2018 i9 MacBook Pro throttling discussion [merged]

[wallysb01]

Sure appears Apple pulled a 2013 Mac Pro and boxed themselves into a thermal corner with their desire to go thin and light above all else. Thus not leaving any thermal room for more power hungry CPUs or GPUs. This is really, really disappointing. I don't typically push my laptops on heavily multithreaded tasks, but I do expect them to more or less hang in there when I do. More typically, I use a good bit of RAM, where the jump to 32 GB is extremely welcome, and only push 1 or 2 threads at a time. Thus, the max turbos are of interest. However, the i9 max turbo to 4.8 is turning out to be a total facade. There is no way the CPU will be at <50 C under any load what so ever. So the max we'll likely see is 4.6. One user reported his i9 only turbo-ing to 4.2, making me suspect bumping up the processor at all an is unwise choice. The 4.1GHz jump is still a decent jump for the base model (assuming the base can hit that) compared to the previous base at 3.8.

 

[CrashTestWalrus]

I have one coming today and I’m curious if this just isn’t apple and their terrible job of applying thermal paste. I would say a full 60% of the many Apple laptops I have dealt with since 2012 have had too much thermal paste applied which has caused premature heating then throttling. It’s a simple repair to have done in the store and has fixed every laptop I’ve had to have serviced this way. What’s terrible is that they leave the factory like that, they are way too expensive for that to be occurring at that kind of rate. Because this has occurred so frequently the first thing I do with my laptop upon receipt is stress test with the power gadget running to watch the clock. If it drops the base frequency incredibly quickly I return it before the window is up. If the reality is with the I9 there just isn’t adequate cooling I would be shocked that Apple would sell a laptop like that. Having too much thermal paste and it causing cooling problems sucks on machines this expensive, it’s an easy repair though. A design flaw where there is just no way to cool the laptop adequately would be terrible.

 

[aevan]

Guys, to those of you bashing the throttle criticism...

I could say the same thing about all the people constantly whining how Apple should make thicker computers.

But I’m not. Forums are a place for discussion.

 

[Ma2k5]

I personally think this was overblown. My office has around 40 2016-2017 MacBooks with the butterfly switch keyboard. As of today take a guess how many have been sent in for repair. One. For a non-keyboard related repair.

We are all hipsters and eat pizzas and chips and drink cokes at our desks. I eat saltine crackers all the time. At least 2 small packs a day in front of my computer and I have yet to experience dead keys.

So I can honestly say, just based on my office, Apple’s keyboard actually works. Whether I love the keyboard, that’s a totally different story.

To be fair, even 1/100 would be a very high failure rate, 0/40 not so much to go on.

Also, do any portion of your office use external keyboards/mice? Then there is a tolerance level - people may be experiencing issues but they just live with it.

In my office, we all get provided surface laptops. The amount of issues experienced vs those raised with IT (judging by tickets) is pretty crazy. A lot of people tolerate poor experiences in an office environment vs at home - myself included.

Until something becomes non-functional or you lose patience, it doesn’t get reported much. I’ve seen people work with horizontal lines on defective monitors and stand alone keyboards with issues for years. I experienced an issue with the keyboard on my 2016 MacBook Pro and returned it - if I experienced the same in an office environment, I wouldn’t have even brought it up (until the point one of the key mechanisms broke).

 

[Schranke]

What do you do to stress the CPU of the system? Just curious.

One could do a torture test with prime95, it will use all core 100% until you stop it. So if there are any problems related to CPU temperature it should catch it. The test is simple and you only need 2 programs, Prime95 and Intel Power Gadget.
Download an run both, intel power gadget will give you the numbers you need while doing the test.
For prime95, run it, go to option, select Torture Test, pick "In-place large FFTS". now see how it goes crazy in intel power gadget, take a screenshot once the idle temperature is almost out of the graph, let it run for a extended period and take another screenshot. we should now have the turbo boost speed and how long it can be sustained + t-max stable speed

 

[mr.anthonyramos]

Perfect, thank you for that, that number alone plus double the ram make it a still very worthy upgrade for me.

I am more of a director / producer for motion work but I will still need occasional use of FCP and Premiere, my gut feeling is that it needs more time to play out and I can do those things on my iMac Pro.

I’ll stick with my order for now.

Sure! Glad this helped somehow. Like I said in another post not a “technical” numbers test but my colleague and I just did it (twice) for the sake of seeing where my 300 bucks went.

From what I’m seeing, comparatively, the i9 does better at most things but apparently in some YouTuber test it didn’t work out too good in FCP. But then another review using FCP compared also with a 2017 i7 top spec was beat by the i9.

Moral of the story is, sure numbers seem daunting and technical tests are great but real life testing might actually prove otherwise.

