(2019) 16" is HOT & NOISY with an external monitor! :(

[bill-p]

This is not a constructive post. Do you have a solution, other than to tell us to simply stick with clamshell mode?

And, to further address your comment that the people who have contributed to this thread are simply "crying" without attempting to find a solution: you will note that I, and others, have posted our observation that adjusting the refresh rate of an external display connected to our MBPs while in clamshell mode may result in a drop in GPU power draw. This is, in fact, a potential solution to a particular problem which many people are apparently experiencing. It also serves to show that there must be something wrong in the drivers and this observation may well lead to a bug fix. So, no, we are not simply continuing to "cry" without trying to find a solution.

I have posted multiple times that if you are experiencing high fan speed for no reason at all, you should try resetting the SMC 3-4 times.

I have also posted that Macs Fan Control (and like TG Pro and many other fan control apps) caused my fans to be stuck at max speed until I reset the SMC 3-4 times and uninstalled the app. This happened even when I was using Macs Fan Control under Bootcamp. So clearly, something in the SMC is not happy with 3rd-party software.

And I have suggested multiple times that it's not refresh rate. My 16" connected to clamshell with Apple's AV adapter or a USB-C/Thunderbolt 3 display does not require me to adjust refresh rate. That means there is something wrong with the dock or adapter that some people are using in clamshell mode.

But... even beyond that, I think it has been long established that the "only" surefire solution to this problem right now is exactly what it is: you absolutely have to run in clamshell mode with the right displays.

Some of us already know it's a driver issue. This actually affects many more machines with AMD graphics than you may think. Some previous 15" MacBooks (like the 2019) are also reportedly affected. If that's indeed the case and there is still no "fix" coming from Apple now, it seems likely that the fix has to come from AMD, not Apple. I have only ever seen AMD drivers getting an update with a major MacOS revision (i.e.: we have to wait for that next OS after Catalina). This issue might as well be a design issue at this point. And if so, you then have two solutions:

1. Return the 16" and get a Windows computer.
2. Get a 13" with eGPU. This was my previous setup.

It's not that I'm not offering possible solutions. It's simply just that you and many others here do not want to consider those solutions at all. But... even beyond that, that post was more in response to people calling me out when I said that "this issue does not affect everyone".

 

[crouch]

It does affect everyone who doesnt want to give up the user of they internal display, keyboard, trackpad, and touchid. To buy another external display, keyboard and trackpad is expensive, and there is no option to buy external touchid. Just because you are happy to live with these limitations does not negate the existence of the problem which affects every mbp 16".

I didn't give up any of those things. I'm running the base i9 16" model open (not clamshell) connected to a Dell U2720Q via USB-C with very minimal fan noise over extended amounts of time.

This is clearly not affecting everyone.

And look, I was furious with my laptop when I received it -- kernel panics out the wazoo. I know it sucks trying to fix something that doesn't work like you expect it to, with people telling you, "Nope, not a problem for me, must be your machine!" but that's the reality: it affects some and doesn't others. Your pain and your situation are not universal.

For me, using the MBP 16" open and dual-screened (laptop screen running *AND* Dell U2720Q running) works without a hitch, as it does others.

 

[jc_9]

I didn't give up any of those things. I'm running the base i9 16" model open (not clamshell) connected to a Dell U2720Q via USB-C with very minimal fan noise over extended amounts of time.

This is clearly not affecting everyone.

And look, I was furious with my laptop when I received it -- kernel panics out the wazoo. I know it sucks trying to fix something that doesn't work like you expect it to, with people telling you, "Nope, not a problem for me, must be your machine!" but that's the reality: it affects some and doesn't others. Your pain and your situation are not universal.

For me, using the MBP 16" open and dual-screened (laptop screen running *AND* Dell U2720Q running) works without a hitch, as it does others.

Can you confirm the wattage of your dGPU in your scenario? (You can use iStats Menu: https://bjango.com/mac/istatmenus/) I thought that everyone using atleast two screens had their dGPU running at 18W+

 

[Camarillo Brillo]

My guess is he's wearing a lot of layers to stay warm in the room he's using the computer in ?

