Slashdot Mirror
Apple Confirms MacBook Pro Thermal Throttling, Issues Software Fix (theverge.com)
An anonymous reader shares a report: For a week, we have been seeing reports that the newly released MacBook Pros run hot, which all kicked off after this video by Dave Lee. They run so hot, in fact, that the very fancy 8th Gen Intel Core processors inside them were throttled down to below their base speed. Apple has acknowledged that thermal throttling is a real issue caused by a software bug, and it's issuing a software update today that is designed to address it.
The company also apologized, writing, "We apologize to any customer who has experienced less than optimal performance on their new systems." Apple claims that it discovered the issue after further testing in the wake of Lee's video, which showed results that Apple hasn't seen in its own testing. In a call with The Verge, representatives said that the throttling was only exhibited under fairly specific, highly intense workloads, which is why the company didn't catch the bug before release. The bug affects every new generation of the MacBook Pro, including both the 13-inch and 15-inch sizes and all of the Intel processor configurations. It does not affect previous generations.
The company also apologized, writing, "We apologize to any customer who has experienced less than optimal performance on their new systems." Apple claims that it discovered the issue after further testing in the wake of Lee's video, which showed results that Apple hasn't seen in its own testing. In a call with The Verge, representatives said that the throttling was only exhibited under fairly specific, highly intense workloads, which is why the company didn't catch the bug before release. The bug affects every new generation of the MacBook Pro, including both the 13-inch and 15-inch sizes and all of the Intel processor configurations. It does not affect previous generations.
It's amazing how remorseful companies are when they get caught doing something silly :|
Here's a thought:
Fix it before you release it to the public and you won't have to apologize and tarnish your reputation.
Why would it be less? Apple's reported battery life is with the CPU at full performance, not throttled down - remember they said they had not seen that case. So it means Apple's battery life figures (usually very realistic) are based on the CPU operating at normal speed.
"There is more worth loving than we have strength to love." - Brian Jay Stanley
QUOTE: "representatives said that the throttling was only exhibited under fairly specific, highly intense workloads" Sure, exporting video from Adobe Premiere Pro. Clearly an unusual workload.
They could have been leaving the CPU at boost speed for too long. Then it has too really slow while the system dumps heat.
John McAfee 'It was like that time I hired that Bangkok prostitute; to do my taxes, while I fucked my accountant'
we’ve identified that there is a missing digital key in the firmware that impacts the thermal management system
Are they just making stuff up now?
Unless you hold the key, there is no way to know now is there....
"File to fit, pound to insert, paint to match" - Aircraft Maintenance 101
I disagree... The price sure is "pro" level.
"File to fit, pound to insert, paint to match" - Aircraft Maintenance 101
Except, the CPU was not overheating. The throttling did not happen because of the CPU
The CPU is powered by a Voltage Regulator Module (power delivery module). It turns out it is this chip that is overheating - when the CPU is going full tilt, it's demanding 125W from the VRM. This causes the VRMs to heat up and when they get close to their maximum thermal limits, they send a signal to the motherboard telling it to throttle the CPUs so they draw less power so the VRMs can cool down.
Part of thermal tuning is to adjust the CPU boost speeds such that it can boost to full speed, then throttle down slightly as the VRM and CPU heat up to a new max steady-state condition where the heat generated can be dissipated.
This is in part due to a documentation error in Intel's docs regarding max thermal power dissipation values.
The good news is if you tweak the throttle settings properly, you can keep the regulators from overheating, but the CPU still performing. This is what Apple did - they optimized the settings so the VRMs will not overheat and force a sudden throttling of the system. Doing this gives you a good 20% speed boost over the old models.
The bad news is if this was caught earlier so Apple could heatsink the VRMs to the CPU like they do with the GPU, you could get up to another 10-20% in performance because you can run the boosts longer since the VRMs would heat up slower.
I wonder how they can fix this issue without either lowering the performance or allowing the VRM to overheat thus killing them faster.
- Raynet --> .
They tested watching cat videos on youtube, what else?
