In terms of the dGPU driver issue causing kernel panics. Apparently engineers are working on a fix.
In terms of excessive power draw. They probably don’t care enough.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
(2019) 16" is HOT & NOISY with an external monitor! :(
- Thread starter Appledoesnotlisten
- Start date
- Sort by reaction score
In terms of excessive power draw. They probably don’t care enough.
Well this sucks. I ordered a refurb 16" then I saw these threads.
Personnally I can't accept that a new $2.4k+ laptop can't even handle a single monitor attached to it without the fans blasting at maximum speed. Especially when the page product mentions: "Supports up to two 6k displays".
I know some people have some success using third-party programs and different adapters but coming from a Windows environment, I thought paying the Apple tax meant no more thinkering for basic usage of the product.
I guess I should wait for the next 16" to come out, hopefully later this year? I just want a somewhat future-proof macOS 15"+ laptop with no keyboard / heat / noise issue when programming with a browser and photoshop opened.
I understand that unidirectional computation (eg. gaming) has less latency than bidirectional computation, but that’s true for both the dgpu and egpu. The question was regarding the difference between the egpu and dgpu performance—namely, how/why the stronger egpu can have worse performance in some cases than the weaker dgpu. Obviously bandwidth is the other key difference here, so I assume that is the culprit. But what I’m stuck on is that Xnorth said the bidirectional data is the reason the egpu is hitting the bandwidth cap. But with full duplex, as I understand it, bidirectional computation in and of itself should have no bearing on whether or not the cap is hit. In other words, even unidirectional computation (gaming) should be able to get faster results on the dgpu than the egpu, if the task is demanding enough to cause the egpu to hit the bandwidth cap, yet the dgpu is able to perform that task higher than that cap.
I’m a little more confused now by what you’re saying. Are you now saying bidirectional data isn’t causing it to max out the bandwidth? Because that was the part that tripped me up. Like I said to the person above (to quote myself), “with full duplex, as I understand it, bidirectional computation in and of itself should have no bearing on whether or not the cap is hit. In other words, even unidirectional computation (gaming) should be able to get faster results on the dgpu than the egpu, if the task is demanding enough to cause the egpu to hit the bandwidth cap, yet the dgpu is able to perform that task higher than that cap.”
By latency, do you just mean the 20% inefficiency loss (different from the bandwidth cap, as I understand it)? Because in regard to the egpu being an extra step/bottleneck, the dgpu is as well. The obvious difference being, as you said, the dgpu has a much higher bandwidth cap. So do you mean something other than the TB3 bandwidth cap or the 20% inefficiency loss when you say that the egpu reduces bandwidth?
The only thing about the 20% performance loss is it doesn’t in itself sufficiently explain why some tasks are better with dgpu and some are better with egpu—because even with that performance loss, isn’t a good egpu at least 20% more powerful than the mbp dgpu (in general, maybe it depends on the task)? So to me it always seems to go back to the thunderbolt 3 bandwidth cap as the prevailing reason. And again, it seems to me that bidirectional computation is irrelevant in that regard.
To recap, I didn’t realize that the dgpu can outperform a TB3 egpu in certain tasks. So thanks for that important tidbit. But in trying to understand the whys/hows/whens, from what I can tell and make sense of, the most logical conclusion just seems to be that certain demanding tasks are able to push the dgpu past the point where the egpu gets capped by the thunderbolt 3 bandwidth limit (whether uni or bi directional computation).
Apologies to others if this is too far off topic.
Likely doing spotlight indexing and other new maintenance tasks..give it a few days after restore, migration or app installs.
Hello everyone, I posted the following over at the Apple Support Communities but I'm reposting it here in case anyone may be interested...
-----------------------------------
This is not a solution. I am presenting my observations when using it with an external 4K monitor, and it is intended for those of us who decided to stick with their 16-inch and live with it for the time being, and just want to hear how others are doing.
My setup:
MacBookPro (16-inch, 2019) / 2.6GHz 6-Core Intel Core i7 / 32GB RAM / AMD Radeon Pro 5300M 4GB
External 4K monitor (LG 27UD68-P)
The resolution of the MBP is set to Looks like 1536 x 960.
The resolution of the external monitor is set to Looks like 1920 x 1080.
Connection: 3rd party USB-C to DisplayPort cable on the left side of the MBP (more on this later)
(I also have a USB-C to HDMI adapter but it didn't make a noticeable difference.)
We all know that many of our MBP exhibits this behavior where the RPM of the fans tend to increase (and the GPU uses more watts) when using it with an external monitor, unless used in clamshell mode.
The following graph from iStat shows the speed of my MBP's left fan under different conditions, when using both the internal monitor and an external 4K monitor.
While it is pretty self-explanatory, I want to point out the following observations:
I am attaching the sensor readings from iStat in case anyone wants to compare it with their own.
This is at point [2] in the previous image, where the system was pretty much idle.
If you have any questions, please let me know.
