Month: September 2020

Apple released another beta build of macOS 11 ‘Big Sur’ (link) – it will be interesting to see whether beta 7 marks the move to the final stretch or whether Apple will keep pushing out updates to address the last few ‘known issues’. There Is a rumour going around that there will be another event held at the end of September regarding the iPhone 12 and the launch of the first Apple Silicon based Mac so it is plausible that they’re wanting to stabilise macOS as much as possible so that the first round of Apple Silicon based Mac’s give customers the best possible experience out of the gate (although like the Intel Mac’s I’m sure there will be a flurry of updates pushed out to fix up bugs developers find, to improve Rosetta optimisation etc). I have a strong feeling (completing guessing) that macOS 11.0 will be released when Xcode 12.2 and the x.2 series of iOS, iPadOS etc. will be released around the same time.

There was a podcast I was listening to a few days ago that claims Apple were looking at moving to their own ARM based designs for Mac as far back as the 2015 based on Intel falling behind the roadmap that Intel promised Apple but I personally disagree with that assessment. I disagree with with the assessment because of Apple’s historical tendency towards wanting to bring as much as possible ‘in house’ which enables them to control the complete process and for the components to suit their needs rather than a third party developing something aimed for a larger market and it kinda meets the needs of Apple. When Apple started to move to their own CPU design (vs. using a reference design) that long term Apple was going to move the Mac platform to their own CPU design – the situation wasn’t if but rather when. I would say that where the decision in 2015 kicks in is the decision to bring forward the time line – that it was always going to go ahead but sometime in the future but given what has taken place with Intel unable to meet it’s roadmap it brings forward the time line of transitioning over to its own CPU.

On Tuesday (NZ Time) Apple announced that iOS, tvOS, watchOS and iPadOS would be released on Wednesday (NZ Time) so when I woke up this morning I was ready to install it. On my iPhone it was a pretty hefty download of around 2.3GB but I was also surprised that Safari 14 was made available to macOS Catalina (I was expecting the rumoured ‘Supplemental Update’ that was talked about on 9 to 5 Mac (link)). With that I did a clearing out my macOS installation (falling down into single user mode and deleting the contents of ~/Library along with Caches etc).

I’m enjoying Safari 14, it loads websites faster – not only making the initial connection but also downloading and rendering the page as well. It doesn’t include 4K Youtube support in the form of VP9 support being added so that’ll have to wait until Big Sur is released but that being said – so far everything is working well. What I am hoping for is that gradually they’ll expand WebExtensions API so that it has the depth and breadth of support that Firefox has. Although Ad Guard does a pretty good job I do wish for uBlock Origin to make its way to Safari.

I’ll be writing up a longer review once Big Sur is released – I’m hoping that’ll be released while I’m on my holiday – I’ll do a clean install on all my devices so I start with a clean slate which will give a fair assessment of how Big Sur performs.

As we draw closer to the release of Apple’s first ARM based Macs it is interesting to watch various people speculate about the future – I generally avoid the clickbait in favour of Snazzy Labs, Rene Richie which have a more measured tone in terms of forecasting where they think Apple will go.

Although there is a lot of speculation about the performance of the SoC, the question is whether we’re going to see software vendors make use of the various parts of the SoC (Neural Engine processor, CPU and GPU) through frameworks provided (Metal, Metal Performance Shaders etc) or whether they use the CPU as a dumping ground resulting in customers getting a subpar experience (see Adobe drag their feet every step of the way when it comes to adopting new technology). There is a certain amount that Apple can do in terms of ensuring that various parts of the operating system’s frameworks are tapping into the underlying hardware but there is a limit particularly if software vendors are writing multiplatform software and need to utilise multi-platform frameworks such as Qt.

