First Apple Silicon Benchmarks Destroy Surface Pro X

As expected, developers with early access to Apple silicon-based transition kits have leaked some early benchmark scores. And it’s bad news for Surface Pro X and Windows 10 on ARM fans.

According to multiple Geekbench scores, the Apple Developer Transition Kit—a Mac Mini-like device with an Apple A12Z system-on-a-chip (SoC), 16 GB of RAM, and 512 GB of SSD storage—delivers an average single-core score of 811 and an average multi-core score of 2871. Those scores represent the performance of the device running emulated x86/64 code under macOS Big Sur’s Rosetta 2 emulator.

Windows Intelligence In Your Inbox

Sign up for our new free newsletter to get three time-saving tips each Friday — and get free copies of Paul Thurrott's Windows 11 and Windows 10 Field Guides (normally $9.99) as a special welcome gift!

"*" indicates required fields

This field is for validation purposes and should be left unchanged.

Compared to modern PCs with native Intel-type chipsets, that’s not all that impressive, but that’s to be expected since it’s emulated. But compared to Microsoft’s Surface Pro X, which has the fastest available Qualcomm-based ARM chipset and can run Geekbench natively—not emulated—it’s amazing: Surface Pro X only averages 764 on the single-core test and 2983 in multi-core.

Right. The emulated performance of the Apple silicon is as good or better than the native performance of the SQ-1-based Surface Pro X. This suggests that the performance of native code on Apple silicon will be quite impressive, and will leave Surface Pro X and WOA in the dust.

Share post

Please check our Community Guidelines before commenting

Conversation 73 comments

  • digiguy

    Premium Member
    29 June, 2020 - 2:46 pm

    <p>And this is with the ipad chip… which, as Craig said in his interview, show how good Apple is when «&nbsp;it’s not even trying…&nbsp;»</p><p>And based on another a recent leak, Apple is going to flex its muscles with the next ARM macs… They have decided not to give any more battery life than current macs, but to go with the maximum performance possible. These macs are going to outperform their Intel counterparts by up to 100% and base on this Rossetta leak, they are going to feel very fast even in emulation….</p><p>I guess this is going change the mind of many of those who believe that ARM chips are a step backwards compared to Intel…</p>

  • brisonharvey

    Premium Member
    29 June, 2020 - 2:50 pm

    <p>Just a couple of additional notes: the multi-core score only included 4 cores instead of maximizing all 8 (probably just using the 4 maximum performance cores). Also, the A12Z score from the 2020 iPad Pro was around 1100 single core and 4650 multi-core. As a reference, Surface Pro 7 (i5 variant) scored around 1200 single core, 4450 multi-core. </p><p><br></p><p>Windows on ARM needs to become a priority for Microsoft. Like 5 years ago. </p><p><br></p>

  • SWCetacean

    Premium Member
    29 June, 2020 - 2:53 pm

    <p>This result, while unfortunate for the SPX, isn't surprising to any Android fan. Apple's chips have always been quite a bit ahead of any Arm chip that Qualcomm has been able to release. Over in Android world, people have been clamoring for a higher-performing CPU for years. Hopefully Arm's new Cortex-X program for semi-custom high-performance architectures will eventually produce better-suited CPUs.</p>

  • shark47

    29 June, 2020 - 2:58 pm

    <p>That is pretty devastating. Microsoft had a multi-year head start here and still messed up. </p>

    • Stokkolm

      29 June, 2020 - 3:06 pm

      <blockquote><em><a href="#550836">In reply to shark47:</a></em></blockquote><p>Apple has been developing their own A-series chips since the A4. Microsoft in no way had a head start. Qualcomm on the other hand….</p>

    • lvthunder

      Premium Member
      29 June, 2020 - 3:50 pm

      <blockquote><em><a href="#550836">In reply to shark47:</a></em></blockquote><p>I think you meant to say Qualcomm messed up. Microsoft doesn't make it's own chips. Maybe Windows on ARM works well if it's on those Apple chips. Maybe all Microsoft needs to do is get emulation for X64 going and convince developers to create ARM binaries.</p>

      • kitron

        29 June, 2020 - 4:43 pm

        <blockquote><em><a href="#550846">In reply to lvthunder:</a></em></blockquote><p><br></p><p>We are not seeing any updates on the X64 emulation front. </p>

      • shark47

        29 June, 2020 - 5:15 pm

        <blockquote><em><a href="#550846">In reply to lvthunder:</a></em></blockquote><blockquote><em>Qualcomm, yes, but no one who buys a Surface Pro X will blame Qualcomm. </em></blockquote><p><br></p>

        • Paul Thurrott

          Premium Member
          30 June, 2020 - 8:33 am

          Especially since this is how Microsoft describes it:

          “Powered by Qualcomm®, the new custom Microsoft® SQ1™ processor delivers multitasking laptop performance, all-day battery life, and fast LTE and Wi-Fi connectivity.”

          It’s only powered by Qualcomm. But it’s a Microsoft processor. So goes the marketing.