-----------------------------------
This is not a solution. I am presenting my observations when using it with an external 4K monitor, and it is intended for those of us who decided to stick with their 16-inch and live with it for the time being, and just want to hear how others are doing.
My setup:
MacBookPro (16-inch, 2019) / 2.6GHz 6-Core Intel Core i7 / 32GB RAM / AMD Radeon Pro 5300M 4GB
External 4K monitor (LG 27UD68-P)
The resolution of the MBP is set to Looks like 1536 x 960.
The resolution of the external monitor is set to Looks like 1920 x 1080.
Connection: 3rd party USB-C to DisplayPort cable on the left side of the MBP (more on this later)
(I also have a USB-C to HDMI adapter but it didn't make a noticeable difference.)
We all know that many of our MBP exhibits this behavior where the RPM of the fans tend to increase (and the GPU uses more watts) when using it with an external monitor, unless used in clamshell mode.
The following graph from iStat shows the speed of my MBP's left fan under different conditions, when using both the internal monitor and an external 4K monitor.
While it is pretty self-explanatory, I want to point out the following observations:
- When the MBP is idle (the flat slope right before [2] in the image), the fans spin at about 2300RPM, and personally I think that this is acceptable.
- Even when doing light web browsing, it is under 3000RPM.
- Watching a Youtube video (the Hermitage video on Apple's channel) at point [5] increases the fan speed but then it starts to stabilize at around 3300RPM (subjectively speaking, this is when I start to notice the sound of the fans).
- When editing/playing back footage, the fans spin at about 3000~4000RPM.
- When analyzing/exporting footage, the fans spin at about 4000~5000RPM.
- If I do the same things in clamshell mode, these speeds are roughly reduced by 1000RPM.
- However, the task of analyzing/exporting a video footage is a computationally expensive task, and it will nevertheless hit 5000RPM even in clamshell mode.
I am attaching the sensor readings from iStat in case anyone wants to compare it with their own.
This is at point [2] in the previous image, where the system was pretty much idle.
If you have any questions, please let me know.
Have a question for you guys since I think in here there's a lot of knowledge in this area. I got the 16inch MPB and it's fine with my current Dell UltraSharp USB-C 1080p monitor. I want an Ultrawide monitor now with higher resolution. USB-C versions are expensive, should I fork out the extra for USB-C or is USB-C to HDMI fine? Something around £500, preferably less though, is what I'm after.
I'm thinking to go clamshell with an Ultrawide 34inch if Citrix supports that. Not a fan of the gap between laptop and monitor screen, but the 16inch screen on its own is amazing. Love it, really beautiful.
I'm thinking to go clamshell with an Ultrawide 34inch if Citrix supports that. Not a fan of the gap between laptop and monitor screen, but the 16inch screen on its own is amazing. Love it, really beautiful.
correct, there is a dedicated bonded lane for tx/rx on the TB3 link which makes it 'full duplex' so saturating one does not affect the other
however the TB3 link is essentially a 4x PCIE 3.0 connection. the dgpu is 8x/16x PCIE. it's far easier to saturate the 40Gbps TB3 link (especially with something like a Radeon VII on the end) and probably near impossible to saturate a 16X connection with a 5500M. and then add the latency of traveling over a cord like i mentioned earlier
Given that there's @XNorth here and another friend of mine with exactly the same model as mine reporting theirs as 'working fine', I've naively ordered a replacement.
MacBook Pro (16-inch, 2019)
2.3 GHz 8-Core Intel Core i9
16 GB 2667 MHz DDR4
AMD Radeon Pro 5500M 4 GB
I didn't see any evidence of anyone on here reporting that they had theirs replaced and it magically started working properly, though. So I'm not holding my breath!
MacBook Pro (16-inch, 2019)
2.3 GHz 8-Core Intel Core i9
16 GB 2667 MHz DDR4
AMD Radeon Pro 5500M 4 GB
I didn't see any evidence of anyone on here reporting that they had theirs replaced and it magically started working properly, though. So I'm not holding my breath!
Hey all! First time poster and sitting on an older 2015 Macbook Pro. Am very close to pulling the trigger on a new 16" Pro but I will need the new 16" Pro to drive two external 49" ultra wide monitors, so this issue is making me hesitate on the upgrade at the moment.
Has this link ever been posted in this thread yet?
How to FIX your 16” MacBook Pro - (Don’t Return it)
Explains why the fans run much faster and often than in the older 2015 Macbook Pro. Basically the 16" Pro is designed to run much cooler than the older 15" Macbook Pro, which results in the significant increased fan usage. There are software tools available to to override this feature and configure it to run the same as the 2015 Macbook Pro.
I don't think this itself explains the crazy wattage increases from the dGPU when using external monitors though, but thought some might find it helpful.
Has this link ever been posted in this thread yet?
