Why Are Apple and Microsoft Using Such Old Processors in Their New Computers?
gizmodo.com
gizmodo.com
The author seems to be living in a fantasy land, it takes time to test, purchase the processors, build the laptops, ship the laptops, and I'm sure I'm missing about 10 other bottlenecks in the process. The Pascal GPU was only released in May, and that is still too soon to have been designed around.
Missing are examples of computers that have shipped in high volume using 1-4 months old components. Missing is some realization that if both MS & Apple are doing the same thing, there's probably a good reason.
The problem is Intel's execution, not Apple or Microsoft. The original retina Macbook Pro launched with Ivy Bridge less than two months after that chip came out. But Intel managed to launch that with a full suite of SKUs. With both Skylake, Intel released the chips in trickles, with ULV parts appearing in August 2015 and Iris Pro parts appearing in May 2016.
I don't know specifics about either Microsoft's or Apple's manufacturing & shipping processes, but I have worked on a product that shipped from Asia, and everything was done, completely done, two months before the product could hit the shelves. Moreover, that two months was the software deadline, the easiest thing to change. The hardware, materials, components, design, that was all done and locked a year before the product shipped.
The vast majority of companies don't qualify for this, but MS and Apple are not most companies.
No idea, but I would speculate wildly the MacBook has a longer lead time than Intel CPUs do, the laptops being a more consumer product and for various other reasons.
Either way, access != volume, Apple can't ship Intel CPUs until it has them. But, it's clear you understand the issues. And so I assume it is obvious that having a laptop on the shelves with a chip that is less than 2 months old is not even realistic, right? You're not arguing that either company actually could have used Kaby Lake, right?
Both companies, according to the article, where in fact using the very latest GPUs and CPU on the market, had Intel and Nvidia not released new chips just as Apple and Microsoft where beginning production.
Samples don't do anything for high volume manufacturing, and a couple of months simply isn't enough time. You have to remember that the release date for MS and Apple products means those laptops are on the shelf. To get to the shelf, they shipped from Asia, which alone takes weeks. Before that they were manufacturing and stocking, before that they were testing like crazy and hotfixing everything they found for a couple of months, and simultaneously purchasing and shipping the components to the manufacturer. It's easy to see that component lockdown has to happen several months before the product is on the shelves, which means there is literally zero chance Kaby Lake was even a choice.
Instead there isn't really a story at all.
For utter noobs to modern technology: If the pipeline from daydream to press day is X months and a new widget was released Y months ago then if X > Y you don't see the widget on press day. Don't matter if its a jet plane or a nuclear reactor or a laptop.
All these machines work great now, but I wouldn't want to go through that pain again just for the sake of getting a little more performance (for average use).
I've got a machine that has a wireless card with the draft version of the wireless N standard, and presumably some promise that updates could fix it when the real thing came out. The real thing came out, and the card's N support has never worked. Fortunately there's a somewhat-deeply-buried setting for both Linux and Windows that lets you turn it off entirely and have the card degrade into a perfectly-functional b/g card, because once real N networks came out it tries really hard to connect to them. And it almost works. You can get about a megabyte or so over the N network before the connection somehow fails. Then another couple hundred kilobytes about 10 seconds later, and so on. Terrible.
How much more concerned about that I would be if I intended to ship a few million such units out. The "fixed drivers in the future" promise isn't that reliable.
Hitting that power and weight envelope is a hell of a lot easier when you make your own CPUs.
Other complaints include a loss of magsafe? I highly prefer being able to purchase numerous non-proprietary chargers for home, work, car, airplane rather than carrying a brick in my bag. Plus, anyone who wants the power connection to be breakaway can get whatever USB C equivalent there is or will be to ZNAPS[1].
[1] https://www.kickstarter.com/projects/1041610927/znaps-the-9-...
Sticking with the more powerful, but power hungry CPUs is one of the few things that allow the MacBook Pro to keep the "Pro".
Edit: the new CPUs use less power, but the new GPUs use... more?
Razer uses it to drive a 4K 100% Adobe RGB display. Seems appropriate enough to power the lower res Macbook 13" displays.