iFixit:

The 2018 MacBook Pro keyboard is a wealth of secrets—it just keeps surprising us. Just when we think we’ve exhausted one vein of tasty tech ore, we find something new. And today, we bring this trove to you.

And:

We pumped this keyboard full of particulates to test our ingress-proofing theory. We started with a fine, powdered paint additive to add a bit of color and enable finer tracking (thanks for the tip, Dan!). Lo and behold, the dust is safely sequestered at the edges of the membrane, leaving the mechanism fairly sheltered. The holes in the membrane allow the keycap clips to pass through, but are covered by the cap itself, blocking dust ingress.

And:

On the 2018 keyboard, with the addition of more particulate and some aggressive typing, the dust eventually penetrates under the sheltered clips, and gets on top of the switch—so the ingress-proofing isn’t foolproof just yet.

I do appreciate the testing, good to know the threshold at work here. Bottom line, don’t dump a bunch of powder or sand on your keyboard and you should be fine.

If you want a more traditional, picture-laden keyboard teardown, iFixit has that too.

And on a related note, here’s John Gruber’s take on the legal/marketing side of the new keyboard, and the quieter vs. better at keeping out debris language.

Joe Rossignol, MacRumors:

Some customers have been hoping that Apple will start swapping out second-generation keyboards with third-generation keyboards, as part of its service program, but MacRumors has learned that isn’t the plan.

When asked if Apple Stores and Apple Authorized Service Providers will be permitted to replace second-generation keyboards on 2016 and 2017 MacBook Pro models with the new third-generation keyboards, if necessary, Apple said, no, the third-generation keyboards are exclusive to the 2018 MacBook Pro.

This isn’t terribly surprising, as the architecture of the new machines has changed and the swap-out goes way beyond simply swapping keyboards. But it is good to know.

[VIDEO] A new video is making its way around the net, under the title “MacBook Pro 15 (2018) – Beware the Core i9”. The video (embedded in the main Loop post), is a reasonably measured analysis of one specific new MacBook Pro model, the highest end, spec’ed with a 2.9GHz 6-core Intel Core i9 processor.

Before we get into the video at all, the issues Dave Lee raises are specific to this configuration. I’ve seen not seen anything to make me believe the over-throttling Dave encountered occurs on lower-spec’ed models. Per usual, ping me if I’ve missed anything, or if you see someone encountering this issue with, say, a 2.6GHz 6-core i7.

On to specifics:

Dave runs an Adobe Premiere render on Mac and Windows, the Mac using the i9, and the Windows machine using an i7. Under high load:

• The Windows laptop (Intel i7) runs at an average clock speed of about 3.1GHz, temp of ~87°C
• The MacBook Pro (high end i9) runs at an average clock speed of 2.2GHz, temp of ~90°C

In this specific case, with this specific configuration, with this specific i9 chip, the MacBook Pro runs hotter and slower under intense load.

Dave then sticks his MacBook Pro in the freezer and repeats the experiment, and the thermal throttling is significantly reduced, as the Mac no longer has to throttle performance to keep the machine from overheating.

I’d be very interested in seeing this experiment repeated by other folks. Thermal throttling is not the villain here. It’s about the ability of the Mac itself to dissipate heat efficiently. Once the chip heats up, that’s when thermal throttling kicks in.

Watch the video, draw your own conclusions.

John Poole, Primate Labs:

Apple announced updated 13-inch and 15-inch MacBook Pros last week. Let’s take a quick look at the performance of these new laptops using Geekbench 4 results from the Geekbench Browser.

For those unfamiliar with Geekbench 4, it is our cross-platform CPU and GPU benchmark. Higher scores are better, with double the score indicating double the performance.

Note that Geekbench scores tend to improve over time, as startup tasks like iCloud syncing tend to eat CPU time when a machine is first configured. Once those “one time” tasks are completed, usually after a few days, they no longer skew the results.

Benchmarks like these offer a nice way of comparing apples to apples when you are considering a move from one machine to another. If your workload includes very specific, repetitive, high-intensity tasks (such as audio or video rendering, for example), you might want to seek out benchmarks comparing results for those specific tasks.