 

[joelhinch]

I didn't give up any of those things. I'm running the base i9 16" model open (not clamshell) connected to a Dell U2720Q via USB-C with very minimal fan noise over extended amounts of time.

This is clearly not affecting everyone.

And look, I was furious with my laptop when I received it -- kernel panics out the wazoo. I know it sucks trying to fix something that doesn't work like you expect it to, with people telling you, "Nope, not a problem for me, must be your machine!" but that's the reality: it affects some and doesn't others. Your pain and your situation are not universal.

For me, using the MBP 16" open and dual-screened (laptop screen running *AND* Dell U2720Q running) works without a hitch, as it does others.

Please connect an external display and open your MacBook Pro lid at the same time. Now open iStat Menus and show us the wattage. It will be 18+w Radeon High Side. Yep the issue affects you too. Fan noise isn’t always the issue. It’s excessive wattage.

 

[HarryPot]

My guess is he's wearing a lot of layers to stay warm in the room he's using the computer in ?

It could be.

When I use my MBP with the AC on, the idle temperature goes down around 10C.
My room without AC is around 80F, with AC is around 68F.

And those extra 10C can very well be the difference between getting fan noise or not.

 

[mrmachine79]

I didn't give up any of those things. I'm running the base i9 16" model open (not clamshell) connected to a Dell U2720Q via USB-C with very minimal fan noise over extended amounts of time.

This is clearly not affecting everyone.

And look, I was furious with my laptop when I received it -- kernel panics out the wazoo. I know it sucks trying to fix something that doesn't work like you expect it to, with people telling you, "Nope, not a problem for me, must be your machine!" but that's the reality: it affects some and doesn't others. Your pain and your situation are not universal.

For me, using the MBP 16" open and dual-screened (laptop screen running *AND* Dell U2720Q running) works without a hitch, as it does others.

Let me rephrase. Instead of "everybody is affected", consider "every 16 inch MacBook Pro is affected". Unless I have missed something, not a single person has said (or shown with a screenshot) that their 16 inch MacBook Pro does not use ~20W with an external display connected and the lid open. Not one, in 1500+ messages in this thread.

Reports of "works for me, I don't notice heat or fan noise for [my particular work load]" is subjective and anecdotal and does not help at all. These reports give prospective users the impression that this issue is a defect that affects *some machines*, when in fact it appears to affect *all machines* but only *bothers some users*.

I do not believe that *anyone* purchasing a 16 inch MacBook Pro today has a "chance of getting a good one" (that doesn't use ~20W for dGPU with external display connected and lid open), which is what you are implying may be possible.

This is where an objective repeatable test (benchmarks and stress tests) is helpful, even if not representative of *your* particular workload.

It seems obvious that if a machine is producing 15-20W more heat for dGPU, it can cool 15-20W less heat generated by the CPU. If it's consuming 15-20W more power for dGPU, it has 15-20W less power available to give to the CPU.

Try the tests I describe in this post:

(2019) 16" is HOT & NOISY with an external monitor! :(

Sorry, but this is just not true. It may be your experience, with your chosen eGPU and your chosen graphics card, but it is not mine. The Blackmagic eGPU is reported to have a noise level of 18dB. Only audible with your ear right next to it. I just got a Razer Core X Chroma and Sapphire Pulse...

forums.macrumors.com

Basically, Cinebench on loop to generate CPU load, and then watch the impact in Intel Power Gadget with varying degrees of dGPU load (idle, video chat, benchmark).

These tests show CPU power dropping from 47-50W (eGPU idle) down to 34-37W (dGPU idle), and down to 26-30W with the dGPU doing a single 1-1 video chat in Safari. Frequency drops from 2.7-2.9GHz down to 1.9-2.1GHz.

And those test results were on a 15" MBP which does NOT have the same problem as the 16". The 15" uses high power (~15W) for dGPU at certain scaled resolutions. The 16" MBP reportedly uses even higher power (~20W) for dGPU at ANY resolution with the lid open.

 

[MrGimper]

Let me rephrase. Instead of "everybody is affected", consider "every 16 inch MacBook Pro is affected". Unless I have missed something, not a single person has said (or shown with a screenshot) that their 16 inch MacBook Pro does not use ~20W with an external display connected and the lid open. Not one, in 1500+ messages in this thread.

