Apple finds the key to Core i9 MacBook Pro performance problems

Much attention paid to Apple's latest MacBook Pros has been focused on the newly-silicone-ensconced keyboards in those notebooks, but another controversy has erupted regarding the new machines' performance under sustained heavy workloads.

As The Verge points out, YouTuber Dave Lee found that a Core i9-equipped MacBook Pro couldn't sustain even its base CPU clock while performing  Adobe Premiere Pro rendering, leading to widespread stress-testing by the community at large. The mobile experts at Notebookcheck noted similarly disappointing behavior with repeated Cinebench runs, for just one example.

Turns out, something was wrong—it just wasn't related to the notebook's cooling capabilities. Apple didn't issue a general press release on the subject, but it did contact outlets with review samples to inform them that a software issue had led to improper thermal management of the processor under certain workloads.

According to the statement given to The Verge, Apple claims “a missing digital key in the firmware” was to blame for the problem, and it affects all Coffee Lake MacBook Pros, not just the Core i9 option. While “digital key” is not a term I've ever heard in my years of PC hardware coverage, the company issued a patch in macOS High Sierra 10.13.6 to correct the issue.

Dave Lee tested post-patch Premiere performance with his Core i9-equipped MacBook Pro and found that it was leaps and bounds better than a similar machine equipped with a quad-core Core i7 CPU—as we would expect. All's well that ends well, I suppose.

Comments closed
    • HERETIC
    • 1 year ago

    Wonder if this is a real fix, or a apple fix (If we tell them enough times-it’s fixed-they’ll believe.)

    My best guess is they’ll stop throttling below base speed, and forget turbo for anything other than very bursty workloads.
    These CPU’s are hard enough to cool in desktops at those speeds-a cheese-cutter chassis has little chance.

    RUMOR-apple has already hand-balled the fault at Intel. “When we designed this chassis,we were promised we would have 10nm CPU’s.”
    Ha ha-everyone in the industry with half a brain (except chuck)were aware 10nm was dead.

    • riviera74
    • 1 year ago

    Almost all other laptops with the new Core i9 processor have extra cooling so that the CPU need not throttle down because of overheating. Apple’s problem (it seems to me) is that they did not want to pay the extra $$$$ to address this issue correctly.

    Now they have some software issue that addresses the i9 and all i7 throttling issues too. Nice, but Apple could have simply put in a better cooling solution to ensure all Coffee Lake CPUs run more smoothly. Then again, maybe this will actually work out OK.

      • tipoo
      • 1 year ago

      [url<]https://i.imgur.com/NrR12yM.jpg[/url<] Laptops that run the i9 faster are proportionally thicker and louder. Now, you could say Apple should build something like that, but it now at least sticks to its base clock speed, and that's technically passable for what should be the assured minimum. That they chose silence and size is the Apple-ey thing to do, but it's at least not as egregious as breaching the base clocks now.

    • Krogoth
    • 1 year ago

    Putting DTR-tier hardware in an ultra-thin chassis and expecting nothing to go wrong.

      • K-L-Waster
      • 1 year ago

      But the Apple magic makes it inconceivable that anything could go wrong!!

      [url<]https://www.youtube.com/watch?v=YIP6EwqMEoE[/url<]

    • barich
    • 1 year ago

    Well, this puts an end to the absurd suggestion that the CPU was idling while under the heavy load of non-GPU-accelerated encoding. People look for the most ridiculous excuses to let Apple off the hook for obvious problems.

      • tipoo
      • 1 year ago

      Yeah, they saw Pooles idling idea and clung to it with no evidence for it (not blaming him himself – an idea is just an idea, and he wanted to test more)

      “Here, it’s dipping to 800MHz under full load in 5 different workloads, not hitting Quicksync or anything”
      “Lol no, it’s idling, the CPU temp isn’t even critical”

      Was too common a discussion. Some would be outright hostile to anyone suggesting VRM related throttling.

      This is why we wait and learn more. Like I said in another post, the gaslighting of legitimate issues is my least favorite thing about [some parts of] the Apple community. Denying issues gets us nowhere, calling them out and improving things does.

    • TwistedKestrel
    • 1 year ago

    Would be very interested to see some thermal testing/results of Apple’s fix

      • tipoo
      • 1 year ago

      From what I see, it still sticks to 100C under load, that’s just the thermal capacity they designed. But it now sticks to base clock, in exchange for not trying to turbo so high and so long, like some of the manual fixes were doing.

      The i9 just ate the MBPs thermal capacity in seconds and forced it to dive lower before, by managing this better they can at least stick it to base clocks, though not turbo for any sustained time.

    • DragonDaddyBear
    • 1 year ago

    Anyone have an idea on what exactly this key was decryption that kept the speeds from being where they should have been?

