Throttling in New i9 MacBook Pros Appears to be Due to Power Delivery Chip

[CaptOblivious]

We live in the real world and it's not always perfect? Why are you so worried? It's not like Apple doesn't provide software updates and launch repair programs.

"Oh, they shouldn't have shipped it like that in the first place. Their quality control is crap. Who is testing this stuff?"

Give me a break. Apple ships over 300M products per year and do a damn good job. Stuff happens. Apple fixes it. End of story.

You also have no idea what is actually wrong from a technical perspective, what Apple knew, what capacity is Intel at fault, and why the decision was made. Maybe they made a mistake? Maybe Intel made a mistake? You people can't think Apple is going to ship junk on purpose if it risks their brand. That is INSANE and they would not do it.

I just wanted to say that its a moot point how many products Apple ships in a year. This is a problem in ALL MacBooks, not an issue resulting of volume. Obviously, during prototype phase, someone either did not see this problem OR decided it can be fixed after release.

If they did NOT catch it, I don't know how you can argue that its not their fault. Yes mistakes happen but that is literally their job to identify problems before the device goes into mass production. So obviously someone at Apple failed at their job.

IF they decided it IS a problem but will be fixed in future firmware updates then, we can argue they did a disservice to their customers by releasing a product before it was ready. Someone might have done a cost/benefit analysis to identify if potential perception problem this will create is worth the monies they would lose by delaying mass production. But yes, in this case, we don't know what really happened. But your point that Apple would not ship something knowing is broken is insane is not true. It all depends on their financial contracts and what delaying mass production would mean.

Just my 2c.

 

[donster28]

What other issues were there? This is the only one I knew of.

I guess he meant all the re-desgined MacBooks Pros from 2016 that included the Touch Bar versions.

 

[Baymowe335]

Right, because greedy corps like Apple make it to the top being only transparent, so much stock price is at it's highest and iconic Mac product line at it's worst. You are funny, please keep embarrassing yourself.

You've proven the point. You're overreacting. Macs sales will be excellent because they are great laptops for people who don't think the entire product line is flawed because of this brand new issue one 1 model that will be fixed.
[doublepost=1532446346][/doublepost]

Because they released a defective product with or without testing and you are still defending them?

The game isn't over. I'm not defending anything...I'm basically saying to stop overreacting and let the process play out because we've seen Apple fix things before.

 

[LordVic]

Seems a bit like the iPhone throttling.

it's a similar concept.

All CPU's these days have built in thermal protection to try and avoid them overheating and burning to death in a smokey halo (I torched a few CPU's back in the day before they started doing this, even 2nd degree burns on my fingers from one)

Whether it's ARM, X86, PPC, etc. Almost all will have thermal protection.

it's up to the designers of the chassis that houses the chip to ensure adequate thermal protection for those chips. In this particular case, apple appears to be unable to provide enough cooling solution for the newer chips so that they are frequently able to hit temperatures that trip these thermal limits.

Apple is no stranger to thermal throttling. Virtually all of their devices have suffered some levels of it. this is a trade off for the ultra-thin form factors that they strive for.

Typically, those thermal limits are hit under long consistent loads. iMac i7 throttled for example, but only after a decent amount of time being hit by CPU load and is really meant for more "burst" like performance.

From my recent memory, this MBPro is probably the worst throttling device they've had since the G4 Cube.

 

[DNichter]

I think that you (and the others who suggest this route) highly underestimate the effort to move to ARM for macOS and all third party applications.

I don't discount the effort needed to do this, but I also don't think this is something Apple just realized they needed to do yesterday. This has likely been in development for years. Marzipan is the writing on the wall, it's just a matter of time. They don't need all 3rd party applications, they aren't Microsoft. They need the handful of large developers on board and the rest of the industry will follow suit.

 

[CodeJoy]

I'd love to agree with you on this one. It's absolutely unfathomable that a company the size of Apple would ship a device without robust valid testing


But yet, here we are... the Devices aren't out even 2 weeks and they're already being reported by a lot of people (look at the other thread on this subject where it's verifiably throttling by users on this very forum)

This is something that Apple should have spotted immediately. Any system builder worth their weight will stress test their cooling solution. And yet, Here we are, Apple's latest MBPro throttles quite aggressively.

