thermal throttling More details about the throttling issues of the 15" core i9 MacBook Pro, this time with Final Cut Pro X

[D13H4RD]

One thing I find interesting is that after seeing P95 + FurMark testing, many of the Kaby CPUs brushed up right against 45W but few exceeded it and if so, it’s usually by 2-3W (mine hits a power limit when trying to go over)

 

The CFL CPUs have significantly higher power draw when boosted, like 62W on the 8750H when boosted but when in stock and running more lenient tasks, it’s lower 

[mr moose]

9 minutes ago, asus killer said:

 

 

Some quotes from people who are wrong

 

I am not sure what that person was thinking, but they are not right.

 

 

[mr moose]

9 minutes ago, D13H4RD2L1V3 said:

One thing I find interesting is that after seeing P95 + FurMark testing, many of the Kaby CPUs brushed up right against 45W but few exceeded it and if so, it’s usually by 2-3W (mine hits a power limit when trying to go over)

 

The CFL CPUs have significantly higher power draw when boosted, like 62W on the 8750H when boosted but when in stock and running more lenient tasks, it’s lower 

quirk of refined process?

[D13H4RD]

2 minutes ago, mr moose said:

quirk of refined process?

Probably. Intel hasn’t made a mainstream 6C 12T CPU for quite a while until recently.

 

i7 Extreme Editions don’t count 

[Brooksie359]

4 hours ago, Trixanity said:

My proposal included more cache and clock speed than the i7 so your point was useless from the get go. You have literally no point because you don't know Intel's lineup and have shown that ignorance in multiple posts now. If you can't at least look into the products we discuss, then move along please.

 

Look at Apple's other launch: MBP with U series. Still throttles to hell too. The i9 throttles in multiple other products (not all), as does i7 but to a lesser degree but it's convenient to ignore, isn't it?

Oh and it's also throttling in gaming laptops by the way. Just to be clear.

In both an Asus and Alienware laptop I believe.

You said adjust the clock speed to cause it to not throttle. The thing can't maintain its base clock so by the time you do that the clocks would be the same as the i7 so I have no idea why you think your proposal would have higher clocks than the i7.... 

[leadeater]

11 minutes ago, Brooksie359 said:

You said adjust the clock speed to cause it to not throttle. The thing can't maintain its base clock so by the time you do that the clocks would be the same as the i7 so I have no idea why you think your proposal would have higher clocks than the i7.... 

Just write a lower TDP on the spec sheet, that'll solve the problem .

 

[Lawliet93]

On 19. 7. 2018 at 12:57 PM, ImpulseRez said:

A bit ridiculous to blame Intel on this one. Apple must have made a conscious design choice to keep the laptop thin at the expense of cooling. Intel chips are pretty good from an efficiency perspective (better than AMD at least), and being the first 6 core laptop chips are always going to be thermally dense to start with.

Also, bringing up the ARM mention is pretty absurd (performance compared to an 8th Gen 6 core Intel CPU is not going to be in the same league!)

No, they are not more efficient than AMDs Ryzen chips. 

[Trixanity]

48 minutes ago, Brooksie359 said:

You said adjust the clock speed to cause it to not throttle. The thing can't maintain its base clock so by the time you do that the clocks would be the same as the i7 so I have no idea why you think your proposal would have higher clocks than the i7.... 

Quite simple. The i7 also throttles. We have to assume the i9 is binned. It should therefore behave better. So if you lower the clocks across the board but keep them above the i7... What happens? The i9 is faster but runs better than it did before. If you adjust the fan curve (which you can easily do yourself if Apple doesn't) in addition to the frequency curve it'll run much much better. You'll have better control of thermals, frequencies and ultimately battery life as well. 