      • derFunkenstein
      • 1 year ago

      there’s a bunch of stuff that relies on a board ID in the various power management kexts. If the board ID is missing, it’ll use a generic full-throttle desktop profile (which is what happened, see NTMBK’s comment). My guess is that a board ID was missing from AppleIntelCPUPowerManagement.kext. Not confirmed, but that’s the kind of thing that could be distributed in an OS update.

        • derFunkenstein
        • 1 year ago

        Or maybe not. The download is 1.2GB. There has to be more to it than that.

        [url<]https://support.apple.com/kb/DL1973?locale=en_US[/url<]

          • derFunkenstein
          • 1 year ago

          Alright, so I downloaded the package and extracted it. The sizes below are from macOS, which reports base-10 values. On any other platform, the numbers would be smaller. This tells me nothing, I guess:

          [url<]https://imgur.com/a/hlGRmp8[/url<] - EmbeddedOSFirmware is 500MB, but that's because it has its own recovery partition bundled with it. - FirmwareUpdate is 200MB. None of the files are executable. Probably applied by the recovery partition. - FullBundleUpdate and macOSbrain are both pretty small, each with a file called "scripts" but it's either some sort of binary or encrypted. I can't read it. - macOSUpd10.13.6Supplemental is a 430MB package that updates the OS - macOSUpd10.13.6SupplementalRecoverHDUpdate is just a few KB with some sort of unreadable script - SecureBoot.pkg probably updates the SecureBoot on the device - The Resources folder has a bunch of language localization files

          • tipoo
          • 1 year ago

          Note that the key was in addition to the regular supplemental update, it wasn’t the update in its entirety.

            • derFunkenstein
            • 1 year ago

            Right. I have a feeling that it’s enough to make it work on the restore partition, though, since the restore partition updater is all of 12kb.

            The supplemental update is strictly for this model, though. The package I downloaded does not apply to my 2017 MBP, and the installer says so when it runs. So the entirety of the supplemental update is only for the 2018 model.

    • tipoo
    • 1 year ago

    Funny how there were people denying the throttling dips were that bad (“it was just idling!”), but they appeared VRM related in some cases even when the CPU wasn’t at critical temp.

    I generally like Apple stuff, it’s the gaslighting of real issues from parts of the community right up until Apple admits something is wrong that I dislike. They’re a company, with human people running it, and not everything is precisely calculated and perfect all the time, quite occasionally there are flaws. Calling those flaws out gets us improved stuff, denying them gets us toxicity and worse products.

    Anyways, I wonder if the “digital key” manages the boost max power draw and time like some redditor tweaks found helped. The other half is it ramps up the fan at 70C now rather than waiting.

      • blastdoor
      • 1 year ago

      Well said — I totally agree.

      edit — to expand just a bit.

      The gaslighting is real and it is very damaging. It’s closely tied to an inability to accept any criticism as genuine. The gaslighters assume all criticism is disingenuous and that it’s impossible to like Apple products and criticize them simultaneously. This mentality exists in many spheres of life — political, religious, etc. It’s a great example of why our species is not ready for immortality.

        • Voldenuit
        • 1 year ago

        >It’s a great example of why our species is not ready for immortality.

        And that, folks, is why Kosh blew up that shuttle.

        • tipoo
        • 1 year ago

        “Death is very likely the single best invention of life. It is life’s change agent. It clears out the old to make way for the new. Right now the new is you, but someday not too long from now, you will gradually become the old and be cleared away. Sorry to be so dramatic, but it is quite true. ”

        -Steve himself put it beautifully

          • Oem
          • 1 year ago

          But ‘life’ is an abstract concept, and what’s more important, perpetuating an unfeeling, unthinking, essentially imaginary thing, or actual thinking, feeling people? If life weren’t so clearly the product of unthinking physical processes, it would make the ultimate supervillain: willing to sacrifice the lives of trillions of beings (through programmed death) to further its own ends. But since there is no actual mind behind this ongoing tragedy, death is entirely pointless.

          In an age making rapid advances in biotechnology and understanding and manipulating the human genome, it is no longer just a thought experiment to consider whether to be for or against extending the human lifespan. Real breakthroughs are likely in our lifetimes. Real politicians, companies and universities have impactful decisions to make to pursue research or not, and they are influenced in turn by the public, by us.

          What makes people deserve to not die? (to ‘earn’ the right to immortality) First, that implies that those alive now are all deserving of the death sentence. Extermination of all that are living is acceptable until proven otherwise. Seven and a half billion living beings, convicted of a crime against some morality that purports to know how people ought to be (though no actual society has ever lived up to this hypothetical utopian standard).

