|
Great performance from following Apple's design lead (clear, 2-level big.little segmentation, rather than the usual 3-4), but some improvements that can easily be implemented by anyone using the ARM ISA, apart from the L3 cache removal - even that could change, though. Expect AMD and Nvidia to bring it on PC after the Qualcomm-Microsoft exclusivity agreement ends this year. Remember that ARM's in-house designers are kinda second-class - look at the Dimensity 9400 on the charts in the YouTube video below. Even that's close to the A18 and M4, and closer to the M3 and A17 Pro. If this continues long-term, Apple is in serious trouble. X Elite was a hack-job on TSMC N4P. This on N3E is dead-even low power performance and a clear lead in high power CPU. Qualcomm has actually been ahead in GPU performance on mobile for about 2 years; Android just needs to clean up the graphics stack and devs need to get on board. (Source: https://www.youtube.com/watch?v=__9sJsKHBmI) Qualcomm supplies Meta and future players like Sony with VR SoCs as well, so there's the Vision Pro buried. The Qualcomm-Microsoft exclusivity agreement ends for PCs this year, and I'm pretty sure Nvidia isn't going to mess up on their attempt in 2025. AMD is also planning ARM for 2025. Linux has been quietly building ARM support, ironically with the help of the Asahi Linux team, and Steam's Proton recently merged ARM patches - there's a real possibility the next Steam Deck goes ARM. This will do wonders for native ARM gaming. There is an actual possibility now that major studios take Linux's early technical advantage, move to Linux to avoid Microsoft's AI bullshit, and Adobe, Ableton and a few other platforms follow. DaVinci Resolve actually already works quite well natively on Linux. Nvidia will probably keep selling high-end GPUs, and PCIe access will probably be good enough that ARM on desktop will be a lot better than whatever Apple's done with the Mac Pro. Apple has burnt bridges with Nvidia, and AMD has never quite settled into the macOS driver environment well, even with a bigger market on offer (Intel 16" MacBook Pro, RDNA2). It is actually possible to have fast M.2 storage with an ARM CPU, as well as expandable RAM with CAMM modules, so Apple will be in trouble for that bullshit too, pretty quickly as soon as Qualcomm, Nvidia and AMD start releasing ARM motherboards. Expect the small form factor case market to do great. Fedora is also incredibly reliable lately, and recent efforts on sandboxing apps with Flathub mean that 99% of people will be able to use immutable, version-controllable versions that are as reliable as, if not better than macOS, let alone Windows. Both Gnome and KDE are progressing fast, and new drivers for ARM will be Wayland-first, as is Asahi Linux. Throw in early AMD/Nvidia ARM SoCs using TSO execution for x86 translation, and some 16k page size patches on Linux, and Elsevier might be supporting Complete Anatomy on Linux before Apple tries to obsolete my M1 MacBook Air. In mobile alone, the Chinese domestic market is a pretty good indicator of what's next - ridiculously fierce competition between BBK (Oppo, Vivo, OnePlus), Xiaomi and Huawei, with Samsung way behind and Apple mostly relying on a unified app store and brand value to retain ~20% marketshare. Play Store doesn't exist there, so Chinese apps on Android are a mess, but they do have the benefit of WeChat being cross-platform. The future in the U.S. and particularly Europe, where iMessage won't help, is arguably even worse for Apple. The claim that high-schoolers are going to stick to iOS for iMessage isn't going to hold up to this much pressure, and no $3 trillion company is going to survive on Mean Girls (2004). No high-schooler, given the option to save ~US$500 and get a device that takes artsy film shots (look at Vivo and Xiaomi, and even OnePlus in the U.S.) and runs every app faster and better, is going to cough it up. Once this frenzy is in full swing, if OnePlus expands the Hasselblad software processing to some cheaper models, improves social media app photo processing pathways, and holds a few workshops on artsy film shots at college campuses or NYC and LA, they're blowing open the U.S. market, ditto Europe. Apple's current weakness on iPadOS is also going to bite really badly if Microsoft or anyone else manages to get a good ARM SoC into a Surface Pro or something similar. As long as the on-screen keyboard appears reliably, power efficiency is good, and app support is there, Windows can do the rest. If Windows doesn't get its act together, Google, DaVinci and Adobe will do it. LumaFusion, the former golden child of iPad vloggers around iOS 12 when everyone was excited about the 2018 iPad Pro, is already on Android, as is GoodNotes. It's an exciting time to be in consumer tech, and a terrible time to be Apple. |
Any insight into the failure/withdrawal of Qualcomm X Elite dev kit? They had every opportunity to hit the ground running in 2024, with huge pent-up demand from the Linux community.
Is there a viable path for Pixel Tablet (Samsung + Google) silicon to compete with Apple and Qualcomm?
> AMD/Nvidia ARM SoCs using TSO execution for x86 translation, and some 16k page size patches on Linux
That's exciting news. Hopefully they can avoid QC fiasco with device trees, by fully implementing Arm SystemReady.