Core blimey! Apple macOS update lifts boot from MacBook Pro neck If your 2018 MacBook Pro laptop is slowing down unexpectedly, then download and install today's macOS 10.13.6 Supplemental Update, and cross your fingers. This patch attempts to address frustrating slowdowns experienced by owners of Apple's latest top-end 13in and 15in notebooks that use eighth-generation quad and six-core …
El Reg ran a story on it last week here
Alot of the comments were that Apple isn't immune to the laws of thermodynamics, reality distortion field or not, the thinner they make their notebooks the more difficult it is to cool a processor like an i9. Then again, why someone's doing rendering or video editing on a notebook and not a workstation is beyond me, but whatever.
Maybe you could point said “someone” to a performant _current_ MacOS desktop?
Sad to see Apple pretty much desert the dev/power user market, these latest MBPs lack in upgradeability and onboard storage. And at what’s getting to be pretty insane price points, way worse than 5-6 years ago.
Lots of devs use macs, but I wonder if that’s going to continue at this rate.
But they do look glitzy in ads.
From a traditionally mac dev shop, a lot of our devs are moving from macs to Thinkpads running Linux.
I have to say from my experience, the Lenovo Thinkpads don't hold a candle to the old IBM tanks. I went through 3 T420s, the fans eventually stopped spinning.
Still you can always use CloudReady and turn your £3k macbook into a Chromebook.
Quite understandable. We have a lot of mobile engineers running ProE and other CAD software on laptops - in our case Dell Precision and HP ZBook machines. We dont expect the same performance as a dedicated tower workstation, however a mobile workstation with Quarto or FirePro graphics should be able to handle moderate workloads without crashing and burning. Glad to see Apple has admitted the issue and pushed a fix, rather than just blaming it on their customers as usual...
Well, you can do video editing and rendering on a laptop, so people do. People could do it on slower computers for years. Sure, you probably don't want to produce your oscar nomination on one, but GPU loaded MBPs are used extensively by all sorts of people for video editing on the go, live visuals etc, there's lots of advantages to having a powerful portable computer that can handle real time video processing when you want to edit video, say, while travelling, don't have many people who want to stick a desktop in their carry on.
No, there was not a hardware design error, but a very stupid bug in the Apple firmware.
There was some user who reported on reddit the exact error made by Apple.
The Intel processors have a MSR (model-specific register) which must be intialized by the BIOS/EFI with the power limit for the processor. Due to some program bug, Apple initialized the"MSR_PLATFORM_POWER_LIMIT" register with the values correct for a desktop processor (with 95 W TDP) instead of the values correct for a laptop processor (with 45 W TDP).
As the reddit user verified, with values even slightly larger than the correct laptop values recommended by Intel, all works OK.
The Apple press release contains some mumbo-jumbo nonsense language like all such press releases.
In fact their updated firmware just replaces the wrong desktop value for that MSR with the correct laptop value.
This is (BTW) explained in the story – Apple used Intel's defaults in a hardware register controlling power usage. Those defaults are unsuitable for MBP laptops.
FWIW, the MSR is supposedly unchanged by the update: it's still on its desktop default. Apple appears to be loading extra software that controls power usage at a higher level.
C.
So you buy an expensive processor that will never be used at max power especially when you need it because the laptop case/motherboard can't keep it cool enough?
So what is important is just the specs on paper are cool enough to get people splash a lot of money on it?
A lot of people are confusing base clock and turbo clock with the maximum performance. The *base clock* is what the processor TDP is rated for. The max turbo boost clock cannot be used for sustained load - it would need the same additional cooling as an overclocked processor (i.e. running outside of manufacturer spec).
ALL processors with an unlocked multiplier *will* hit a point where it overheats for a normal cooling system. The design makes sure that the cooling system meets the tdp rating, not an arbritrary number.
It's like complaining that notebooks haven't got liquid nitrogen coolers.
It is definitely a bug going below base clock for sustained loads, it is not a bug that it does not stay at turbo for sustained loads. This is explicitly enabled by Intel with their cTDP feature amongst other things. Read the data sheet. That is why it is marketed by Intel as Turbo "Boost" i.e. temporary.
Also both the mobile i7 and i9 have the *same* TDP. So *Intel* (and not Apple/Dell etc) assures the system integrator that the same cooling/power system is enough.
I'm no Apple fan, but it makes no sense to have a reality distortion field that irrationally amplifies their negatives just because it's Apple. That's just as bad.
You got this completely wrong.
People (including the blogger reporting the problem originally) have re-run their tests with the software fix. And they all get the same picture: Instead of wildly varying clock speed and temperature, the clock speed know stays quite high, and the temperature stays quite high. It's very evident from the charts that the original software slowed down the CPUs much more than was needed, lowering the temperature much more than was needed. It's like driving in a 50mph speed limit, and when you spot a speeding camera you slow down from 55mph to 25mph. Instead the 50mph that a reasonable person would do. That's fixed.
