Oh please. You haven't read 5% of this thread, otherwise you'd know there is no solution for some specific problems. I won't mention which, so that you can continue your obdurate condescension unabated... ya'll.
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
Oh please. You haven't read 5% of this thread, otherwise you'd know there is no solution for some specific problems. I won't mention which, so that you can continue your obdurate condescension unabated... ya'll.
I've noticed a bit of lag with turbo boost shut off... or I may be imagining it, not sure. Wish it wasn't necessary.
I do find leaving the laptop in clamshell vs running an external with the screen open to be helpful, though the wattage to the radeon ramps up quick if you do anything like watch video. But it starts at only 5 watts of draw vs 20 if the screen is open. We should just be able to use the iGPU but until apple let's us we have to choose the least bad option.
I left it in an elevator that allows air flow, helps a bit more.
I can live with the processors but... just let us turn off the gpu when plugged into a monitor!
I do find leaving the laptop in clamshell vs running an external with the screen open to be helpful, though the wattage to the radeon ramps up quick if you do anything like watch video. But it starts at only 5 watts of draw vs 20 if the screen is open. We should just be able to use the iGPU but until apple let's us we have to choose the least bad option.
I left it in an elevator that allows air flow, helps a bit more.
I can live with the processors but... just let us turn off the gpu when plugged into a monitor!
I feel like I've noticed this too, although it's a bit subjective. Just a sense of being not quite as "snappy". I felt the same when using TBS to avoid overheating on my old 2012 MacBook Pro (turns out it just needed new thermal paste - oops!)
As I said a few posts back, I'm having pretty good results setting Turbo Boost Switcher to auto activate once the fan hits a certain speed (I'm still experimenting with the threshold point). That seems to allow the CPU to boost up to max speed to maintain that extra responsiveness, then curtails it as the heat & load increases.[/QUOTE]
I'm pretty new here, but I've read the whole thread (god help me) and I do kind of see his point. There seem to be two cohorts here - a group who wants to agitate for some official fix/recall from Apple, and another who are just looking to see if/how they can make their machines work well enough that they can work with it.
I've kind of got a foot in each camp, but I can see how both sides would tend to annoy the other from time to time. If you're looking for tips and workarounds the tilting at windmills gets a bit tiresome. By the same token if you're trying to create a broad, consistent record of an unacceptable design fault to get Apple's attention, it's not helpful to have people coming in saying that actually it's working OK for them...
*shrug*
I reckon I'd like both, to be honest. I'd strongly welcome a proper fix, but in the meantime I'll gladly take any workaround I can find
That's a handy app. Had a bit of a play with it - seems to work as it suggests - can limit CPU power usage to whatever wattage you want. Works better I think than turning off turbo boost - single CPU scores on Geekbench are around double those when turbo boost is off even with a low power limit.
On a 5600M model initial testing just using the Intel power gadget app - All thread scalar test shows I can run the CPU at 100% at 27W on Volta with the Radeon around 10W (2 x 4K monitors + MBPro display) and fans are under 3000RPM.
As a bit of a comparison here are some Geekbench results. This is with the above display configuration - i.e. Radeon 5600M at around 10W - my machine is a 5600M with 2.3GHz CPU and 64GB of RAM.
No throttling - 1227 / 7331
Limit CPU to 27W - 1154 / 4721
Turbo boost off - 657 / 5002
So those of you using turbo boost switcher might want to give Volta a try.
Is this the app?
Website states that it doesn't work on machines 2016 and up
Volta - Ultimate power control for Mac
volta.garymathews.com
Is this the app?
Website states that it doesn't work on machines 2016 and up
Volta - Ultimate power control for Mac
Just says that for the undervolting. The power limit option still seems to work on later models.
Correct - adjusting the power usage worked - confirmed via the Intel Power Gadget app - can see the CPU power usage match the setting in Volta.
Yes I think it will be good in my case to limit fan noise in those situations where I want to ensure I have a quiet machine (even with the 5600M).
One slightly ironic thing I've just noticed with Turbo Boost Switcher is that it actually seems to cause CPU spikes every time it refreshes its sensor data. The first half of this Intel Power Gadget graph is with the TBS app running, the second half I switched it off. (There's some other noise in the graph because I opened a Chrome window to type this post!)