User: The computer is slowing down when it gets too hot
Engineer: Yes
User: You sold us something that is not performing as expecting
Engineer: We told marketing about this, too. This was not a good idea.
User: We want full power.
Engineer: Okay, enjoy your third degree burns then.
Marketing: Let's call it a "software fix". Fucking engineering at it again ruining our reputation
User: Everything is melting but I gained 20 seconds on a video render! Yay thanks Apple!
Apple didn't think any of their users would put a heavy workload on their computer, what does that tell you?
That "Pro" is just a dickwaving nameplate, not an indicator of intended market.
So a software update will make it not a "form over function" overly-thin disaster with an insufficient heat pipe and fan? That'd literally be magic. What I assume this does is just let it run insanely hot without down-clocking it. Wow, what a great fix. I wonder why they implemented the throttling in the first place? Also, Apple's quite implied they never ran a stress test of any kind on it? Not even a simple prime number calculation for 10 minutes? REALLY? Maybe they took one look at the heatsink and just didn't bother.
Thank you for providing a detailed explanation. I appreciate being educated instead of berated.
"A person is smart. People are dumb, panicky dangerous animals and you know it." - K
What was the chain of missteps that enabled what can only be called an engineering fiasco?
* AMD's high performing 8 core Ryzen's selling like hotcakes, Intel gets envy.
* i9 rushed to market
* The reason Intel didn't release an 8 core desktop processor in the first place: sucky thermal envelope, and no time to fix the process
* Marketing solution: lie about the power envelope in the spec sheet
* Apple getting shade for obsolete laptop lineup just as PC market showing signs of life
* Apple rushes Macbook refresh to market
* Apple engineers believe Intel's specs, design the VRM too small.
* Apple project manager telescopes the project by cutting out the QA stage
* First production units show significant issues
* QA fudges the problem resolutions for team player reasons
* Product already shipped to retail, just got to cross fingers and hope for the best
* Unhappy customers are not fooled
* Oops.
See, it starts with Intel's well known process issues then envy and hubris take it from there. Apple should pull this product and re-engineer it. Dump the remaining stock for a discount as befits the low performance. Apologize for the mistake. Next one should be AMD.
Apple isn't going to do any of that. "You're holding it wrong."
When all you have is a hammer, every problem starts to look like a thumb.
The VRM bit is an important part of the story, but it's not the only important part. We also learn that i9 does not operate inside its published thermal envelope, otherwise no competent engineer would have wiffed the thermal design so badly. Intel lied, Apple didn't verify, marketing jumped the gun on both sides. It's a circus.
When all you have is a hammer, every problem starts to look like a thumb.
Apple is trying to get their own chip to market so that they don't have to give money to intel or AMD. Good luck to them, though. The number of architectures has been shrinking for a reason, and that reason is that amd64 has won. The most competition produced the best products and now there's no reason for anything else to exist but amd64 at the high end, and arm at the bottom. And so sad for intel, they have no arms. They had strongarm->xscale, but it didn't xscale down in power consumption so it got its ass whipped by arm implementations which did and now intel has nothing but some insecure-by-design antiques that it has to rehash.
But maybe I'm wrong and apple will manage to make an arm that can compete with amd64 on highly parallel workloads. Most creative tasks are highly parallelizable, which was how Apple managed to retain their professional user base during the powerpc era. The primary processing unit in the G4 is limp even by the standards of the day, but there was a fast vector coprocessor in there, and content creation apps could be accelerated by it. It's therefore not impossible, only unlikely.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
Apple is trying to get their own chip to market so that they don't have to give money to intel or AMD. Good luck to them, though. The number of architectures has been shrinking for a reason, and that reason is that amd64 has won.
X86 has won the battle but most probably not the war. Nothing technical stops ARM from invading X86 space at increasingly high performance points. When X86 tries to invade the low power space, its complex instruction set bites hard because that big chunk of die that decodes it (don't understate this, it is a good chunk of any modern x86 chip) eats battery power and takes up space that could be used for more cores or more integrated system components. In the long run ARM is going to catch up with X86 in single core throughput and clobber it in cores per dollar, thus winning the war. X86 will be relegated to a sliver or the market in a legacy role. ARM is already roughly tie with X86 in total ops/sec already shipped and ahead by an order of magnitude in units shipped.