They obviously knew about it. They might have run out of time to fix it, management might have put unrealistic deadlines on release, they might have decided to not fix it for cost reasons, or any number of possibilities. But them not knowing about it is just not within the realm of possibilities. It's something they know about long before this even gets to testing. They know about it before they have the first prototype logic board in place. They know about it before they have sample silicon from Intel. They know about it from the minute they see the first spec sheet for the chip. It's just not something that got missed in testing.

 

[madKIR]

Anyone who thinks that their laptops are not terribly overpriced is delusional! From the terrible keyboard design to flaws like this one, I really can't justify paying about 2K for a machine that is absolutely unreliable. They fix one thing (allegedly with the new keyboard design) and introduce something else. On top of that, Apple is never really honest about their mistakes either.

 

[Baymowe335]

I just wanted to say that its a moot point how many products Apple ships in a year. This is a problem in ALL MacBooks, not an issue resulting of volume. Obviously, during prototype phase, someone either did not see this problem OR decided it can be fixed after release.

If they did NOT catch it, I don't know how you can argue that its not their fault. Yes mistakes happen but that is literally their job to identify problems before the device goes into mass production. So obviously someone at Apple failed at their job.

IF they decided it IS a problem but will be fixed in future firmware updates then, we can argue they did a disservice to their customers by releasing a product before it was ready. Someone might have done a cost/benefit analysis to identify if potential perception problem this will create is worth the monies they would lose by delaying mass production. But yes, in this case, we don't know what really happened. But your point that Apple would not ship something knowing is broken is insane is not true. It all depends on their financial contracts and what delaying mass production would mean.

Just my 2c.

Great. Say they failed if you want...I don't think you know, but let's just assume it was a mistake. They'll fix it. Get over it. It doesn't mean the entire product line is flawed or the company is in turmoil.

 

[H2SO4]

Explanation below. Intel developed the power delivery chip. Apple certainly plays a factor here as well, but I still think it doesn't bode well for the future of Intel in Apple machines.

"The user goes on to explain that one of the internal power limits set for the device may not be appropriate for the power draw of the CPU and identical to previous MacBook Pro models, causing the power delivery chip (known as a voltage regulation module, or VRM) to report an over power condition that forces the clock of the CPU down to scale back power. This sets up the same conditions to allow throttling to occur once again."

Right, the way that reads is that the wrong item or setting was chosen. That makes it an Intel issue?

 

[netdoc66]

Apple continues to upgrade some parts of their hardware while leaving old chips from previous models in place. This has happened before. It’s still happening. This company has lost my loyalty.

 

[CodeJoy]

This is a Core i9, not an i7, not an i5, not an i3, not a Pentium and not a Celeron. High end processors have higher requirements to deliver higher performance.

This argument is nonsensical when you know that the i7s and the i9 are effectively the same chip. Same with the i5 and the i7 in the 13". i3/i5/i7/i9 are just marketing terms, they mean very little in terms of anything technical.

 

[flat five]

These are just simple benchmark stress tests. They peg the CPU at 100% to see how the system handles the load, from both heat and performance.

This is standard methodology for testing heat and burn-in of a device, and accurately represents any load that will use 100% CPU power. For many people, this is a valid test as they do have their CPU's pegged at 100% while doing certain tasks.

i get that.
and i really do believe i personally have one of the very few (ie— not ‘many’ people.. relatively speaking) instances of doing actual work while the computer is hitting all cores 100%..

most instances of when these stress tests can be related to real world has the user not actually interacting with the process and are instead eating lunch or sleeping.
(and typically, not happening on a laptop anyway)

but even then, when i get to the point of needing to use and wait and use and wait, i’ve already completed most of the work.. say, spent 30hrs on the project and now have 30mins of this weird working condition (preview renders prior to sending to final where material properties need fine tuned)..

most of the project, i want super fast single core stuff.. or a very responsive no-lag environment..
do these 2018MBP handle that well or not?

none of these tests reflect this or even show any concern for actual working conditions. but everybody is fighting about how ‘non-pro’ the machines are?
it doesn’t make much sense

 

[LordVic]

This is a Core i9, not an i7, not an i5, not an i3, not a Pentium and not a Celeron. High end processors have higher requirements to deliver higher performance.

