Sidebar Sidebar
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.

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

For myself, even the slim chance that I do purchase a new MacBook with an i9 in it, the chances of me pushing it to the point of it slowing down are probably in the realm of 0.08% and 0.06% chance.
So it is a non-issue. Again for me.

Now, am I going to personally drop that level of cash on this thing just for the sake of doing it? No
Do I have a job that requires a mobile computer with that level of power? No
If I did have a job that required that level of power would I decide to purchase a laptop rather than a desktop? Answer is still probably a no.
Do I just want for the sake of wanting a laptop with that power? I am still leaning towards a no simply because of the additional size and weight of the laptop.
I purchase my portable computers for travel. Meaning the least amount of weight coupled with decent CPU power is my sweet spot. 50% plus of my time is spent at a desktop when I need to do extended work sessions so I would never (at this particular point in my life/career) choose a laptop costing north of $2500 maybe $2800 max. Just simply because I do not have any value in a laptop costing that much, so that makes the cash that it costs a waste in my eyes.
I have the ability to buy it I just am not going to.
 
This is from Extremetech

Intel said back in 2015 that it was okay for manufacturers to apply their own thermal throttling approaches, even if that resulted in wildly different performance from the same CPU.

Intel explained that it had given OEMs more freedom to set certain specifications for their own systems. One example the company gave was skin temperature: If OEMs wanted to specify a low skin temperature that had the side effect of keeping the CPU clock speed from turboing as high as a competitive system from a different OEM that traded a higher skin temperature for better thermal performance, that was fine by Intel — even if it also meant two systems with the same CPU might perform very differently.
Click to expand...

Author of the article Hruska added

Forget the idea that Apple, Dell, HP or any other OEM can’t possibly know what kind of workloads their customers are going to run. The major OEMs know exactly what their high-end professional customers run because, when push comes to shove, there aren’t all that many high-end applications that compete at the top of the market […]

OEMs could certify that their laptops don’t throttle […] Define a standard suite of software using common applications. Explain the reasoning. Perform the tests.
Click to expand...

Source: https://www.extremetech.com/computing/273917-cpu-throttling
 
This is an Apple Maps situation. There needs to be an acknowledgment + apology from Apple with a promise to do better.
 
Does anyone think that this would still be an issue if Apple’s professional laptops weren’t so thin?

Of course it wouldn’t be.
 
There were pages of people trolling suggestions of 800MHz dips before, based on the CPU not being at critical temperature when that happened. Now it looks like those dips are very real and VRM related, and they've been corroborated by many sources now.

This is my least favorite thing about the community - the gaslighting of legitimate issues before all the information is out, often right up until Apple admits to and fixes a mistake. Blind defense gets us nowhere, why shouldn't we always want things improving?
 
  • Like
Reactions: VictorTango777
tipoo said:
"This fix will not address total system power draw becoming excessive, such as long sustained loads from the CPU and GPU, but it is possible Apple could issue a fix similar to the one outlined in the reddit post that is stable. "


Fix is a strong word - it's reducing the max power state, and time spent in that state. The VRMs are underspecced for the job in the end, and I wonder what running them to their limits will do for board life.

It looks like there certainly is improvements to be made, and those will reduce load on the VRMs, but the i9 should have been paired with higher rated ones.
Click to expand...
Have we actually seen a solid and educated VRM and power delivery analysis of the i9 model anywhere? I'm not fully sure that this is an area of expertise of MacRumors.
 
Rogifan said:
Who added additional thickness to accommodate this (besides maybe gaming laptops). Doesn’t the XPS have similar issues?
Click to expand...
Other manufacturers would not make such drastic and profound sacrifice on performance just to keep with the aesthetics and design of the laptop. Last year i7 outperforms this year i9 in normal room temperature. This is unacceptable.
 
I think the biggest thing Apple lost with the passing of Steve Jobs is the ability to hold a product back until they got it right. Steve always emphasized that they didn't ship junk, and while there were a few duds (iPod HiFi, MobileMe, etc.), they were still great products and they "just worked". The i7 2018 MBP is already a very powerful machine. No one was really expecting an i9, yet apple didn't hold it back and shipped it, charging people over $3000, without working out any of the cooling issues. Hope Apple can pull it together and not forget us Mac users.
 
All was going alright until we got to the last paragraph:

"Along with the heatsink path provided for the IR parts, it's clear they will not be capable of driving the same amount of load in any sustained mode. This makes sense given GPUs can see high loads for longer periods,"

So the parts "will not be capable of driving the same amount of load in any sustained mode", but "GPUs can see high loads for longer periods", so this "makes sense". Forgive me if I am wrong but according to my understanding of the English language and grammatical constructs, what is presented here is a juxtapositional anomaly.
 
DNichter said:
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."
Click to expand...
Both the Cube and the Trashcan mac pro should prove to you that Apple can still get heat management BADLY wrong, despite all the practice they have in building thin enclosures....
 
jsalda said:
I feel this is more Intel's fault. Apple is caught in the middle, power users want the latest and greatest chips and Apple is trying to fulfill that request, but Intel's ability to deliver is lacking. This isn't just an Apple issue, any other manufacturer that uses these chips is going to have the same issues.
Click to expand...
Throttling is the fault of the company that created the cooling solution for its computer. There are PC laptops that don't throttle down below i7 performance using the i9.
 