          Secondly, who or what are we trying to impress? In an unthinking and purposeless universe, there is no one to disapprove of us achieving immortality outside those achieving it. Unless there is some moral absolute by which we can all be judged. But if there was, that also would be an inanimate, unthinking abstract concept. If we sacrifice our lives for the benefit of a concept, will the concept feel gratitude? If that were even possible, the already dead would hardly be in a position to appreciate it.

            • trackerben
            • 1 year ago

            Properly, you’re doing infinitely too much thinking and purposing given an unthinking and purposeless existence on your part. That irrationality undercuts any kind of reason you can present, if such a cosmological model was somehow the case.

            • Oem
            • 1 year ago

            Ha! Points rewarded for the wordplay, points deducted for (deliberately?) misinterpreting ‘unthinking universe’ — which here was meant (I thought clearly) to refer to a universe that arose naturally as opposed to being created deliberately by a thinking being. The latter could conceivably have a purpose or plan for you to follow; the former provides no built-in point to existence. That doesn’t mean existence itself is purposeless — and how can it be unthinking if even one sentient being exists in the universe?

            IMO a purpose emerges as a matter of course from our biology and genetics (our natural mental architecture).

            These points are interesting but overlarge and getting a bit off track. Really to boil down the original motivation of the previous post, it’s an objection to casual misanthropy. “The human race is clearly immoral / evil / undeserving” — because it fails in the judgement of ideology XYZ — is an all too common view, and that’s sad. And one that falls apart on close inspection. Trivially since all you have to do is ask, “OK, so what makes XYZ correct? Prove that’s truth and not just a bag of ill informed opinions and / or ill founded traditions and / or faulty assumptions & logic”. You can’t, of course — we’re divided into opinionated groups yelling at each other without making any real progress. So take your (not you personally of course) misanthropy and file it with all your other unfounded ideological opinions. Spoken in a public forum as if it is a well known fact (it certainly is a popular one) is not that different from spouting a particular political view as if the majority, or the right-thinking minority you belong to, knows is the truth. One should be prepared to face objections from those who actually consider it just as wrong as you think it is right.

      • K-L-Waster
      • 1 year ago

      [quote<]it's the gaslighting of real issues from parts of the community right up until Apple admits something is wrong that I dislike.[/quote<] This. Before fix: "Apple is perfect, you're stupid if you think otherwise, if anything on your system isn't working it's your fault for doing it wrong, fake news." After fix: "Look how awesome Apple is, they fixed everything perfectly!"

    • Chrispy_
    • 1 year ago

    Now that they have fixed #Throttlegate, how much slower is the i9 in the MBP than, say, an Alienware R4 15″ which has seemingly [url=https://img2.parts-people.com/products/TGV49a1.JPG<]double[/url<] the cooling the solution of a [url=https://c.slashgear.com/wp-content/uploads/2018/07/macbook-pro-2018-teardown.jpg<]2018 MBP[/url<] but also allows the intakes to [url=https://www.notebookcheck.net/fileadmin/Notebooks/Alienware/15_R3/15R3_12.jpg<]actually breathe [/url<]rather than forcing it to suck in fresh air through some razor-thin slits at the edges.

      • tipoo
      • 1 year ago

      [url<]https://i.imgur.com/NrR12yM.jpg[/url<] The MBP now appears to stick to base at least in most loads, everything proportionally larger gets further into boost territory

        • Chrispy_
        • 1 year ago

        Oh wow, Apple are leaving 40% of the potential i9 performance on the table? That sucks.

        I didn’t realise that you only get a Radeon 560 with the Apple, too. I guess you could game on it but any CAD packages or OpenCL applications aren’t going to perform like anyone would reasonably expect on a $4000 laptop. The cheaper competition is rocking GTX1080 MaxQ, mostly.

          • End User
          • 1 year ago

          My T480 was $4,000 (CDN) and all I got was a lousy MX150 😛

            • tipoo
            • 1 year ago

            That must be nearly half SSD cost then? They start at a grand and 2K is pretty pimpin’ on them, 2700 even gets you a 1TB SSD

            [url<]https://www.lenovo.com/ca/en/laptops/thinkpad/thinkpad-t-series/ThinkPad-T480/p/22TP2TT4800[/url<]

          • tipoo
          • 1 year ago

          It’s a pretty reasonable OpenCL card at 35 watts, the only issue is they don’t provide anything larger that can house higher wattage cards. The 1080 destroys it, heck the 1060 does, but both in proportionally higher wattage and larger laptops. Polaris when dialed back to these wattages is actually pretty comparable with Nvidia cards on perf/watt, unlike on desktop and higher wattage mobile.

          Hence the calls for a Macbook Pro-Pro that’s thicker and can house the chips some want. Even then, Nvidia seems a no go for Apple.

          • Bauxite
          • 1 year ago

          Is anyone actually surprised by this fashion company that makes computers anymore? Everything they make is thermally constricted and starved as if they are above the laws of thermodynamics. Thin bodies and smooth shells are more important than anything to them.