Incorrect.

Thermal limits aren't based on what core I version they are but the specced TDP.

an i3, i5, i7, i9 all rated at 28w, all produce 28w heat energy. same with 15w, 95w, 115w. Intel specs them to fit within those thermal envelopes.

Now, it's possible that Intel has wrong specs. I can't claim otherwise, But at the same time, it's apples responsibility to test thoroughly their design choices with the CPU's they choose.

 

[Ploki]

i get that.
and i really do believe i personally have one of the very few (ie— not ‘many’ people.. relatively speaking) instances of doing actual work while the computer is hitting all cores 100%..

most instances of when these stress tests can be related to real world has the user not actually interacting with the process and are instead eating lunch or sleeping.
(and typically, not happening on a laptop anyway)

but even then, when i get to the point of needing to use and wait and use and wait, i’ve already completed most of the work.. say, spent 30hrs on the project and now have 30mins of this weird working condition (preview renders prior to sending to final where material properties need fine tuned)..

most of the project, i want super fast single core stuff.. or a very responsive no-lag environment..
do these 2018MBP handle that well or not?

none of these tests reflect this or even show any concern for actual working conditions. but everybody is fighting about how ‘non-pro’ the machines are?
it doesn’t make much sense

running too hot == diminished performance. as simple as that.

 

[nvmls]

You've proven the point. You're overreacting. Macs sales will be excellent because they are great laptops for people who don't think the entire product line is flawed because of this brand new issue one 1 model that will be fixed.

One issue? good one.

 

[LordVic]

i get that.
and i really do believe i personally have one of the very few (ie— not ‘many’ people.. relatively speaking) instances of doing actual work while the computer is hitting all cores 100%..

most instances of when these stress tests can be related to real world has the user not actually interacting with the process and are instead eating lunch or sleeping.
(and typically, not happening on a laptop anyway)

but even then, when i get to the point of needing to use and wait and use and wait, i’ve already completed most of the work.. say, spent 30hrs on the project and now have 30mins of this weird working condition (preview renders prior to sending to final where material properties need fine tuned)..

most of the project, i want super fast single core stuff.. or a very responsive no-lag environment..
do these 2018MBP handle that well or not?

none of these tests reflect this or even show any concern for actual working conditions. but everybody is fighting about how ‘non-pro’ the machines are?
it doesn’t make much sense

I do agree that we still need to do a lot more testing of the thermals and see what exactly triggers them. I'm one of those nuts that will run 100 different tests just to prove something to myself.

does the device throttle if it's 100% load on a single core? what are the temperature rampups based on what specific load? there's a LOT to be tested.

For my work, I can easily peg the CPU's on multiple cores at 100% while also managing 100% IO utilization (Heavy database maintenance work). Sometimes I do testing on my laptop to avoid potential destructive work on a live system. CPU throttling could make my work go from 2-3 hours of processing to 5-10 hours, which for me would be completely inexcusable.

But the simple fact that we already have evidence of throttling at 100% load means the thermals aren't being handled well enough at that 100% load.

 

[DNichter]

Its not an IntelBook. It’s a MacBook.

I agree and Apple needs to take that next step and remove Intel from the scenario.

 

[hudson1]

Right now... ALL of Apple's laptops are "thin" including the supposed "pro" laptops. You have no choice but to buy a thin Macintosh laptop.

Macbook - THIN
Macbook Air - THIN
Macbook Pro 13" - THIN
Macbook Pro 15" - THIN

It's too bad Apple can't have at least one laptop that prioritizes horsepower/airflow instead of looks. A little extra thickness wouldn't be that bad if it serves a purpose.

Seriously... these are Apple's pro machines. Can a laptop be TOO SKINNY...?

The MBP is 15.5 mm thick. The Dell XPS, also available with the new i9 chip, is 17 mm thick. Not a big difference.