      • Paul Thurrott

        Premium Member
        30 June, 2020 - 8:36 am

        “Powered by Qualcomm®, the new custom Microsoft® SQ1™ processor delivers multitasking laptop performance, all-day battery life, and fast LTE and Wi-Fi connectivity.”

      • bkkcanuck

        05 July, 2020 - 12:45 pm

        <blockquote><em><a href="#550846">In reply to lvthunder:</a></em></blockquote><p>Microsoft messed up in selecting Qualcomm – and yes Qualcomm messed up… but the fact is that Qualcomm is not hungry enough to care enough about it… Microsoft should have partnered with and taken a minority position in a company over a long development period with the assurances (i.e. hard to cancel) that they would become the premiere select supplier recommended for all authorized partner products and Microsoft would commit to optimizing its OS for all products that company supplies. AMD would have been a better company to partner with (could even make it a subsidiary of AMD where MS owns maybe 20% and AMD 80% – and gets all the technology support from AMD the parent but also funds to do required acquisitions). i.e. Microsoft provides the capital, AMD provides the technology.</p><p><br></p><p>Microsoft too often seems to do last minute hacks without enough long term committed strategy behind it. </p>

    • red77star

      29 June, 2020 - 4:05 pm

      <blockquote><em><a href="#550836">In reply to shark47:</a></em></blockquote><p>Windows does not need ARM. Microsoft just need to focus on actual Windows targeting x86-x64. The biggest mistake Microsoft did is to try to copy Apple, they will always fail at it. They should play their game called Windows and those things Google and Apple will never be. That's where Windows success came from in the past.</p>

      • lvthunder

        Premium Member
        29 June, 2020 - 4:15 pm

        <blockquote><em><a href="#550856">In reply to red77star:</a></em></blockquote><p>It may not need ARM, but it should pursue it anyways. Servers in datacenters are starting to go ARM for the energy consumption and Intel and AMD really need competition. Also remember x86 goes back to the 70s. Maybe an updated chip instruction set is needed.</p>

    • wright_is

      Premium Member
      30 June, 2020 - 2:59 am

      <blockquote><em><a href="#550836">In reply to shark47:</a></em></blockquote><p>Not Microsoft, Qualcomm… It is a comparison of Qualcomm vs. Apple silicon.</p>

    • ewanmakepeace

      02 July, 2020 - 4:11 am

      <blockquote><em><a href="#550836">In reply to shark47:</a></em></blockquote><p>What is devastating (to me) is that Apple has announced the end of Boot Camp. Why can the new Arm MacBook Pros not be used to run Windows on Arm? Seems like a match made in heaven…</p><p>Although a Parallels (virtualisation) icon with a Windows logo was visible in the keynote, so maybe that will the the way forwards.</p>

      • Paul Thurrott

        Premium Member
        02 July, 2020 - 8:14 am

        There’s a fundamental misunderstanding here. Windows 10 on ARM is really Windows 10 on Snapdragon, not “ARM” generally. And Apple silicon is not “ARM” in that it is somehow compatible with what Qualcomm or other companies that license the ARM architecture offers. So in addition to Apple supporting Boot Camp for some reason—it’s simply not necessary anymore—Microsoft would also need to create, support, and endlessly update yet another version of Windows that would only run on Apple’s hardware. And that would be dumb and pointless.

        Virtualization and remote desktop are absolutely the way forward.

        • wright_is

          Premium Member
          02 July, 2020 - 9:58 am

          <blockquote><em><a href="#551527">In reply to paul-thurrott:</a></em></blockquote><blockquote>Virtualization and remote desktop are absolutely the way forward.</blockquote><p>As Apple have stated that x86_64 virtualization won't run on their silicon, only ARM based VMs, that leaves just remote desktop going forward.</p>

          • Paul Thurrott

            Premium Member
            03 July, 2020 - 9:18 am

            The goal here is the same as ever: To just that one legacy app you really need. If it’s a bit slower because of emulation, whatever. At least there’s the option.

      • wright_is

        Premium Member
        02 July, 2020 - 9:57 am

        <blockquote><em><a href="#551502">In reply to ewanmakepeace:</a></em></blockquote><p>Because "ARM" isn't ARM. Every SoC is slightly different and not 100% compatible. As Paul says, WoA is the wrong name, it is WoS. They'd need to put a lot of work into getting it to run on Apple, plus drivers and it is doubtful that Apple would provide the drivers.</p><p><br></p>

        • bkkcanuck

          06 July, 2020 - 1:10 pm

          <blockquote><em><a href="#551574">In reply to wright_is:</a></em></blockquote><p>[Because "ARM" isn't ARM. Every SoC is slightly different and not 100% compatible.]</p><p><br></p><p>Actually every ARM CPU is ARM compatible as an architectural license REQUIRES it. SoC is just an implementation of other components on the same silicon as the CPU — that might have resided on a motherboard separately (like the accelerometer, or neural engine co-processor) — this is no different in reality than what exists on different motherboards as not all motherboards are the same and you require a bundle of different drivers for different system components. </p>