The most competition produced the best products and now there's no reason for anything else to exist but amd64 at the high end and arm at the bottom.
Mostly agree, but IBM shows no signs of giving up on Power arch and System Z for its big iron. Also, the high high end segment keeps getting smaller and I think we will pretty quickly hit a situation where ARM and X86 are sharing it, and only legacy effects keep ARM out of the Windows sector. Most probably, my next laptop will be ARM based, maybe a repurposed Chromebook running Linux.
And so sad for intel, they have no arms. They had strongarm->xscale, but it didn't xscale down in power consumption so it got its ass whipped by arm implementations which did and now intel has nothing but some insecure-by-design antiques that it has to rehash.
But maybe I'm wrong and apple will manage to make an arm that can compete with amd64 on highly parallel workloads. Most creative tasks are highly parallelizable, which was how Apple managed to retain their professional user base during the powerpc era. The primary processing unit in the G4 is limp even by the standards of the day, but there was a fast vector coprocessor in there, and content creation apps could be accelerated by it. It's therefore not impossible, only unlikely.
Intel's sun is starting to set, I really can't think what they should do about it. But it's setting slowly, there are still many billions to be had out of the old Wintel cartel. Intel won't get into ARMs simply because they dread cannibalizing their own x86 market. It is just not possible to command the same margin for an ARM part as Intel has become accustomed to in its monopoly segment.
When all you have is a hammer, every problem starts to look like a thumb.
Man, who would have thought cramming two more cores into a system that already was thermally insufficient would case heat and throttling issues?
I never really believe Jobs was the heart of the company but considering the trend-chasing that Tim Cook has led apple into and the garbage hardware pumped out because no one is around to tell Ives when he has stupid ideas I'm beginning to think there may have been something to that.
What does it say? When a customer more brutally stress tests your brand-new product than your own V&V department, it says that the fruit of your labors isn't the highest quality produce, best calibre nor the embodiment you claim.
Its bad apples. It's bruised fruit on sale which AAPL should send packing and sell the next fruit off the tree. To hell with a Bandaid® patch. Its still bad fruit. Cutting out the part that's bad with a patch around and over the bad spot still leaves a bad taste!
SteveJobs would fire V&V, re-engineer and sell fresh apples with no apologies and free upgrades for those who got bad apples
Hi, could the VRM theoretically also still make the graphics card overheat or throttle?
Apologies for being anonymous, I did this post without having an account.
Well obviously you need cryptographic protection involved when addressing hardware, otherwise everyone would be making drivers for it. You don't want people to go & run linux on it and have actual good performance.
"Intel won't get into ARMs simply because they dread cannibalizing their own x86 market."
Perhaps there is a simpler explanation: incompetence. We've already seen that even with a process advantage, they can only outperform AMD by ignoring security. Perhaps they simply can't make a decent ARM.
"You're right," Fisheye says. "I should have set it on 'whip' or 'chop.'"
I have many words to describe Intel but incompetent is not one of them. I guess they could make a nice ARM chip but they are afraid to do it. To maintain their share price they need both a huge share of the market and a gross profit in the range of 60% as they still command in their monopoly Wintel business. Looking at the trend for Qualcomm isn't encouraging, their gross margin is trending down, currently around 55% and likely to go lower in the face of robust competition from the likes of Samsung. And it can get worse fast... what happens when some monster Chinese manufacturer introduces its own ARM offering, or Amazon? It's going to happen.
AMD could move (back) into the ARM business for the counterintuitive reason that they aren't encumbered by massive profits and therefore not as beholden to shareholders, but there is this weird dynamic where Intel holds the price of X86 parts artificially high for both of them, which tips the balance in favour of staying right where they are.
When all you have is a hammer, every problem starts to look like a thumb.