Reports of "works for me, I don't notice heat or fan noise for [my particular work load]" is subjective and anecdotal and does not help at all. These reports give prospective users the impression that this issue is a defect that affects *some machines*, when in fact it appears to affect *all machines* but only *bothers some users*.

I do not believe that *anyone* purchasing a 16 inch MacBook Pro today has a "chance of getting a good one" (that doesn't use ~20W for dGPU with external display connected and lid open), which is what you are implying may be possible.

This is where an objective repeatable test (benchmarks and stress tests) is helpful, even if not representative of *your* particular workload.

It seems obvious that if a machine is producing 15-20W more heat for dGPU, it can cool 15-20W less heat generated by the CPU. If it's consuming 15-20W more power for dGPU, it has 15-20W less power available to give to the CPU.

Try the tests I describe in this post:

(2019) 16" is HOT & NOISY with an external monitor! :(

Sorry, but this is just not true. It may be your experience, with your chosen eGPU and your chosen graphics card, but it is not mine. The Blackmagic eGPU is reported to have a noise level of 18dB. Only audible with your ear right next to it. I just got a Razer Core X Chroma and Sapphire Pulse...

forums.macrumors.com

Basically, Cinebench on loop to generate CPU load, and then watch the impact in Intel Power Gadget with varying degrees of dGPU load (idle, video chat, benchmark).

These tests show CPU power dropping from 47-50W (eGPU idle) down to 34-37W (dGPU idle), and down to 26-30W with the dGPU doing a single 1-1 video chat in Safari. Frequency drops from 2.7-2.9GHz down to 1.9-2.1GHz.

And those test results were on a 15" MBP which does NOT have the same problem as the 16". The 15" uses high power (~15W) for dGPU at certain scaled resolutions. The 16" MBP reportedly uses even higher power (~20W) for dGPU at ANY resolution with the lid open.

Is it a defect tho?

 

[joelhinch]

Is it a defect tho?

iGPU doesn’t need 20w to run 2 displays at 4k60p. So why should a dGPU 4x better require it? It should be a walk in the park for the dGPU.

 

[mrmachine79]

Is it a defect tho?

It appears to be a design limitation (a combined CPU/dGPU and general system cooling capability that is unable to adequately cool *both* CPU and dGPU under load to maintain the *base* CPU frequency) and/or possibly a driver issue causing excessive power draw that is not advertised with the features and tech specs. I suspect that the purchase decision would be different for many people, had they known that the claim:

Simultaneously supports full native resolution on the built-in display in millions of colours and:

• Up to two displays with 6016‑by‑3384 resolution at 60Hz in over a billion colours
• Up to four displays with 4096‑by‑2304 resolution at 60Hz in over a billion colours

should have been listed with the caveat "use of ANY external display at ANY resolution will significantly increase heat and noise, and significantly reduce CPU performance, even to below the advertised base frequency". As it stands, the claims (above) made by Apple in their tech specs page are misleading or deceptive.

Also, it seems that while dGPU usage in all Mac notebooks have used additional power, and taken away power from the CPU, the 16 inch MBP appears to be the first time that the additional power usage has been so high at idle, and so high regardless of the resolution on the external display. For example, a 2019 15 inch MBP uses ~6W at low-DPI 2560x1440, ~12W at native 4K and ~15W at scaled "looks like 2560x1440". The 2019 16 inch MBP appears to use ~20W at all resolutions.

 

[bill-p]

It's worth pointing out that my 16" never drops below base frequencies under sustained load, and yes, this is even after 30 minutes of looping Cinebench. The sustained CPU speed stays above base.

If the lid is open, dGPU draws 20W, CPU is limited to around 65-70W sustained power and gets around 2.4 - 2.7GHz. That's still above base. Base is around 2.3GHz.

If the lid is closed, dGPU draws 5W, CPU is able to go up to 85-90W sustained power and gets around 3.0 - 3.2GHz. That's far above base.

I think you're running into thermal limitations with your 15". I used to have the 2018 15" so I know the machine hits its thermal limits pretty soon. The 16" MacBook doesn't seem to run into thermal limits that easily, but it seems there is a hard ceiling (about 100W) that limits how much power the MacBook can draw in a sustained load environment.