 

[billhinge]

My 2015 i7 throttles to just under 3.0ghz (3.1 base) and reaches 102C running prime95, fortunately I don't run prime95 all day.

 

[mr.anthonyramos]

To be fair, even 1/100 would be a very high failure rate, 0/40 not so much to go on.

Also, do any portion of your office use external keyboards/mice? Then there is a tolerance level - people may be experiencing issues but they just live with it.

In my office, we all get provided surface laptops. The amount of issues experienced vs those raises with IT (judging by tickets) is pretty crazy. A lot of people tolerate poor experiences in an office environment vs at home - myself included.

Until something becomes non-functional or you lose patience, it doesn’t get reported much. I’ve seen people work with horizontal lines on defective monitors and stand alone keyboards with issues for years. I experienced an issue with the keyboard on my 2016 MacBook Pro and returned it - if I experienced the same in an office environment, I wouldn’t have even brought it up (until the point one of the key mechanisms broke).

I guess what I was trying to point out is how people have this mentality that the keyboards were MADE to fail. Sure they aren’t perfect but I doubt they are THAT fragile.

We use our computers with and without external monitors and maybe a few of us use external keyboards but even if they do, we run around meetings and towns dragging out macs along so usage I would say is slightly more than normal.

 

[Queen6]

Of course, that is great if you want to lug a 17” which is exactly 1kg heavier than the 15” MBP, btw. This is not a small difference.

Again, this is fine and great if you want that kind of computer. Many do, but many don’t. This is why you have an Asus and I have a MacBook Pro.

I am not writing posts about how Asus doesn’t care about their customers when they make computers that weigh 3kg, because I understand that is a tradeoff some people are willing to make. Emphasis on some.

Whole 1Kg my, my what's the world coming to...

It's simply too early to say with any certainty regarding the 2018 MBP, equally initial reports are not encouraging. If I or others drop $4K upwards I think it's fair to expect the advertised performance...

I have Acer, Asus, MSI, 3 15" MBP's, 13" rMBP, 2 Air's, and a Retina MacBook, Surface Book I gifted to my daughter and I'll comment on whatever I want on public forums.

Your more than welcome to join in on other forums, but if you can...

Q-6

 

[Elektrofone]

Sure! Glad this helped somehow. Like I said in another post not a “technical” numbers test but my colleague and I just did it (twice) for the sake of seeing where my 300 bucks went.

From what I’m seeing, comparatively, the i9 does better at most things but apparently in some YouTuber test it didn’t work out too good in FCP. But then another review using FCP compared also with a 2017 i7 top spec was beat by the i9.

Moral of the story is, sure numbers seem daunting and technical tests are great but real life testing might actually prove otherwise.

I need to do some testing myself on my new i9, but I wish I had the 2.6 i7 to test and compare with in After Effects and Premiere.

 

[duervo]

Not surprised it's looking this bad on Premiere Pro. Throttling is there for sure.

Just wondering why people are ignoring iPhonedo's video though? Seems like rendering times are faster on FCP X across the board.

Might have something to do with the Salt n Pepa parody at the beginning of it. Speaking only for myself, I stopped playing their video after about 5 seconds (not a fan of Salt n Pepa.)

 

[kotlos]

I don't know if this has been posted before but here it shows how much difference there is between XPS15 and macbook pro 15:
https://www.notebookcheck.com/Das-n...-vom-XPS-15-deutlich-geschlagen.317258.0.html

By the way I just received mine and there are definitely a lot of things that can be optimized. The computer hangs quite a bit (similar to what has been shown in some YT videos) and the fans seem to kick in late but stay on for quite some time after the temperatures drop. I have to investigate further but I really hope Apple fixes these issues.

 

[HerbertDerb]

Hopefully, Apple realizes the issues now and adjust this to the assembly line.
Probably this issue could be minimized (to some degree) with better thermal paste between the cooling and CPU.

And imo, stress testing laptop CPU with prime95 is stupid. It's useful if you want to find out if your overclocked CPU is stable. But in real world, the CPU will be never hitting 100% at all 6 cores in the same time for that long.

 

[Smeaton1724]

Nonsense, they were able to cool down Skylake and Kaby Lake just fine. Since 8th gen has the same thermal specs, it they should behave similarly.



We are at a point where Intel sells overclocked chips that draw double of their spec power under normal operation. Intel should get their game together and release chips with true perf-per watt increases.

If Apple know/expect/test and establish Intels chips draw double the power of their spec sheets then they can’t simply put those processors in a tiny case and just point at a spec sheet and go “intels fault” they don’t comply. They are just as complicit as these chips go on their machines, they have a duty of care to the customer!

Soon as those machines ship with those chips Apple are just as compliant as Intel.

 

[Hitrate]

So anyway this might mean the 2013 macpro is a better bet, since it has no throttling issues...oh well a cheap car and a used macpro doesn’t sound too bad either.

