 21% versus 89% multithreaded improvement gen-over-gen). If Qualcomm continues this extremely steep trajectory of gen-over-gen performance rate of improvement, they will easily surpass Apple M series in one and half generations. As another note, on Microsoft SQ3 (using Geekbench 5 as a rough guide) emulated x86 performance should be roughly close to what native ARM performance was on SQ2 or how native apps felt like on the Surface Pro X. That’s huge, meaning running x86 programs should be transparent or mostly indistinguishable to most people. I can totally live with what the native performance was on the Surface Pro X for my x86 programs.
Single-threaded Geekbench 5
% Increase Gen-Over-Gen
Multi-threaded Geekbench 5
% Increase Gen-Over-Gen
SD 8cx Gen 3 / Microsoft SQ3
SD 8cx Gen 3 / Microsoft SQ3 (x86 to ARM emulation)
Too soon to argue the numbers yet, but a couple of points.
First and foremost, a huge amount of what Apple has done with the A series and now M series chips is Apple proprietary and there is a mountain of IP protecting it.
So the obvious thing here is that a lot of the things that Nuvia could do to improve performance over the 8cx, they likely wont be able to copy, thus making the overall task orders of magnitude harder.
Second I also think it’s highly unlikely that MS/Qualcomm will be able to/make the sacrifices needed to tune the OS to the chip and vice versa the way that Apple could. That is unless MS gets all new upper management that is much less risk averse.
Don’t get me wrong I’d love in this case to be totally wrong, but we are talking MS and Qualcomm here.
BTW: We’ve heard from one of our relatively well placed sources that Qualcomm’s internal targets are for a 30% improvement over the 8CX gen 3. So take that as the rumor it is however.
I will agree that the OS tuning allows Apple to wring out more performance from their chips than others, but even without it, if Qualcomm’s rate of improvement holds as it consistently has before NUVIA was in the mix, then it would not be unrealistic or a stretch to say that the Surface Pro 11 will outperform the iPad Pro in the next two generations even if Windows as an OS could draw out even more performance (see also: Linux versus Windows benchmark results). Heck, Intel and AMD are also both turning out high double digit percentage gains like Qualcomm has (e.g. Ryzen 7000 series is 30-40% generation over generation and Intel’s 13th Gen shows similar gains) for the last several generations as well and have not stepped on any of Apple’s IP legal landmines, so as they say, there is more than one way to skin a cat (sorry, @Dellaster ).
Apple isn’t the end all, be all to chip design and there is a prime reason other than design limitation why their gen-over-gen improvements have been slowing so much while other players have been flooring the pedal: complacency. Cook’s incrementalism is both a blessing and a cursing because it has a synonym: it’s what we call only doing “good enough.” Without Job-like drive and passion to shake things up, you are not going to get the thirst or ingenuity to improve at a radical rate of change necessary to maintain market leadership indefinitely. And especially when your competitors are improving at rate as much as four times as much as you are (and the numbers I shared earlier are real and not a guess at this point), it doesn’t take a rocket scientist to say that they are going to lap you eventually even if you are a mile ahead right now and even if you can’t see them in your rearview mirror right now. You might say it’s a variation of Einstein’s 8th Wonder of the World applied to tech-sense.
Hmmm, not so sure. Maybe if you are a hard core computer nerd. I used to fall into that category, but these days I view factors like longevity, productivity, and ROI as much more compelling. Computers are mainstream and mostly bought with other peoples’ money these days. Phones and tablets are probably headed in that direction too.
I wonder why they can’t reach for the stars, delay the product a little to try to aim for a 50%+ improvement. Then again they don’t have much competition in this space since Nvidia left & Samsung(phones) use Exynos in only certain regions.
Agreed, the handheld gaming space is has almost single-handed kept alive innovation in the mobile x86 space, with devices like the Steam Deck, Ayaneo Air and GPD Win Max 2. These OEMs are hungry for market share and you can see it how quickly they iterate components and snatch up discounted dies from AMD.
I used to see that hunger in the Surface team with ambitious designs like the Book, Neo, and X. I guess they’ve ‘graduated’ from all that.
Ok, real talk. How many of you guys are motivated to go WOA, now that it’s in the Surface Pro chassis?
Going head-to-head like this, MS better make ■■■■ sure that WOA just flies, or it’s going to solidify skepticism in the whole platform. Are you optimistic that MS can deliver an M1-like near seamless transition and performance parity this time?
Nope. Other than 5g, I see no delta to owning ARM vs x86, and I have NO compatibility issues to deal with. At least with my experiment to go ARM (with the M1 Pro) I got incredible integration across devices, an advantage I still miss…