I have only ever seen my 16" drop below base frequencies when it's idling. Otherwise sustained load performance is nothing short of amazing for a MacBook.

 

[mrmachine79]

It's worth pointing out that my 16" never drops below base frequencies under sustained load, and yes, this is even after 30 minutes of looping Cinebench. The sustained CPU speed stays above base.

If the lid is open, dGPU draws 20W, CPU is limited to around 65-70W sustained power and gets around 2.4 - 2.7GHz. That's still above base. Base is around 2.3GHz.

If the lid is closed, dGPU draws 5W, CPU is able to go up to 85-90W sustained power and gets around 3.0 - 3.2GHz. That's far above base.

I think you're running into thermal limitations with your 15". I used to have the 2018 15" so I know the machine hits its thermal limits pretty soon. The 16" MacBook doesn't seem to run into thermal limits that easily, but it seems there is a hard ceiling (about 100W) that limits how much power the MacBook can draw in a sustained load environment.

I have only ever seen my 16" drop below base frequencies when it's idling. Otherwise sustained load performance is nothing short of amazing for a MacBook.

Yes, there is supposedly ~12W better cooling capacity in the 16 inch and I have seen that play out in benchmarks reported online. I am going to re-run my tests on a 16 inch as soon as I receive it. I am hoping that the improved cooling capacity is enough to eliminate the crippling performance issues I have seen on my 15 inch.

However, I believe that CPU performance will still be significantly impacted, and that this is a problem serious enough that Apple's claims are misleading or deceptive. That 15-20W we are forced to use for dGPU has to result in heat, noise, and less power for CPU.

I would like to see your Cinebench and Intel Power Gadget test results, though. I believe you are correct in that there is a total system power limit of 96-100W (on AC power or battery), but I have never heard of or seen the CPU sustain more than 62-67W in a 16 inch MBP. You are saying the CPU is able to go up to 85-90W sustained power.

Ideally:

- Install latest Catalina on an external SSD with no other apps or extensions running
- Connect an external display
- Run Cinebench on loop for at least 10 passes and watch Intel Power Gadget
- Take a screenshot just before the last pass finishes (which will dip the Intel Power Gadget numbers while the test restarts)
- Do the same with a 1-1 Discord video chat in Safari to simulate a light dGPU load (for an apples to apples comparison with my earlier test results)
- Do the same again with Luxmark (with only dGPU selected) to simulate a heavy dGPU load.

If it is too much hassle to run these tests on a fresh install of Catalina, you can still run them in your everyday environment. The important thing to note is not the benchmark result, but the sustained CPU power draw and frequency.

 

[Tnanman]

I have posted multiple times that if you are experiencing high fan speed for no reason at all, you should try resetting the SMC 3-4 times.

I have also posted that Macs Fan Control (and like TG Pro and many other fan control apps) caused my fans to be stuck at max speed until I reset the SMC 3-4 times and uninstalled the app. This happened even when I was using Macs Fan Control under Bootcamp. So clearly, something in the SMC is not happy with 3rd-party software.

And I have suggested multiple times that it's not refresh rate. My 16" connected to clamshell with Apple's AV adapter or a USB-C/Thunderbolt 3 display does not require me to adjust refresh rate. That means there is something wrong with the dock or adapter that some people are using in clamshell mode.

But... even beyond that, I think it has been long established that the "only" surefire solution to this problem right now is exactly what it is: you absolutely have to run in clamshell mode with the right displays.

Some of us already know it's a driver issue. This actually affects many more machines with AMD graphics than you may think. Some previous 15" MacBooks (like the 2019) are also reportedly affected. If that's indeed the case and there is still no "fix" coming from Apple now, it seems likely that the fix has to come from AMD, not Apple. I have only ever seen AMD drivers getting an update with a major MacOS revision (i.e.: we have to wait for that next OS after Catalina). This issue might as well be a design issue at this point. And if so, you then have two solutions:

1. Return the 16" and get a Windows computer.
2. Get a 13" with eGPU. This was my previous setup.

It's not that I'm not offering possible solutions. It's simply just that you and many others here do not want to consider those solutions at all. But... even beyond that, that post was more in response to people calling me out when I said that "this issue does not affect everyone".