  • gmenfan

    29 June, 2020 - 3:08 pm

    <p>I don't see how this is all that bad for Microsoft. To me, this comparison is almost meaningless, or at a minimum, it might be used as evidence that Qualcomm is no match for Apple. But I think that most folks already knew this. Given Microsoft is dependent on Qualcomm for the "best available" ARM silicon one could say this is bad news for Microsoft and Surface Pro X but its not like Microsoft has much of a choice. The same statement could be made for ARM based Chromebooks, other ARM based tablet not using Apple chips.</p>

    • bkkcanuck

      05 July, 2020 - 7:43 pm

      <blockquote><em><a href="#550838">In reply to gmenfan:</a></em></blockquote><p>Google has decided that the general ARM chips out there are NOT sufficient for their purposes and are following Apple's lead of developing their own chips in-house. (code named: whitechapel apparently).</p>

      • Paul Thurrott

        Premium Member
        06 July, 2020 - 7:10 am

        Well, so has Microsoft. It’s using Qualcomm, which also develops its own chips based on the ARM architecture.

  • lilmoe

    29 June, 2020 - 3:37 pm

    <p>I don't think that's an issue, at all. First, you can buy a better performing x86 PC with a similar power envelope. No emulation needed. Second, by the time Apple hits the market, SoCs with ARM's X1 cores will be ready for Windows X devices, and can more cores with significantly higher clocks than their smartphones counterparts for devices with more thermal headroom and larger batteries. </p><p><br></p><p>The "devastating" blow is when you see Apple employ dedicated co-processors that will destroy any CPU/GPU core in specialized workloads, like video editing… </p><p><br></p><p>It's too early to see the implications of that move. Time will tell. </p>

    • lvthunder

      Premium Member
      29 June, 2020 - 3:51 pm

      <blockquote><em><a href="#550842">In reply to lilmoe:</a></em></blockquote><p>Why don't you think AMD and NVidia will make their own co processors and stick them in the graphics cards that can do the same thing?</p>

      • red77star

        29 June, 2020 - 3:55 pm

        <blockquote><em><a href="#550847">In reply to lvthunder:</a></em></blockquote><p>Nvidia does not have to. Have you guys ever used Nvidia GPU for video editing? It is crazy fast.</p>

        • lvthunder

          Premium Member
          29 June, 2020 - 4:11 pm

          <blockquote><em><a href="#550850">In reply to red77star:</a></em></blockquote><p>I don't edit video very often, but I have an NVidia GPU in my Surface Book 3.</p>

    • red77star

      29 June, 2020 - 3:52 pm

      <blockquote><em><a href="#550842">In reply to lilmoe:</a></em></blockquote><p>No Apple co-processor can compete Nvidia GPU Core for video editing. Let's get that BS out of way. Afterall Intel is also working on new GPU. This Apple thing will work only for them and won't have any meaning in PC world.</p>

      • bkkcanuck

        05 July, 2020 - 12:29 pm

        <blockquote><em><a href="#550848">In reply to red77star:</a></em></blockquote><blockquote>No Apple co-processor can compete Nvidia GPU Core for video editing. Let's get that BS out of way. Afterall Intel is also working on new GPU. This Apple thing will work only for them and won't have any meaning in PC world.</blockquote><p>Never say never…. Even if Nvidia GPU in isolation is faster than anything Apple can product in isolation – does not mean the holistic solution that Apple produces cannot surpass that of what it would be with nVidia (since nVidia requires you to go through an extra layer). I suspect (without proof) that Apple is working with Adobe on more than just porting the application over to the ARM based solution – but helping them transition fully to using Metal to use full low level access to all components that Apple may build into the machine. i.e. low level access to multiple GPU processors, FPGA specialized processors, Machine Learning and Neural Engine processing. There is a lot of optimization left on the table by Adobe products at this point in time (IMHO).</p><p><br></p><p>Apple at this point in time likely employs more chip designers than a company like AMD. Worst case scenario – AMD has shown a willingness to license it's technology.</p>

      • lilmoe

        06 July, 2020 - 1:39 am

        <blockquote><em><a href="#550848">In reply to red77star:</a></em></blockquote><p>Maybe for GPGPU compute. But fixed function hardware and neutral networks are a completely different beast. If not in performance then in power efficiency. Have you tried playing 4k video on a cpu that has dedicated blocks for 4k h.265 VS cuda cores? Now apply that to Adobe's filters, layers, NR, etc. Yea, totally not BS.</p>

  • mattbg

    Premium Member
    29 June, 2020 - 3:49 pm

    <p>The raw performance shouldn't be a huge surprise – the iPad Pro 2020 was shown to be outperforming the Macbook Air in GeekBench, but the figures in emulation are very interesting.</p><p><br></p><p><a href="https://arstechnica.com/gadgets/2020/03/ok-computer-the-ipad-pro-2020-review/&quot; target="_blank">https://arstechnica.com/gadgets/2020/03/ok-computer-the-ipad-pro-2020-review/</a></p&gt;