 

There is something wrong with Apple's software implementation though. It goes beyond just fan curves. It spikes the power and thermals but never really recovers after the thermals settle after the fan kicks in and the frequency drops. Changing the curves will also avoid the spikes causing this slow down.  So either it's just shit or it's not just thermals but power throttling as well. Which makes sense given the fact the frequently used FCP hammers the Intel GPU as well. But then we're back to it becoming Intel's problem if hammering both breaks the power limits. Because all the Windows laptops run dedicated GPUs and don't run QuickSync. So you'd need to run a similar benchmark on a Windows machine to determine that (preferably XPS 15 or similar to account for similar cooling systems and power delivery but a Gigabyte laptop should work too). In fact it'd be funny if it was Apple actually embracing Intel's tools that broke their chip because it seems no one else cares about the IGP.

[Brooksie359]

5 minutes ago, Trixanity said:

Quite simple. The i7 also throttles. We have to assume the i9 is binned. It should therefore behave better. So if you lower the clocks across the board but keep them above the i7... What happens? The i9 is faster but runs better than it did before. If you adjust the fan curve (which you can easily do yourself if Apple doesn't) in addition to the frequency curve it'll run much much better. You'll have better control of thermals, frequencies and ultimately battery life as well. 

 

There is something wrong with Apple's software implementation though. It goes beyond just fan curves. It spikes the power and thermals but never really recovers after the thermals settle after the fan kicks in and the frequency drops. Changing the curves will also avoid the spikes causing this slow down.  So either it's just shit or it's not just thermals but power throttling as well. Which makes sense given the fact the frequently used FCP hammers the Intel GPU as well. But then we're back to it becoming Intel's problem if hammering both breaks the power limits. Because all the Windows laptops run dedicated GPUs and don't run QuickSync. So you'd need to run a similar benchmark on a Windows machine to determine that (preferably XPS 15 or similar to account for similar cooling systems and power delivery but a Gigabyte laptop should work too). In fact it'd be funny if it was Apple actually embracing Intel's tools that broke their chip because it seems no one else cares about the IGP.

Yes but the i9 is meant to be at those very high clocks for the very high end gaming laptops. You lower the clocks on the i9 and it would change it's original intent. I think at the end of the day the i9 is essentially putting an i7 8700k in your laptop so you had best be ready to put cooling adequate enough to keep up with an 8700k. 

[D13H4RD]

Wat

 

(That's the 8950HK on what I presume is an MSI GT75. The CPU is at 3.8GHz) 

[Trixanity]

10 minutes ago, Brooksie359 said:

Yes but the i9 is meant to be at those very high clocks for the very high end gaming laptops. You lower the clocks on the i9 and it would change it's original intent. I think at the end of the day the i9 is essentially putting an i7 8700k in your laptop so you had best be ready to put cooling adequate enough to keep up with an 8700k. 

It's up to the designers to make it work in whatever way. Microsoft worked with Intel to get an i5 to run fanless. There's an obvious performance penalty by doing that. It's stupid to use the i9 in that manner but it should be workable and in a fashion that's at least slightly faster than the i7 alternative. As I said beyond the cooling system itself there are some software issues that exacerbate it. No amount of software changes can fix it and certainly not without a cost but it should still be possible to make a passable solution and still see a decent enough result. I would really like to see multiple samples tested with temperature and power benchmarked across every point in the frequency curve because it'll be important to determine wherein the problem lies because just saying it's the cooling and shrugging is a simplification and a cop out. Because of course the cooling is terrible but multiple systems with the same problem indicates the issue isn't so cut and dry as just Apple being Apple.

 

And this doesn't even account for MBP13 throttling and running hot with a U series processor of which Apple is the sole customer and the target audience.

[mynameisjuan]

Wow the people in this thread that say its intels fault. Bullshit. 

 

Plenty of laptops have the same chip and dont thermal because their actually test their coolers and make sure its up to par. Its a tested CPU, gives off this much heat at this clock, draws this much power. Not fucking hard to test Apple.

 

Im sorry but anyone who thinks this is in any way intels fault is mental. And dont give me the "but they should have soldered it" because that is not the point.