          Its almost like they are doing a long deconstruction or subtle troll of the concept of “Form over Function”.

          Their flagship device, a [i<]handheld[/i<] phone, is literally too thin for hands to the point basically everyone gets a case for it. Ergonomics is blasphemy to them, and all the me-too asia oems that joined them. Meanwhile every other handheld tool in the world has evolved a useful form, starting from when the first pre-human picked up his favorite sharp rock.

            • MathMan
            • 1 year ago

            What does it say about the tech companies when a fashion company has 2 operating systems and a chip design division that is head and shoulders above their offerings?

            • tipoo
            • 1 year ago

            We’re really letting Michael Kors off easy!

            • moose17145
            • 1 year ago

            Think Different indeed.

          • MathMan
          • 1 year ago

          Who cares that the equivalent Windows laptop is faster?

          The fact that you have to run Windows instead of OSX is a much bigger negative.

          • JustAnEngineer
          • 1 year ago

          Have you looked at the innovative intakes on the Asus [url=https://www.asus.com/us/ROG-Republic-Of-Gamers/ROG-Zephyrus-M-GM501/<] RoG Zephyrus M GM501GS[/url<]?

    • FireGryphon
    • 1 year ago

    This was originally included as a feature. The CPU has extra time to finish processing because you’d be tilting your laptop and blowing compressed air into the keyboard, then driving to an Apple store, it didn’t have to finish quickly.

      • sreams
      • 1 year ago

      Correction:

      They expected the compressed air used to clean the keyboard would also cool the CPU.

    • K-L-Waster
    • 1 year ago

    Yet another exhibit for “don’t buy on launch day, wait for the first few patches.”

      • blastdoor
      • 1 year ago

      I thought it was an exhibit for “buy on launch day, discover stupid bug that should have been fixed in QA, post to YouTube , become famous “

        • K-L-Waster
        • 1 year ago

        Those groups do have some venn-diagram overlap…

        • Redocbew
        • 1 year ago

        I thought it was “buy on launch day, gush profusely while removing from box, post to youtube, become famous. Complain about stupid bug later.”.

          • Pancake
          • 1 year ago

          That’s not quite how it works. By now it’s “THERE NEVER WAS A BUG! IT DID NOT EXIST! IT NEVER EXISTED! WHY DO YOU KEEP LYING!”

    • NTMBK
    • 1 year ago

    According to some reddit sleuthing (https://www.reddit.com/r/macbookpro/comments/91256u/optimal_cpu_tuning_settings_for_i9_mbp_to_stop/), the problem was that the CPU power management was set to allow up to 100W(!) of continuous power draw. This meant that the VRMs would overheat, and then the CPU would throttle down to 800MHz until the VRMs cooled down enough.

    Clearly there was some screwup in the firmware, which meant that desktop power management settings were being applied to the laptop.

      • DPete27
      • 1 year ago

      I know the i9 MBP has only been out for 2 weeks, but do you suppose there’s any threat that the VRM could be weakened/damaged for someone who would’ve taken delivery of one on launch day and (presumably) needed to use that machine to it’s full potential for the past 9-13 working days? Meaning it’s been overheating/overloading its VRMs multiple times per day for that period of time.

        • techguy
        • 1 year ago

        We would need the datasheet from the VRM manufacturer to determine the VRM’s capabilities. I was hoping ifixit would have high res pictures in their teardown but it appears they may only have posted a short Youtube video rather than a thorough written summary with pics.

        We can’t make any definitive statements with certainty without this data. Generally speaking though, throttling behavior is designed to prevent components from experiencing conditions which will cause them to degrade. If we assume these mechanisms were functioning properly, there should be no long term effects or damage to sensitive components.

    • Dudeface
    • 1 year ago

    How does an issue like this make it through into a shipping product? Did Apple do literally zero QA?

      • jihadjoe
      • 1 year ago

      Seems like they did the QA *before* the product “went gold”, if we were to speak in software terms.

      • NTMBK
      • 1 year ago

      Apparently they performed the thermal tests before this last minute firmware update.

        • blastdoor
        • 1 year ago

        I bet they don’t do that again!

      • Chrispy_
      • 1 year ago

      Yes.

      • Pancake
      • 1 year ago

      “But it worked fine running Windows!”

      • tipoo
      • 1 year ago

      Especially with that much ballyhooed professional workflow team. A long CPU intensive render finding the VRM throttling dips wasn’t a thing they found before launch, and it was up to youtubers?

      Weird, but at least they were fast this time.

      • chuckula
      • 1 year ago

      Don’t be ridiculous they did plenty of QA!

      The problem is they did the QA at their new mothership headquarters, and the building’s RDF generators meant the problem didn’t show up there.

Pin It on Pinterest

Share This