  • red77star

    29 June, 2020 - 3:50 pm

    <p>And both are joke compared to x86-x64. ARM has no place in Windows world. It is that simple. How about Microsoft focuses on x86-x64 and Windows. Also Geek Bench is not a real indication of anything.</p>

    • anton1900

      01 July, 2020 - 8:21 am

      <blockquote><em><a href="#550845">In reply to red77star:</a></em></blockquote><p>You rather Microsoft continues down the path that for sure will make Windows even more irrelevant than it already is? Sorry, but you dont have a clue what you are talking about.</p>

      • Paul Thurrott

        Premium Member
        01 July, 2020 - 9:12 am

        Disagree with the ideas. Let’s not make it a personal attack.

  • lvthunder

    Premium Member
    29 June, 2020 - 3:53 pm

    <p>So maybe most of WOA's problems are the chips they are running on and not the OS itself. Maybe all Microsoft needs to do is get X64 emulation working and convince developers to create ARM binaries. Does anyone know how much work it is to create an ARM binary for an existing app?</p>

    • Rafael Rivera

      Premium Member
      29 June, 2020 - 5:16 pm

      <blockquote><em><a href="#550849">In reply to lvthunder:</a></em></blockquote><p>For some (most?) apps, it's not particularly difficult. A recompile typically suffices. But some factors that slow that progression include (but aren't limited to): a. the developer needing to purchase hardware for test; and b. there's no native .NET Framework (e.g. WPF, WinForms) support in Windows 10 on ARM.</p>

      • lvthunder

        Premium Member
        29 June, 2020 - 7:34 pm

        <blockquote><em><a href="#550889">In reply to WithinRafael:</a></em></blockquote><p>Microsoft should really fix B in a hurry. I had just assumed it was there.</p>

    • wright_is

      Premium Member
      30 June, 2020 - 3:30 am

      <blockquote><em><a href="#550849">In reply to lvthunder:</a></em></blockquote><p>The latter. All the time they are running emulation, it is never going to be an optimal experience.</p><p>And, maybe, this is the thing that will get Qualcomm's rear into gear. The talk has always been about ARM scaring Intel, but with Qualcomm at the helm, it has been more of a light tap as opposed to a hefty kick in the rear.</p><p>As to Rafael's b) the problem is, WoA was supposed to be "the new", so converting "legacy crud" was not a priority. The problem is, Microsoft has been so back-and-forth and cannot force the "new" on anyone, let alone let go of the old, that most software still uses the legacy APIs.</p><p>That is a problem Apple doesn't suffer from. They just tell devs to get with the program, we are dropping that old rubbish next week! </p><p>Due to the different markets, that is something Microsoft can never do. Corporates have so much tied up in that "legacy crud", that they can never drop it, and whilst the devs are already comfortable with it and Microsoft doesn't force them to use the new stuff, the apps remain "legacy" apps and the new stuff fails, time and again.</p>

  • curtisspendlove

    29 June, 2020 - 6:05 pm

    <p>There was an interesting quote from the Apple guys (I think Craig Federegi) about the DTK hardware.</p><p><br></p><p>That kit is intended only for developers and is using a non-modified version of an iPad Pro chip. </p><p><br></p><p>He said “that is without them even trying.”</p><p><br></p><p>The real machines that start to come out this fall are “when they are actually trying [to make a Mac version of the chip]”. </p><p><br></p><p>I’m looking forward to it. I’m guessing a new MacBook will be the best hardware with which to run Windows on ARM. ;)</p>

  • IanYates82

    Premium Member
    29 June, 2020 - 11:58 pm

    <p>Keen to see more. I'm wary this still isn't quite a direct comparison as we'd like due to the surface being bound thermally and the dev kit being some AC powered thing.</p>

  • Philotech Mueller

    30 June, 2020 - 4:04 am

    <p>Important when comparing the results:</p><p>Keep in mind that most likely this is no just-in-time emulation, but the code was translated upon load and is then being executed natively. So apparently we're missing all sorts of optimization, but it's not like the DTK has to interpret x86 code, translate it to ARM and then execute it. That would definitely result in low three-digit single-core scores and high three-digit multi-core scores.</p>

  • geoff

    30 June, 2020 - 4:20 am

    <p>I don't understand the negativity here.</p><p><br></p><p>Apple is just 6.15 percent faster on single core, and about 3.75% <u>slower </u>on multicore.</p><p>And most people, including Paul, generally accept that benchmarks are untrustworthy most of the time.</p><p><br></p><p>To use words like Apple 'destroyed' SPX, or that it was 'left in the dust', is crazy. </p><p><br></p><p>Those figures are more or less equal. </p><p><br></p><p>It tells me that a SPX is on a par with Apple's developer kit, which is fine.</p>