 

[DjanSeriyAnaplian]

I could say the same thing about all the people constantly whining how Apple should make thicker computers.

But I’m not. Forums are a place for discussion.

I think people would rather Apple utilised their $256 billion cash pile to RnD a decent cooling solution instead of selling something that can't perform for an outrageous price.

But perhaps i'm wrong...

Lol nah ofc i'm not haha

 

[akis-k]

The irony is that Apple is the last company that placed these cpu's in their laptops. They ought to know the throttling issues, since these cpu's where launched about 4 months ago, and laptops like the xps had already problems with that. Doesn't the 1 trillion market cap company, have the money to do more strict quality checks on their products, before release?

 

[wallysb01]

The irony is that Apple is the last company that placed these cpu's in their laptops, They ought to know the throttling issues, since these cpu's where launched about 4 months ago, and laptops like the xps had already problems with that. Doesn't the 1 trillion market cap company, have the money to do more strict quality checks on their products, before release?

You are assuming they care enough to both check and subsequently do something about it if they even do check.

I'm guessing they just did this drop in because they were starting to get seen as too far behind, plus they wanted to roll out a preventative measure for the keyboard. However, I don't think they care enough to do a redesign to make these processors actually work for their computers. They may be hoping that the move to 10nm fixes this, allowing them to push off a redesign until 2020 or later.

 

[swerve147]

What about fan control? Are there any good fan control apps for Mac OS out there? An aggressive fan profile might just work, even to the point of allowing normal or *gasp* aggressive boost. Damn the noise, I just want my computer to be cool and fast.

Or maybe a firmware update?

 

[fate0311]

Correct me if I am wrong, these performance tests put the laptop at 100% CPU utilization.

Is it possible the results would be much different if a test was done that limited the utilization to say 80% or 85% over a period of time?

This would be to simulate someone who has workflow that is for sure making the laptop work but not putting it into 100% utilization territory.

 

[leman]

If Apple know/expect/test and establish Intels chips draw double the power of their spec sheets then they can’t simply put those processors in a tiny case and just point at a spec sheet and go “intels fault” they don’t comply.

Doesn't the 1 trillion market cap company, have the money to do more strict quality checks on their products, before release?

You are right. Still, what would you want them to do? Skip Coffee Lake alltogether? While everyone else is shipping it? How would that look like if all competitors are shipping these CPU (and throttling as well)? A chassis redesign would take too much time — until everything is tested and new production lines are in place, we'd probably already have next gen CPUs with improved IPC and much more reasonable thermals.
[doublepost=1531941954][/doublepost]

Correct me if I am wrong, these performance tests put the laptop at 100% CPU utilization.

I suspect that the real problem of these tests is simultaneous CPU and GPU utilisation. Likely, offloading work to the GPU causes the CPU to throttle. Looking at the videos in the first post, CPU is never really heavily utilized and its temperature is also not high enough to cause thermal throttling. Incidentally, these videos rely on video production software which is utilising the GPU.

 

[yesyesyoaa]

I am thinking of buying a new MacBook Pro 15, coming from a 15“ 2015 mid. Most of the time i am working with big excel files simultaneously, also doing a lot of photoshop.

Would I feel a difference between the i7 and the i9 in daily usage? I assume with my tasks I would never need the boost of the i9. So would it then never heat up and run smooth with 2.9 ghz base frequency?

 

[akis-k]

Correct me if I am wrong, these performance tests put the laptop at 100% CPU utilization.

Is it possible the results would be much different if a test was done that limited the utilization to say 80% or 85% over a period of time?

This would be to simulate someone who has workflow that is for sure making the laptop work but not putting it into 100% utilization territory.

True, but the same applies for other vendors too, who according to the tests are faster, even though they are using ''slower'' cpu's... And i am not referring to the bulky ones... https://www.notebookcheck.net/The-n...nd-clearly-beaten-by-the-XPS-15.317264.0.html

 

[subjonas]

Whole 1Kg my, my what's the world coming to...

Whaaaa? 1kg is insignificant in a car. It’s HUGE in a laptop, which is weighed by the gram. 12” MacBook is 1 kg!

 

[ronincse]

One could do a torture test with prime95, it will use all core 100% until you stop it. So if there are any problems related to CPU temperature it should catch it. The test is simple and you only need 2 programs, Prime95 and Intel Power Gadget.
Download an run both, intel power gadget will give you the numbers you need while doing the test.
For prime95, run it, go to option, select Torture Test, pick "In-place large FFTS". now see how it goes crazy in intel power gadget, take a screenshot once the idle temperature is almost out of the graph, let it run for a extended period and take another screenshot. we should now have the turbo boost speed and how long it can be sustained + t-max stable speed

No I wanted to know what s/he in particular does to stress his/her computer