Your solution isn't even a solution.

It didn't work? Get rid of it!

I've never seen so many words written to really say nothing that helps anyone in this thread.

 

[jc_9]

I'm returning mine.

Apple called me back to tell me that the only way to make their engineers aware is to either send a detailed report at www.apple.com/feedback or send the macbook for repair at an Apple autorised repair store (since Apple stores are still closed during the pandemic).

The issue is present on ALL macbook pro 16". Connect it to an external display while keeping the lid open and its dGPU automatically consumes 18W+ on idle. This is half of the whole laptop wattage on idle. Unacceptable.

 

[bill-p]

I would like to see your Cinebench and Intel Power Gadget test results, though. I believe you are correct in that there is a total system power limit of 96-100W (on AC power or battery), but I have never heard of or seen the CPU sustain more than 62-67W in a 16 inch MBP. You are saying the CPU is able to go up to 85-90W sustained power.

I hope iStat works for you.

Here's what I did: I shut the MacBook down, let it cool for 10 minutes to make sure everything is swell before I started the computer up and ran Cinebench. Here's what happens with the lid open (you can see Radeon High Side is 20W)

And then what happens if I just close the lid:

Immediately, the MacBook jumps up to 78W, corresponding to a 13W increase in performance headroom.

Some runs later with the lid closed:

But you are right. In constant sustained heavy load (all 8 cores stressed at 100% for more than 1 hour straight), 65W is what I'm seeing eventually. It translates to about 3.00GHz. My room has been pretty toasty recently, so I'm also starting to see 2.90GHz. It seems the CPU starts to drop frequencies when it reaches 90C, regardless of power draw.

With the lid open, and after 30 minutes of sustained heavy load, frequencies will drop steadily until about 2.50GHz. Admittedly, I have seen that, and it happens much faster than with the lid closed.

Your solution isn't even a solution.

It didn't work? Get rid of it!

I've never seen so many words written to really say nothing that helps anyone in this thread.

Then what solution, pray tell, do you have to solve everyone's problems when the lid is open?

Please note, once again, that when the lid is closed, there are those of us who do not see any problem at all.

 

[mrmachine79]

I hope iStat works for you.

Thanks for that. Intel Power Gadget would be better, as it shows CPU power/PKG and frequency both as a graph alongside temp and utilisation. It's hard to tell from your screenshots, some taken shortly after a new Cinebench pass has started where the CPU may be bursting after a dip, is really stable and sustaining the figures reported by iStat menus.

I don't know exactly what is included in "CPU Computing Highside" but I assume that is where you are getting your figures from (78W, 65W). But that does not match up with what Intel Power Gadget reports for CPU Power/PKG.

Your screenshots do at least show a drop in available power and frequency with the lid open.

However, you did not do any tests to show the drop in CPU performance (under load) with concurrent light and heavy GPU load (e.g. Discord 1-1 video chat and Luxmark benchmark with only dGPU selected).

 

[bill-p]

I'm not sure why a flat graph is so interesting, but... okay, here you go:

I think you'll get your 16" soon, so you'll most likely be able to tell for yourself how the rest of the other scenarios play out.

I think Intel Power Gadget lies about CPU power consumption, which would be consistent with Intel's M.O. as they have tried to "blind" the world to the fact that their CPUs have stayed on the same manufacturing process for 4 years now. Or otherwise, I don't see how 45-50W reported in Intel Power Gadget + 20W for dGPU translates to >90W power draw. Or if iStat is "over-reporting" power consumption now, we do have to question whether the 20W GPU power draw is even accurate (kinda brings this whole thread into question then). I'd err on the side of iStat here.

But beside the point, all of that is not to show that I have a Godly 16" MacBook that can draw more power than any other 16". It's just to say... I have not seen CPU frequencies drop below base clocks yet, with my own usage. I'm pretty certain that you can probably devise a usage where both the dGPU and the CPU need to be 100% stressed at all times, and sure, something will have to give. The 100W max power consumption is actually even more daunting then given the discrepancy between what Intel Power Gadget reports and what iStat reports.