The charts also show that with or without fix, the expensive processor runs at max speed. Just for a short time. And that's fine, because you are sold a "base speed" that the new software can sustain 24/7 and a "turbo speed" that both old and new software runs for maybe 25 seconds. Exactly what the specs on the paper say.
>> You got this completely wrong. People (including the blogger reporting the problem originally) have re-run their tests with the software fix.
>> It is definitely a bug going below base clock for sustained loads, it is not a bug that it does not stay at turbo for sustained loads.
So they fixed the bug. And the blogger gets my point.
Go to redditt or the comment I replied to above, who complain that the issue is that the cooling system and power regulation is not up to scratch. Because it isn't *staying* at the turbo clock that they paid for.
@diodesign
Apple appears to be loading extra software that controls power usage at a higher level.
I'd suspect AppleIntelCPUPowerManagement.kext (or a related kext) was left unsigned. This way the kext won't load (no power management?) and the Apple press release would make sense.
To test this, one should get an affected MBP and compare all kexts to see which ones get loaded pre- and post-update, in addition to turning on "kext-dev-mode=1" in NVRAM and disabling SIP to see if the (supposedly unsigned according to Apple) kext loads (provided it was a kext issue, but probably is, again, if we were to make any sense of Apple's press release).
This post has been deleted by its author
"you're holding it the wrong way"
At the time this happened I had a Nokia phone. I read the instruction manual, being curious. And guess what: It had instructions how to hold it. Because holding it wrong meant the reception wasn't anywhere near as good as you would like it to be. I bet lots of phones had similar problems.
"we invent rectangles with rounded corners, no one else may"
Nonsense. A clear demonstration that you don't understand the difference between. utility patents and design patents. Samsung has design patents as well, for a Galaxy kind of design, with Galaxy shaped rectangles with round corners, and they will sue you if you try copying it.
I'm wondering now how successful I'm going to be in shoehorning a 4.2 litre V6 into my Fiat 500. I hadn't counted on upgrading the radiator or the fan.
Perhaps I can fix it in software, as in
if(ownerPowerTestDetected()){
while(engineNotActuallySeized){
setAllowEngineOverheat(true);
} else {
setAllowEngineOverheat(false);
}
It could be done, but you would need to remove the rear seats, and probably install side cooling fins as per the old Clio V6 for good measure.
A lot of modern cars the grille is huge and provides more than adequate cooling, typically only a small amount is needed. Look at a pre-facelift early Ford Sierra V6 - it didn't even have a grille!
Similarly the Passat VR6 of the late 80s, which was grilleless to ape the old water cooled models.
I think you've just been doing donuts all over my little joke.
However, I don't think the fan holes on an Apple laptop are likely to be oversized, somehow.
Also, the current FIAT 500 is front wheel drive, and I think removing the front seats to shoehorn in a VR6 would bring novel problems. I don't think you would get it between the wheels of the original FIAT 500.
(I used to like watching the Abarth in Cambridge race Jaguars and the like from the lights though.)
It's not a problem at all.
The limiting factor for all CPUs is heat nowadays. The CPUs could go at much higher speed but they would melt. In a desktop, you can have tons of cooling and run at very high speed.
So what do you do with a laptop with less cooling? You don't want a slower chip because it's slower. Instead you use something almost identical to the desktop chip, and you get the benefit of the high clock speed either for short times, or for longer times with single or dual core usage.
You see the same effect actually with Diesel engines. The same engine provides 90hp in a normal car, 110hp in a car with better cooling, and 160hp as a boat engine where you have unlimited amounts of water for cooling.
The problem started with pentium 2nd gen. burn-outs (I suffered from two of them)
The work has since been out-sourced to GPUs, along with the heating problem.
As mentioned by many replies, the PC industry has forgotten a key basic engineering process: design, develop specs,, add a safety buffer, meet specs. Using spftware to cover hardware gaps is a cludge.
"Using spftware to cover hardware gaps is a cludge."
Why is it a cludge? It lowers costs, gives better bang for you buck, is flexible and is a non-critical application.
Meltdown, spectre, this all were mitigated because HW is now more SW programmable. Saved some landfill too.
>>design, develop specs,, add a safety buffer, meet specs
A spec does not maketh a perfect design.
When Apple stopped selling the 17 inch MBP, the last remaining ones were available in the UK as "refurbished" for about a year. (That's what Apple does if they can't shift old gear. They mark it as "refurbished"). Took a year to sell them all. So apparently there wasn't that much of a market.
The Register 
Biting the hand that feeds IT
Copyright. All rights reserved © 1998–2025