    • gazaian

      30 June, 2020 - 5:39 am

      <blockquote><em><a href="#551022">In reply to Geoff:</a></em></blockquote><p>Look closer, the A12Z is running Geekbench with Rosetta emulation, while the Surface Pro is running it natively. Emulating the x86 instruction set means a ton of overhead is introduced, so you lose a ton of performance to the emulation. If Geekbench was running using WOW with x86 emulation then your comment would be true, and this would be no big deal. But these numbers essentially mean that when ARM to ARM performance is compared, the A12Z should completely destroy the SQ1. </p><p><br></p><p>I'm excited honestly, perhaps this will be the push we need for better ARM chips. </p>

      • Philotech Mueller

        30 June, 2020 - 5:52 am

        <blockquote><em><a href="#551025">In reply to gazaian:</a></em></blockquote><p>Hm. As written in the post just below, it needs to be taken into account that "emulation" can mean a ton of things. From the keynote I understand that Rosetta 2 tries to translate all code upon load (resulting in longer load times) and then executes the translated code natively. You're missing all the optimizations of real natively compiled code, but there shouldn't be any (or not much at least) overhead.</p>

    • Paul Thurrott

      Premium Member
      30 June, 2020 - 8:22 am

      Benchmarks are no measure of real-world performance, yes. But we’re all very interested to see where this thing lands, and this is our first apples to apple comparison. So it’s notable.

      You seem to have missed the bit about the benchmark running natively on WOA/Surface Pro X and under emulation on the Mac/Apple silicon. The latter being roughly comparable in performance to the latter while running under emulation indicates that the real-world performance gap will be much, much higher. We’ll see: There are ways to game benchmarks, and I’ve always wondered if Apple’s A-series advantage was tied to this. But this is what we have to go on now.

      • michael_babiuk

        30 June, 2020 - 11:02 am

        <blockquote><a href="#551061"><em>In reply to paul-thurrott:</em></a><em> </em> I thought I might have read somewhere that for this DTK Mac Mini A12Z machine, Apple chose to use ONLY half its available cores for developer work.</blockquote><blockquote>In an article by Mike Tyson, he states that the “…<span style="background-color: rgb(255, 255, 255);">Another important observation is that the A12Z-based system under test is only utilising half of its eight CPU cores in Geekbench, presumable the big, rather than LITTLE, cores.”</span></blockquote><blockquote><span style="background-color: rgb(255, 255, 255);">If true, even more amazing, Paul.</span> </blockquote><p><br></p>

        • Paul Thurrott

          Premium Member
          30 June, 2020 - 12:17 pm

          Yep.

    • nbplopes

      01 July, 2020 - 8:10 pm

      <blockquote><em><a href="#551022">In reply to Geoff:</a></em></blockquote><p><br></p><p>“And most people, including Paul, generally accept that benchmarks are untrustworthy most of the time.”</p><p><br></p><p>The thing about benchmarks is that if it does not fit our believes are <span style="background-color: rgb(255, 255, 255);">untrustworthy, if it does than confirms our believes.</span></p><p><br></p><p><span style="background-color: rgb(255, 255, 255);">The fact is the benchmark software has evolved to the point that results are close to what one might expect in a daily work environment. Since around 2011 you can actually see benchmarks as a reliable tool to compare performance across systems … even though it can be tricked. </span></p><p><br></p><p><span style="background-color: rgb(255, 255, 255);">Welcome to the era of passive cooling, and no throttling at unchallenged speeds. If Apple delivers this it will be like the iPhone again but now on the Desktop and Laptops. It took the industry 5 years to catchup and still …</span></p><p><br></p><p>What intrigues me is the GPU. Because that’s is were the Desktop PC mostly evolved in the last 15 years. Apple up until now has not presented much on that regard and the demos were not really that convincing. For instance, in the Maya demo very little was shown.</p><p><br></p><p>I was more impressed with the Tomb raider demo … it meant that iPad Pro games are not really pushing the graphics capacity that much But was still far away from what a medium to high end GPU can deliver. </p><p><br></p><p>What intrigues me is that Apple set the transition time so short. Two years. This timing will only work if the gains are without question quite noticeable. Otherwise if the Mac had a small market share in comparison with other it will end up being even shorter. So either the short time its a bluff or they actually have something. Because I’m sure they are not aiming coming out only with something destroying a dead horse like the Surface Pro X… hopefully crack sessions stayed with the marketing when naming the macOS 11.0.</p>

    • ewanmakepeace

      02 July, 2020 - 4:07 am

      <blockquote><em><a href="#551022">In reply to Geoff:</a></em></blockquote><p>We already knew the native Geekbench scores of this chip from iPad, so this story is looking at the numbers in the wrong way. Rather than using them to evaluate the chip, we should be using them to evaluate the Rosetta emulation. It seems that Rosetta is imposing a penalty of 27% vs native scores. I for one am mildly disappointed – I had hoped that with all of their LLVM compiler experience and with the advantage of translation on first install, they might have got this penalty down a bit.</p>

      • Paul Thurrott

        Premium Member
        02 July, 2020 - 8:14 am

        That’s what we are evaluating here. The prototype Mac is running Geekbench in emulation.