H2SO4 said:
Both the Cube and the Trashcan mac pro should prove to you that Apple can still get heat management BADLY wrong, despite all the practice they have in building thin enclosures....
Click to expand...
I have both of those computers and both run well, When I upgraded the CPU and GPU of the Cube I did add a quiet fan. I know the upgrade path on the trash can is limited but it is a great work horse and has pretty strong specs, even now. I paid $1,500 for my trash can - Apple referb and got AppleCare, I upgraded the SSD and the Ram and it is a beast, I do a lot of video encoding and it warms the room up eventually but that means it is handling the heat and pushing it out.
 
Both the iMac not having hardware encoding and this flaw in the MBP are obviously to push the youtubers towards the upcoming Mac Pro which probably will have custom hardware encoding like the iPhone has.
 
Apple has been so focused on the iPhone that the true neglect of the Mac is evident if this information is true. It also says that Apple really needs to rid itself of Intel as fast as it can.
 
CodeJoy said:
My understanding is that they handle single core stuff extremely well. Maybe this was your point.
Click to expand...

sort of my point but it's also where my concern lies..
when i see the stress tests, they're showing the throttling happening in a sporadic fashion.. the chip appears to drop to very low frequency then spike to a high one etc.. and averaging out somewhere in between.

i'd expect it to draw more of a straight line instead of all this up and down stuff.

so my concern is what happens when one may be pushing linear calculations.. (single core stuff... the vast majority of CAD processes).. is there a chance a MBP i9 is going to give you 800MHz one second then 4.5GHz the next? even when the process wants all of the core?
or is the CPU more stable when turbo_ing with a single core?

i just haven't seen much in the way of this type of testing which would correlate better with my use case (and i believe most other people's usage requirements)


----
or, to be a little more clear.. i'm not simply trying to make a point about real world usages..
instead, i'm most likely going to be buying one of these machines.. the 2018 i9 MBP looks freaking awesome on paper for what i do.. 4.8GHz turbo in a laptop? sweet, sign me up.. but all of these multicore tests, and more specifically, the way the throttling is unstable in appearance, has me concerned..

so yeah, 1/2 concerned 1/2 making a 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.
Click to expand...
yep, i see.
thanks for your thoughts
 
  • Like
Reactions: CodeJoy
Spizike9 said:
I’m going to do the same. I was going to get the 2.6 but now I’m just going to get the base model and upgrade the HD
Click to expand...

Same.

Seems like Apple did a great resign on the cooling system of the iMac Pro. I've heard the 5K iMac i7 sounds like a wind tunnel, with its single fan straining. Meanwhile the iMac Pro, with double fans and a more robust cooling system, makes almost zero noise. (At least, I've heard/read that... I don't have either system.)

The iMac pro was a clever tweak of an existing design from Apple.

What would the clever tweak to the MPB have been, to avoid throttling and dissipate heat better?

I'm thinking, there wasn't a possible tweak. It's just too much heat. Maybe there is no possible solution to this problem, without a total re-do on the MBP chassis/basic design.

Ideas:

MPG Space Shuttle Tile NASA Edition

MBP where heat gets absorbed in creating energy to recharge the battery?
 
IPPlanMan said:
“They’ll fix it”.... is that the new standard for Apple now?
[doublepost=1532447697][/doublepost]

Spot on. Exactly right.
Click to expand...
It’s only a standard if you only focus on isolated problems.

Btw, they are fixing it, so I was right. Not a big deal.
 
flat five said:
sort of my point but it's also where my concern lies..
when i see the stress tests, they're showing the throttling happening in a sporadic fashion.. the chip appears to drop to very low frequency then spike to a high one etc.. and averaging out somewhere in between.

i'd expect it to draw more of a straight line instead of all this up and down stuff.

so my concern is what happens when one may be pushing linear calculations.. (single core stuff... the vast majority of CAD processes).. is there a chance a MBP i9 is going to give you 800MHz one second then 4.5GHz the next? even when the process wants all of the core?
or is the CPU more stable when turbo_ing with a single core?

i just haven't seen much in the way of this type of testing which would correlate better with my use case (and i believe most other people's usage requirements)


----
or, to be a little more clear.. i'm not simply trying to make a point about real world usages..
instead, i'm most likely going to be buying one of these machines.. the 2018 i9 MBP looks freaking awesome on paper for what i do.. 4.8GHz turbo in a laptop? sweet, sign me up.. but all of these multicore tests, and more specifically, the way the throttling is unstable in appearance, has me concerned..

so yeah, 1/2 concerned 1/2 making a point :)



yep, i see.
thanks for your thoughts
Click to expand...
I suspect single score stuff would be completely fine. Linear calculations would presumably be AVX workloads, which are extra power hungry, but for a single core it should still be ok. I can't imagine that you would see the same kind of VRM throttling then. Frankly though, get the 2.2 base model and save the money for the upgrade. I've been saying this from the day they were released, it's just the wiser decision. This is not a quick fix situation, and if Apple eventually revises the logic board, I'm sure you can sell the 2.2 and get the i9 then.

I don't have an i9 to test on, so take this for what it is. (and I don't intend to get one)
 
  • Like
Reactions: flat five
You must log in or register to reply here.
Register on MacRumors! This sidebar will go away, and you'll see fewer ads.
Top Bottom
Back