What has also happened recently was the announcement by nVidia to buy ARM from SoftBank (link). It’ll be interesting to see how it all turns out but given that it would provide nVidia with a CPU of their own, it will provide them with the opportunity to make use of ARM based SoCs for cloud servers and maybe even open up the possibility of ARM based laptops, desktops, workstations in the mainstream. Regarding how this impacts Apple – it doesn’t as Apple has a licence which gives them access to the ARM ISA but the design of the SoC itself is up to Apple to build from the ground up (vs. licencing a design from ARM as with the case of Samsung’s upcoming Exynos which will revert to standard CPU designs straight from ARM).

It will be interesting to see whether long term we might even see console vendors like Microsoft and Sony look into using ARM based nVidia SoC’s with nVidia GPU particularly if it means they can get improved performance but at a lower cost per unit which can either be passed on as savings to customers the ability to develop consoles beyond consoles – imagine a XBox running a stripped down modernised version of Windows where you boot up and choose between ‘gaming mode’ and ‘desktop mode’ – where a student could purchase an XBox, hook up a screen, mouse and keyboard, and have a games machine and computer that they can use for study all in one.

Something that I realised when thinking about the fact that the move to ARM based Mac’s will enable the running of native iOS/iPadOS applications on macOS about what it might mean for Google applications that run on iOS/iPadOS. For example, for years people on various Apple forums have complained about the subpar integration between macOS and Google services but with the movement to ARM based Mac’s might open up the opportunity for the native Gmail application appearing on MacOS along side a native YouTube Music app, Google Maps etc. Keeping in mind that for these applications to make their way onto macOS it is something that Google would have to opt in on – it’ll be interesting to see whether they go, “meh, might as well, it’s no skin off our nose” or whether they rigidly stick to their current position of “small screen’s get the apps, everyone else uses a web browser to access our services”.

Rumour has it that there is an ARM iMac being released next year – 24 inch but in the same case as a 21.5 inch by stretching the screen to the edge (27 inch being upgraded to a 32 inch by stretching the screen to the edge) but as noted in the past – I always let the early adopters take the 1.0 version and I’ll be waiting for the the next revision where hopefully most of the bugs will be found and corrected.

After much excitement was created after Samsung announced their return to using bog standard designs from ARM rather than a custom core architecture there was also the rumour that the next Samsung device next year (rumoured to be the Samsung Galaxy S21) will be shipping with it along with an AMD GPU. Unfortunately it appears, based on information from a source with a good reputation for reliability that although it will be using a bog standard ARM CPU design it will be staying with the Mali GPU rather than making the switch.

Although S5E9840（or Exynos1000） performance is not the strongest, but I think the power consumption will be better.

— Ice universe (@UniverseIce) August 27, 2020

One thing to keep in mind that although it will be using bog standard ARM designs it is also important to realise that they can be customised – for example when it comes to the Mali G78 it is possible to choose from between 7-24 cores, from 512KB to 2MB of L2 cache etc. not to mention the customisations that I’m sure can be done at the processor level. As long as the performance is ‘close enough’ to what Qualcomm offer then the only people who will quibble over a matter of a few points on synthetic benchmarks are those in the press and in online forums all while the average person will happily rock up and buy a Samsung device.

There is also a rumour that with US carriers phasing out their CDMA2000 networks and the rise of nationwide LTE support (including VoLTE support) that Samsung may look at dropping Qualcomm. This is particularly interesting given that there are handset vendors deciding to go with the Qualcomm 765G SoC as a success to the Qualcomm 865 SoC, Samsung using its own modem and SoC could mean that Samsung decides to standardise on their own SoC worldwide which would either allow them to bank higher margins, pass on the savings yet still maintain better profit margins that if they kept with Qualcomm. If the sales in the UK of the Exynos based Note 20 is anything to go by, Samsung not using Qualcomm outside of the US hasn’t hurt their sales so it would make sense – to make that leap and standardise on a single hardware platform in much the same way that Apple has done with it’s iPhone when stopped sourcing modems from two suppliers and decided to go ‘all in’ with Intel modems.