 

[_meds]

Looking at the AMD drivers it appears there's a 'tuning' option when you use the gaming drivers, unfortunately the latest drivers on bootcampdrivers.com does not have the 'tuning' option in the control panel.

The tuning options apparnetly have some stuff around memory clock speeds, I'm wondering if I can forcefully keep the memory clockspeed lower to keep the GPU from overheating. On Windows the latest radeon drivers are so buggy that it keeps the memory clock speed running at peak even when no external monitors are connected, it's utteraly ridiculous.

Trying with older drivers now will report results.

 

[matram]

What we see with the 5 - 18W jump is peculiar. It seems to indicate that the GPU has to do more than 3x the amount of work to drive the internal screen in combination with an external one.

I tried to read up on the RDNA architecture used in the "Navi" GPU. The architecture white paper says that to go beyond 4k, 60Hz requires Display Stream Compression (DSC).

Any GPU-guru here that could say if this (going beyond 4k with the two screens combined) could be an explanation?

Testing clock speed is interesting but I doubt it could have a major effect?

 

[wegster]

Your solution isn't even a solution.

It didn't work? Get rid of it!

I've never seen so many words written to really say nothing that helps anyone in this thread.

Guys - lay off. What Bill has said isn't incorrect (not an issue in clamshell), although I also thought he was originally saying 'not an issue for anyone.' He's added info into the thread, versus the 20th time hearing someone say 'my fans aren't bad' which completely ignores the actual and real issue (connect external display = unreasonable GPU power consumption, which can cause possible CPU throttling as well as excessive heat and 'wear' (e.g. battery longevity before replacement) over time).

I'd far rather see Bill P's posts versus another link to the video on fan controls, etc.

 

[wegster]

Your solution isn't even a solution.

It didn't work? Get rid of it!

I've never seen so many words written to really say nothing that helps anyone in this thread.

Actually, he provided information on part of the problem which is defining the extent of the problem.
He showed that lid closed has reasonable performance while lid open in similar/same conditions, CPU is throttled sooner although in his case is still above base CPU frequency.

That's pretty valuable for when Apple eventually decides to address, as well as to others who are considering - 'there is a problem; how much does it affect <them> for their use, with lid open or closed.'

Besides an eGPU, AMD or Apple are most likely the only ones able to provide a real 'solution,' if one is provided.
Right now, we're just trying to get clarity on the actual impact.

 

[covertsurfer]

Has Apple publically acknowledged this? I assume not but haven't seen anything to confirm/deny. If this is that much of an issue how long until class action law suit starts?

 

[Bearygoodfries1]

I wasn’t aware this was a problem...

Darn, then I guess using 2 4K USB-C monitors is going to be a problem? ?

 

[Platonist]

I have been following this thread as I am reasonably certain a 16" MBP is in my future. I use my current 15" (2016, i7, Radeon Pro 460) in both clamshell mode and open along with a 27" LG 4K monitor--depending on the task at hand. In clamshell mode, the GPU is drawing 17w, while drawing 32w when open using both the internal and external screens. Looking through this thread this seems pretty bad, and makes me think that the new 16" will be far superior in managing these things than my current 15".

Am I wrong in thinking this? I wonder if some of the commentary on the new 16" is really about optimization rather than a flaw in the design, given the difference between it and some older generations of MBP.

 

[wegster]

Has Apple publically acknowledged this? I assume not but haven't seen anything to confirm/deny. If this is that much of an issue how long until class action law suit starts?

Not publicly acknowledged, but see large post count topic on Apple forum.
File feedback or call in a case if under warrantee or AppleCare, or in 90 day call in support.
Hard to say where it goes.
If it’s a driver issue, hopefully Apple sees enough calls, cases and possibly returns to ‘motivate them’ into resolving sooner vs later or not at all.

If the do nothing, it may take time to see if the added heat impacts longer term component reliability, then an extended warranted or replacement program like retina batteries or the retina/flakegate or older Nvidea GPUs coverage.

Or lastly, if enough solid info is collected it’s possible someone goes after a non-suitability for purpose, false advertising or similar type claim or action. Maybe.