[DeScruff]

From all I'm reading this seems like 1% Intel's fault (The TDP being only slightly higher is flat out misleading. It's a significantly hotter chip, and everyone knows it.)

 

But 99% on Apple for: A) not giving this chip a beefier heatsink/more aggressive fan curve, or B) flat out offering this chip as an option.

I doubt there would be any hubbub if they didn't offer an i9 config. I don't even think anyone would have even noticed. This isn't at all like the max 16gb of ram thing.

 

I have to imagine Apple did testing on a machine with an i9 config. Even just to get a nice "it's 30% faster then the 20__ model!" And saw that it gets too hot.

[captain_to_fire]

8 minutes ago, Sypran said:

From all I'm reading this seems like 1% Intel's fault (The TDP being only slightly higher is flat out misleading. It's a significantly hotter chip, and everyone knows it.)

 

But 99% on Apple for: A) not giving this chip a beefier heatsink/more aggressive fan curve, or B) flat out offering this chip as an option.

I doubt there would be any hubbub if they didn't offer an i9 config. I don't even think anyone would have even noticed. This isn't at all like the max 16gb of ram thing.

 

I have to imagine Apple did testing on a machine with an i9 config. Even just to get a nice "it's 30% faster then the 20__ model!" And saw that it gets too hot.

Could it be that Apple tested the i9 MacBook Pro on an industrial cold room with temperatures as low as below 0^C?

 

[XenosTech]

22 minutes ago, mynameisjuan said:

Wow the people in this thread that say its intels fault. Bullshit. 

 

Plenty of laptops have the same chip and dont thermal because their actually test their coolers and make sure its up to par. Its a tested CPU, gives off this much heat at this clock, draws this much power. Not fucking hard to test Apple.

 

Im sorry but anyone who thinks this is in any way intels fault is mental. And dont give me the "but they should have soldered it" because that is not the point.

Shhhh you're not allowed to use logic here. Plus this is a chip that has no IHS so it's direct die cooling 

[Arika]

5 minutes ago, captain_to_fire said:

Could it be that Apple tested the i9 MacBook Pro on an industrial cold room with temperatures as low as below 0^C?

 

hey don't give them ideas to release an iFridge.

 

want to plug in all your stuff to your new laptop? buy this dongle

want your brand new i9 macbook pro to not throttle? buy this fridge

[captain_to_fire]

7 minutes ago, Arika S said:

hey don't give them ideas to release an iFridge.

 

want to plug in all your stuff to your new laptop? buy this dongle

want your brand new i9 macbook pro to not throttle? buy this fridge

Carrier will manufacture it for Apple but without the Carrier branding 

[D13H4RD]

2 hours ago, mynameisjuan said:

Plenty of laptops have the same chip and dont thermal because their actually test their coolers and make sure its up to par. Its a tested CPU, gives off this much heat at this clock, draws this much power. Not fucking hard to test Apple.

There are a couple with issues that have the same i9 8950HK

 

• Dell XPS15
• Alienware 15 R4
• Eurocom Q8

[XenosTech]

41 minutes ago, D13H4RD2L1V3 said:

There are a couple with issues that have the same i9 8950HK

 

• Dell XPS15
• Alienware 15 R4
• Eurocom Q8

Out of those 3, one brand is known for over heating out the ass.

[mark_cameron]

On 19/07/2018 at 11:45 AM, captain_to_fire said:

Source: 9to5 Mac

Download the Intel Power Gadget for PC, Mac and Linux https://software.intel.com/en-us/articles/intel-power-gadget-20

 

I didn't know until now that when the CPU goes to just 800 MHz it's not throttling but going to idle. So the dips in the middle is not throttling but some cores are going to idle. Is this some kind of power saving feature or does the CPU knows when cooling solutions suck so it resorts to some cores going to idle?

I'd say the result between the simulated quad core and the six core MBP is rather marginal when it comes to FCP X and 18 seconds is not significant in my opinion. The graph shows that even with four cores, it still throttles although not as pronounced as the six cores.