  • darlingtonpear

    30 June, 2020 - 6:49 am

    <p>.NET Core 5 Preview 6 SDK now includes support for Windows Forms. That means you can build and run Windows Forms apps on Windows ARM64.&nbsp;</p><p><br></p>

    • red77star

      30 June, 2020 - 8:03 am

      <blockquote><em><a href="#551034">In reply to darlingtonpear:</a></em></blockquote><p>That's all nice but who cares about ARM on Windows?</p>

      • Paul Thurrott

        Premium Member
        30 June, 2020 - 8:14 am

        Obviously, many people reading this site care very much about Windows on ARM.

    • Paul Thurrott

      Premium Member
      30 June, 2020 - 8:18 am

      What’s old is new again! Well, it’s still old. 🙂 But compatibility is compatibility.

    • lilmoe

      06 July, 2020 - 1:23 am

      <blockquote><em><a href="#551034">In reply to darlingtonpear:</a></em></blockquote><p>Yes, but WinForms is anything but a modern UI framework. WPF has been fully supported since .NET 3.1, but it's also a donkey at this point. Both were ported to .NET Core and 5+ for the sole purpose of getting everyone on board in .NET Core and beyond. </p><p><br></p><p>What they need is a true, modern, feature-equivalent successor to Win32 and its subsets (WinForms, WPF, MFC, etc). I don't mind if it's Windows-only. But it has to be Direct2D first, low latency, with no software rendering layer to minimize overhead. </p>

      • Paul Thurrott

        Premium Member
        06 July, 2020 - 7:04 am

        Windows-only isn’t happening in 2020. That’s over.

        Embrace the point of .NET Core: Interoperability and cross-platform is the way forward.

        • lilmoe

          11 July, 2020 - 12:15 am

          <blockquote><em><a href="#552380">In reply to paul-thurrott:</a></em></blockquote><p>Feature equivalent does not mean Windows-only, but it could start there. If anyone can pull off a universal desktop UI framework, it's Microsoft. As I said before, Maui is a good start, but exactly desktop first. If they won't go it, that leaves Apple as the only ones who do. </p>

          • Paul Thurrott

            Premium Member
            11 July, 2020 - 9:24 am

            What previous experience suggests that Microsoft could pull that off? 🙂

  • christian.hvid

    30 June, 2020 - 8:34 am

    <p>For reference, the Dell XPS 13 9300 (i7) scores 4194 / 10883, or about 300% faster. Even with a more optimized chip and no translation overhead, Apple will have a hard time keeping up with mainstream Windows laptops. Not to mention high-end Windows workstations.</p><p><br></p><p>Unless, of course, they have something truly spectacular up their sleeves.</p>

    • michael_babiuk

      30 June, 2020 - 10:52 am

      <blockquote><a href="#551067"><em>In reply to christian.hvid:</em></a><em> Regarding “Unless, of course, they have something truly spectacular up their sleeves.”‘ </em>conjecture.</blockquote><blockquote>Tim Cook indicated that the transition to ARM based “Apple Silicon SoC chipsets would take two years. 2 years and even the new Mac Pro Intel Xenon based workstation will have Apple Silicon chipsets. Just think about that for one second and the conjecture about “something truly spectacular” might be proved true in just under two years!</blockquote><p><br></p>

      • christian.hvid

        30 June, 2020 - 12:21 pm

        <blockquote><em><a href="#551102">In reply to Michael_Babiuk:</a></em></blockquote><p><br></p><p>Here's what I don't get: assuming that Apple is going to replace Intel's silicon in <em>all </em>of their models, including the Mac Pro; assuming (generously) that their best processors today can reach 20% of the performance of a Xeon W; and assuming that Intel will be able to double the performance of their Xeons in the next two years…. then Apple would have to achieve a <em>ten-fold</em> performance increase just to keep up with Intel, and they would still be <em>vastly</em> outperformed by AMD.</p><p><br></p><p>On the other hand, if I can figure this out, then so can Apple. And they wouldn't announce this two-year target unless they knew they could meet it. It's all very mysterious.</p>

        • lvthunder

          Premium Member
          30 June, 2020 - 7:18 pm

          <blockquote><em><a href="#551124">In reply to christian.hvid:</a></em></blockquote><p>I highly doubt Intel will be able to double the performance. Maybe they can get 10-20% max.</p>

          • christian.hvid

            01 July, 2020 - 5:54 am

            <blockquote><em><a href="#551229">In reply to lvthunder:</a></em></blockquote><p>Clearly you haven't heard about the upcoming 14nm+++++++ technology. :)</p><p><br></p><p>I think a doubling of performance in two years sounds achievable even for Intel, but that's kind of beside the point. The point is that from what I can see, it sounds almost impossible for Apple to catch up with Intel, let alone AMD, in the HEDT/workstation arena in just 24 months. But that's likely more due to my limited knowledge than anything else. </p>

            • Paul Thurrott

              Premium Member
              01 July, 2020 - 9:14 am

              Suspect they’ve been working on that in secret for years and have met some milestone.