How to FIX your 16” MacBook Pro - (Don’t Return it)
Explains why the fans run much faster and often than in the older 2015 Macbook Pro. Basically the 16" Pro is designed to run much cooler than the older 15" Macbook Pro, which results in the significant increased fan usage. There are software tools available to to override this feature and configure it to run the same as the 2015 Macbook Pro.
I don't think this itself explains the crazy wattage increases from the dGPU when using external monitors though, but thought some might find it helpful.
Yes it was posted, and rated trash as it should be. He's curing the symptoms and not the causes, which is basically the same as taking cough syrup and continue to ignore the fact you got COVID-19.
If you're planning to use internal + external, you might want to hold on the purchase because it always pulls 18W no matter what. If you're using external only, you might want to use a direct USB-C > Displayport cable and hope your monitor doesn't play strange. Still, even in the best case scenario, (5W dGPU) you'll be always on the verge of fans spinning up during light workloads, with your internals hovering around 60°C and your batteries at 35°C+.
macOS Catalina 10.15.5 Release Notes | Apple Developer Documentation
Update your apps to use new features, and test your apps against API changes.
Resolved Issues :
- Large data transfers to RAID volumes no longer cause Finder to become unresponsive. (61307708)
Not holding my breath, how can Apple stay silent for this long? The issue was also first reported last November on discussions.apple.com
macOS Catalina 10.15.5 Release Notes | Apple Developer Documentation
Update your apps to use new features, and test your apps against API changes.developer.apple.com
Resolved Issues :
- Large data transfers to RAID volumes no longer cause Finder to become unresponsive. (61307708)
Not holding my breath, how can Apple stay silent for this long? The issue was also first reported last November on discussions.apple.com
Did you see this post in that forum? I'll quote it below and it's not my post:
Yes I saw the post, however the kernel panic is related to the 10.15.4 update and the overpower draw from the GPU is there since the beginning.
Love me some apples
Last edited:
I find it interesting that you are experiencing the issue on the i7/5300m model. So appears to not be cpu/gpu specific. Must be a driver issue surely?
On my LG 34UC79G connected via DisplayPort at either native resolution 2560x1080 or the slightly lower 2048x864 I get these "Radeon High Side" measurements when the lid is closed:
I can't explain why at 60 W it draws almost 4x as much power compared to the much more resource intensive 99.88 Hz.
At 1920x1080 the only options are 50 Hz and 60 Hz and both draw only 5 W.
But when the lid is open, it's 18W, always, in every configuration, in every resolution.
To compare, on my late 2016 MacBook Pro I get only 11 W with the lid open and external display at 120 Hz. And the fans run at their absolute minimum of ~2000 rpm, completely silent.
Massive step back considering that I only wanted to upgrade because the battery started to bulge on my old one!
| 144 Hz | 18 W |
| 120 Hz | 18 W |
| 99.88 Hz | 5 W |
| 74.88 Hz | 5 W |
| 60 Hz | 18 W |
| 59.88 Hz | 5 W |
| 50 Hz | 5 W |
I can't explain why at 60 W it draws almost 4x as much power compared to the much more resource intensive 99.88 Hz.
At 1920x1080 the only options are 50 Hz and 60 Hz and both draw only 5 W.
But when the lid is open, it's 18W, always, in every configuration, in every resolution.
To compare, on my late 2016 MacBook Pro I get only 11 W with the lid open and external display at 120 Hz. And the fans run at their absolute minimum of ~2000 rpm, completely silent.
Massive step back considering that I only wanted to upgrade because the battery started to bulge on my old one!
This may be a dumb question, but hear me out .... I assume the laptop screen switches off when in clamshell mode, but it there a way to keep the screen on with the lid shut? Trying to ascertain if it's the screen being on that causes the 18-20w or if it's the lid being open.
I don't want to make any speculations, but it looks like that is the case. According to another user in the Apple Support Communities, that person was told by CalDigit that Apple is aware of this issue and is working on a fix. Of course, we still can't party until such a fix is officially released.
On a side note, my late 2013 15-inch retina MBP (with GeForce GT 750M) ran the same 4K display just fine until High Sierra or something came along with the Metal updates and then using the display became a pain. While it's a complete different hardware, this makes me want to believe that the cause of the current issue is also software related.
I have a guess. USB-C monitors which charge MacBook pro, I suppose the fan noise won't happen with these monitors. What do you think?
As far as I know it doesn't matter what display, what resolution, what cable, if HDMI or USB C - nothing matters with the MBP 16", except if it is in clamshell - but even then the right refresh rate must be set. So in short:
MBP 16" with external monitor and not in clamshell = 18+ W
MBP 16" with external monitor in clamshell + the right settings/refresh rate = about 5W.
I think there is a case where someone used x2 LG 4k UltraFine monitors (daisy chained and in clamshell) and I wouldn't trigger the 18-20Watts but other than that it seems to make no difference with open lid.
Lol, this is so weird but inline with what I'm seeing. I'm not a hardware engineer, so I can't logically think through this...but it seems very silly to me.
I think this would be really good as a sticky at the beginning of this long thread.