This is obvious that the freezer cooled MacBook Pro yielded with better results than the air cooled six core and quad core MacBook Pro and all CPU cores don't go to idle. Oddly enough, the iMac Pro with Intel Xeon processors performs slower than the MacBook Pro when it comes to Final Cut Pro X because of the fact that Xeon processors doesn't have a dedicated Quick Sync core unlike other Intel's consumer desktop and mobile processors (including the core m3). So all of that talk about Apple's magical optimization of FCP X often touted by Apple's fans users is just Quick Sync.

 

 

Then the last few words from the author is somewhat misleading and disingenuous in my opinion especially the one in bold.

I don't think Apple engineers are oblivious of the consequences of putting something as hungry as a core i9 to their laptop while retaining their skinny design and putting shitty cooling solutions. And yet the 9to5 Mac author tries to shift the blame towards Intel? This is some not-so subtle shilling from Jeff Benjamin and I don't think bringing up the rumored ARM powered MacBook would prove his arguments that Intel is somehow responsible for the throttling issues  of the MBP and I don't think Intel's alleged thermal issues (according to Benjamin) is the reason why Apple might transition to their in-house ARM chips for the 2019 MacBooks.

Intel didn't put the i9 chip into the MacBook. 

 

Apple did.

[JoostinOnline]

14 hours ago, D13H4RD2L1V3 said:

The closest actual thing I found was that Reddit article I linked above. 

 

At 3.8GHz, the CPU alone draws between 105-115W of power under load according to HWInfo. I wished I had one to test at 4.8

HWInfo can't accurately detect power draw.   You also seem to fail to realize that not all the cores boost to 4.8GHz.  It probably uses a lot less power.

 

Regardless, the whole system is definitely using less than 87W.

[D13H4RD]

11 hours ago, JoostinOnline said:

You also seem to fail to realize that not all the cores boost to 4.8GHz. 

It's more like 4.1GHz when all 6 cores are active. 4.8GHz is when a single core is active or overclocked. That I should have mentioned earlier. 

 

However, I doubt that the MBP even allows an OC in the first place. 

 

Something I found interesting is that the i9-powered Zenbook Pro with the 1050 Ti (slightly lower TDP compared to the Radeon 560X) in total consumes around 90-100W in average under heavy load from the adapter according to NBCheck, which is par for the course for a 45W CPU + 65/70W GPU, though I would assume it is maintaining stock speeds as I don't think the Zenbook's chassis allows turbo frequencies. 

 

(UPDATE: It's apparently 4.3GHz max with all 6 cores active at stock settings) 

[BlueChinchillaEatingDorito]

Thermal throttling out of the box isn't new for Apple. We saw this with what was it? The 5K iMac with the K series chip? 

[captain_to_fire]

So basically Jonathan is arguing that it's Adobe's fault that's why the i9 MBP thermal throttles? What do you think @D13H4RD2L1V3?

For whatever reason, he seems to be an Apple apologist by saying Dave2D used the wrong program and the wrong codec that's why the i9 MBP sucked.

[D13H4RD]

45 minutes ago, captain_to_fire said:

~snip~

I appreciate John's thorough tests. However, I don't think this changes much of the narrative because while it does show that the i9 is significantly better in some scenarios, there's one thing that he left out. 

 

The clockspeed. Most of the complaints aren't stemming from how the i9 is slower or only a tad faster in times than the i7. It's due to how the CPU is unable to maintain its base clockspeed due to how aggressive the thermal throttling is. Given how there's so many on that, I'm surprised that John didn't even check that. 

 

And I'm not especially keen on his statement that the problem is Dave2D using Premiere Pro rather than FCPX. Yes, FCPX is much faster due to the use of QuickSync and it's my preferred choice if I have to use a Mac for video editing on a professional level. But saying that the results are skewed because he used Premiere kinda reeks of the whole "You're using it wrong" thing.