 

[Michael Scrip]

The MBP is 15.5 mm thick. The Dell XPS, also available with the new i9 chip, is 17 mm thick. Not a big difference.

And if there was a laptop that was 19mm thick... it wouldn't be noticeable either.

I'd rather have a 4.5 lb laptop that doesn't throttle... than a 4 lb laptop that does throttle.

There's only so much heatsink and airflow you can put in a "thin" laptop.

 

[nikhsub1]

Wouldn't sustained high power usage (gaming/video-encoding) after applying the "patch" mentioned in the article cause damage due to over-heating? Or are these components throttled based on temperature independently from the chips mentioned in the article?

There are definite thermal throttling safeguards built into every CPU independent from power delivery. FWIW, when overclocking on a desktop system, you can up the voltage to the CPU though the bios - obviously with more vcore comes more heat. From this article here, it seems like this is an apple component thinking the CPU is using too much power, a simple firmware update should be able to fix that pretty easily.

 

[CodeJoy]

most of the project, i want super fast single core stuff.. or a very responsive no-lag environment..
do these 2018MBP handle that well or not?

My understanding is that they handle single core stuff extremely well. Maybe this was your point. But beyond that, it's also worth considering why modern CPUs function the way they do at all. User usage patterns get analysed by CPU manufacturers, and they align the design so that the most frequently usage patterns go faster, whereas less frequently usage patterns get a lower priority. Many current usage patterns involve short bursts of single core performance, and then backing off again. There may even be multiple such patterns happening in parallel, but the point is that this is something modern processors (and the 2018 MBP) do exceptionally well. But this is harder to test for, so it's not what gets tested in reviews. Reviews instead tend to test what's easy to test, not what's actually relevant to test.

 

[DNichter]

Right, the way that reads is that the wrong item or setting was chosen. That makes it an Intel issue?

I am not sure who is responsible for choosing that item or setting, but yes I would think Intel is at least partly responsible for the problem. Apple likely tested the same way they had with previous MacBooks and did not account for Intel not increasing the thermal design power. It's an Apple machine so of course it falls on their shoulders to fix, but I still don't feel this bodes well for the future of Intel in Apple machines.

"These conditions may be presenting themselves due to the new six-core design of the i9 CPU featured here. While Intel increased the core count of the CPU, they did not increase the thermal design power (TDP), or the amount of dissipated power manufacturers should plan to have to cool for a proper CPU design."

 

[chrmjenkins]

Explanation below. Intel developed the power delivery chip. Apple certainly plays a factor here as well, but I still think it doesn't bode well for the future of Intel in Apple machines.

"The user goes on to explain that one of the internal power limits set for the device may not be appropriate for the power draw of the CPU and identical to previous MacBook Pro models, causing the power delivery chip (known as a voltage regulation module, or VRM) to report an over power condition that forces the clock of the CPU down to scale back power. This sets up the same conditions to allow throttling to occur once again."

The chips in question are not of Intel design. Apple choose specific parts to regulate voltage for the CPU and GPU, and they also set the limits in software for power draw. Those individual components have their own thermal shutdown mechanisms as well.

Intersil and IR are well known names in the enthusiast motherboard industry on the PC side. The latter is usually preferred and featured on high end options aimed at overclockers.

 

[CodeJoy]

Incorrect.

Thermal limits aren't based on what core I version they are but the specced TDP.

an i3, i5, i7, i9 all rated at 28w, all produce 28w heat energy. same with 15w, 95w, 115w. Intel specs them to fit within those thermal envelopes.

Now, it's possible that Intel has wrong specs. I can't claim otherwise, But at the same time, it's apples responsibility to test thoroughly their design choices with the CPU's they choose.

TDP doesn't define thermal limits, at least the way Intel uses it. But that's known to any hardware engineer in the whole world, nothing new.

 

[IPPlanMan]

The entire product line isn't flawed. We are only talking about certain MB Pros and this issue can likely be fixed with some tweaking.

See how you're overreacting?

There’s no overreacting here on my part.

Apple released a flawed product. Period.

It might be fixable. Then again, it might not be.