              • bkkcanuck

                05 July, 2020 - 11:23 am

                <blockquote><a href="https://www.thurrott.com/apple/237225/first-apple-silicon-benchmarks-destroy-surface-pro-x#551067&quot; target="_blank"><em>In reply to christian.hvid:</em></a></blockquote><p>First your benchmarks DO NOT match the Geekbench 5 numbers in the browser – the absolute TOP of all the ones I see for: Dell Inc. XPS 13 9300 – Intel Core i7-1065G7 (1302 / 4840) about 55% / 64% faster than a crippled iPad Pro processor meant to run in a more restricted thermal environment).</p><p><br></p><p>Benchmark (native) running on the iPad Pro are: 1118 / 4626 or basically that XPS running about 16.5% / 4.6% faster than a non-crippled iPad Pro processor (still meant for a much lower thermal limit with no fans).</p><p><br></p><p>This IS NOT the processor that will be in the Macs released later this year… they are not including pre-production chips because those are under wraps so they have as much of a wow factor as possible.</p><p><br></p><p>https://browser.geekbench.com/v5/cpu/search?utf8=%E2%9C%93&amp;q=Dell+XPS+13+9300+i7</p><p><br></p><p>–</p><p>So how much of a handicap are you placing on this unreleased processor?</p><p><br></p><ul><li>reject chip meant for the iPad Pro and is effectively a 2018 chip design</li><li>designed for a VERY LOW TDP (thermal) envelope…. with no fans (effectively the same thermal envelope of the MacBook 12" – for which the Intel product that they used was widely panned for being a slug)</li><li>You are comparing it based on it not even using all the power of the cores (all efficiency cores not used)</li><li>You are comparing it based on translated binary code (probably equivalent to what Java JIT for binary). (with Java it can be the difference of being a multiple slower than good C code)</li></ul><p><br></p><p>That is not the chip they will use, those are the discards which could not run at the same speeds as what are released in the iPad Pro).</p><p><br></p><p>Being an ARM instruction set processor (not an ARM processor) (sort of like AMD vs Intel – can vary widely dependent on the quality of your chip designers) does not inherently make it slower. The most powerful supercomputer in the world runs on ARM instructions. AWS (Amazon) new AWS Graviton2 instances – are ARM based.</p><p><br></p><blockquote><a href="#551302"><em>In reply to paul-thurrott:</em></a></blockquote><p>Likely absolutely correct, Apple wouldn't show up the day before at the fab and say – make these chips and then put them in a released product. Apple would already have preproduction (finalized design) chips for which they are using at the end of this year – all taped out and likely being produced in the quantity they expect to need before the first computer comes off the assembly line. And no, that would not be the one in the DTK, the DTK ones are 'discards' from the iPad Pro. The ones Apple are planning to use in the Mac Pro have likely had samples made and are in beta testing internally. Apple would have everything they need to know whether they hit a milestone or not. Apple wants to have the greatest press impact once the first computers start showing up on store shelves — so Apple wants to keep expectations in check and then exceed as much as possible on release.</p><p><br></p><p><strong>Additional Thoughts</strong></p><p><br></p><p>The cost to fab an A series (Apple) chip is closer to something like $40 (not including R&amp;D which is a sunk cost upfront for the entire chip design team) — while the cost for an equivalent Intel chip would range from $250 to $500. Intel typically handicaps it's entry level processors and offers a series of different tiered processors (up to the high end for that thermal envelope)… to drive revenue and profits. It will not be worth it to Apple to do this sort of thing, so they will be designing the most powerful chip for a thermal envelope… resulting in the highest power chip for a complete line (i.e. the complete MacBook Air line or the complete MacBook Pro 13" line). There price will be driven by screen size and options (SSD, memory, LTE modem, etc.). The entry level price for the MacBook will likely drop ($100), but Apple could (not necessarily in year 1) do the same with the MacBook (Entry), MacBook Air, and MacBook Pro that they did with the iPad line… where you have a very high volume, very reasonable entry level price. This will lead to a much better processor in the entry level model (much bigger improvement than a fully optioned model), and a much lower price for the fully decked out model (maybe $400 cheaper for the same options). </p><p><br></p><p>Microsoft partners will be demanding Microsoft have an answer, but with Microsoft leaving planning to the last minute it will likely lead to another last minute hack solution that may make things worse for those partners. </p><p><br></p><p>In the end though this dropping of Intel will benefit consumers in the long term.</p>

        • wright_is

          Premium Member
          01 July, 2020 - 5:45 am

          <blockquote><em><a href="#551124">In reply to christian.hvid:</a></em></blockquote><p>You are making the assumption that the A12z(?) from the iPad is the only and most powerful chip they have. They have been planning this for several years and will probably have performance chips as well as energy saving chips waiting in the wings.</p><p>ARM has been synonymous with low power, mobile solutions, but it started off as a desktop chip in the late 80s and early 90s, before it was adopted for low power solutions. There are also server manufacturers that use the ARM architecture for server chips and the most powerful supercomputer in the world is ARM based.</p><p>It is theoretically possible that Apple have a 24 or 36 core desktop processor in testing at the moment. With the increased TDP for desktop designs, there is a lot more theoretical performance than the miserly, power sipping silicon we know today, and which is no slouch.</p><p>But it is typical Apple, they are playing everything close to their chests, until products are ready to roll. It is already rumoured that Apple will sacrifice battery life for performance with the first laptop SOCs, meaning they will have MacBook Pro like power consumption/battery life, not iPad like power consumption/battery life and will therefore run much quicker than an iPad.</p><p>We will just have to wait and see, what exactly they have been doing for the last decade.</p>

          • christian.hvid

            01 July, 2020 - 6:02 am

            <blockquote><em><a href="#551270">In reply to wright_is:</a></em></blockquote><p>Well, it's more of an assumption that Apple is going to iterate on the low-power iPad chips (or a variant thereof), because that's the impression I got from the announcement. And while these probably offer more performance per watt than any other processors on the planet, it's kind of inevitable that a 200 W+ powerhouses like Xeon and Threadripper will have an edge. But I agree completely, we'll just have to wait and see.</p>

            • wright_is

              Premium Member
              01 July, 2020 - 6:46 am

              <blockquote><em><a href="#551281">In reply to christian.hvid:</a></em></blockquote><p>But, if they are going to make a desktop, they will use desktop TDPs, so a 200W Apple SOC isn't out of the question. The question then becomes, just how efficient is their design at that sort of TDP.</p>

              • christian.hvid

                01 July, 2020 - 7:25 am

                <blockquote><em><a href="#551283">In reply to wright_is:</a></em></blockquote><p>Considering that both Apple and AMD source their silicon from TSMC, I won't be surprised if they will end up having similar performance/watt. It's just that Apple made a point about using low-power chips in the new Macs – but maybe they were implicitly referring to just laptops. </p>

    • ewanmakepeace

      02 July, 2020 - 4:03 am

      <blockquote><em><a href="#551067">In reply to christian.hvid:</a></em></blockquote><p>I think you are confusing Geekbench 4 and 5. Under Geekbench 5 (which was used for the results in the story) I think the Dell XPS 13 9300 (i7) scores around 1427/4597. For comparison the A12Z used in the Developer Preview and also in the iPad Pro 2020 benchmarks (natively under iOS) at 1118/4621.</p><p>Natively this chip is almost 50% faster than the MS SQ-1 (not a huge surprise?) but is only 20% slower than your i7 per core – and is actually faster in multi core.</p><p>It is that last that should give you pause. This is not a story about Apple coming for Surface. It is about Apple exceeding Intel i9 on CPU, and completely demolishing it for (integrated) GPU before the end of the year.</p><p>Xeon in 2022? Absolutely.</p>

  • Chris Payne

    30 June, 2020 - 1:19 pm

    <p><span style="color: rgb(0, 0, 0);">Keep in mind the chip in the Apple DTK is a 2 year old off-the-shelf iPad processor. The ARM chip in the Surface Pro X is the latest hotness. When Apple sells products with the latest A14 chip and scaled up to whatever its "Apple Silicon" standards are, it will be that much more impressive.</span></p>

  • SYNERDATA

    03 July, 2020 - 2:50 am

    <p>Let's cut to the chase… In any case, Qualcomm would be increasing performance of the #cx chipset with Apple's solutions as their compare. The Win64 emulating SQ1 Plus in the SPX2 is expected to present performance increases even before SQ2. Already Snapdragon 8cx runs things fast enough and it is only going to get faster, meaning, either Windows on ARM or Bionic ARM are going to be wonderful experiences moving forward.</p><p><br></p><p>RISC ate CISC for breakfast.</p><p><br></p><p>Moving forward, like Macs, end-user PC's would transition to ARM and the experience of using either platform would be wonderful and impressive.</p><p><br></p><p>ARM is the road ahead.</p><p><br></p>

  • bkkcanuck

    05 July, 2020 - 7:21 pm

    <p>Paul, saw you on Windows Weekly and another show. To answer your question of why Apple will likely choose the MacBook Pro model and an iMac… is because Apple does not like prolonged confusion in their market — most times (and they also don't like telling you what they will deliver in advance)… If they chose to come at it by replacing their lowest end computer first and work up the line — questions would continue with regards to … ok, you can do it in low end hardware which is not a surprise … but are you still going to have problems in delivering it in the professional line. By delivering a higher 'pro' (laptop) first, and it exceeds the current line substantially and provides that 'wow' moment in the press… it will go a long way to soothing the nerves of the more demanding users… and limit the confusion in the marketplace.</p>

Windows Intelligence In Your Inbox

Sign up for our new free newsletter to get three time-saving tips each Friday

"*" indicates required fields

This field is for validation purposes and should be left unchanged.

Thurrott © 2024 Thurrott LLC