)
I've kind of got a foot in each camp, but I can see how both sides would tend to annoy the other from time to time. If you're looking for tips and workarounds the tilting at windmills gets a bit tiresome. By the same token if you're trying to create a broad, consistent record of an unacceptable design fault to get Apple's attention, it's not helpful to have people coming in saying that actually it's working OK for them...
*shrug*
I reckon I'd like both, to be honest. I'd strongly welcome a proper fix, but in the meantime I'll gladly take any workaround I can find
I think most owners would like both workaround tips and a fix to be fair. That’s certainly the case for me. Based on learnings on this thread I’ve had a full week now of working exclusively in clamshell mode, with turbo boost disabled using Turbo Switcher Pro. No issues with heat and fan noise in this way of working and the performance is fine for my usage (mainly MS Teams, MS Office apps, Chrome, Safari and Apple Music), so as a workaround I’m definitely satisfied with it. That doesn’t mean I won’t continue to chase Apple for a permanent fix as the heat and fan noise should not ramp up as aggressively as they do just when connecting an external display with the lid open.
You might need to set it higher than 2500. Boost doesn't only work when you do long running intensive tasks (when you possibly don't care about fan noise). It boosts all the time for any trivial tasks. Even a single core boosting can raise fan speeds significantly.
If your machine is often at (for example) 3500 rpm and it is annoying, then you put a limit at 2500 to keep fan noise down (max you can tolerate). You might find it just constantly cycles. Full speed burst cycle to over 3000 quickly for trivial tasks and then disable boost and slowly fans reduce then immediately repeat.
I find that tb switcher pro works well to disable boost when on battery (to save power and heat when it might be literally on you lap) and otherwise it works well if you set the max limit to a value that is HIGHER than what you typically hit under normal medium load usage. E.g. I set it to disable boost at 5000 (the max) and re enable when it drops below 4000.
This doesn't prevent the machine from ever hitting the max but it prevents it from staying there, while still allowing boost and also not cycling boost off and on all day long.
There's another tool that lets you set a different level of max boost (frequency and duration) but it requires you disable SIP. It works better by allowing you to set boost as high as you can tolerate (fan noise). The best option would be an app that lets you set a new max fan speed and have it dynamically balance CPU power to maintain that under load without cycling. But I don't think that exists.
I find that leaving boost disabled all the time does have a significant performance penalty. You might not notice it in actual use at the time but it's basically 33% slower across the board, not only when trying to render or compile something heavy on all cores. A minute waiting here a minute waiting there adds up to a lot of minutes across my day.
With boost enabled, I do notice often that "wow that finished already? That was quick". Basically many tasks I used to wait for on my old computers, I feel like I'm not waiting for at all now.
Yeah, absolutely - I kind of pulled that number out of the ether, mostly based on what would be quiet. I'll definitely experiment to find the sweet spot (as well as trialing Volta, which seems to come at the same idea from a slightly different angle).
From my initial playing around, I found I could keep it pretty calm and collected but still allowing spikes above 2.3Ghz fairly regularly at that 2500rpm threshold, but that was very light use.
I'd be keen to check it out if you can share? Not too fussed about disabling SIP - I've already had to for Volta anyway! 😁
Agreed.
All this has me flashing back to my early early PC days when I manually overclocked a 486DX50 up to 66Mhz - The POWER!!!!
(oh god I'm old...)
I think Volta is better, but it requires a custom SIP state.
TB switcher does not, and has auto modes for power state, fan speed and apps.
Setting the max power limit in Volta will prevent boosting to the max while still allowing some boosting.
The one that's even better but more advanced (CLI tool) is voltage shift. It also requires custom SIP and has no GUI. But it allows you to individually set short and long term boost. So you can still get max boost in short bursts but drop down to whatever long term boost produces fan speeds you can tolerate.
I've had to temporarily do this for an app install in the past, but could turn it back on again after that. Is that the same situation for Volta, do I permanently have to disable SIP?
you can do that even with the free version, read this thread https://github.com/rugarciap/Turbo-Boost-Switcher/issues/4
btw guys I don't know how you can live without turbo boost switcher... my mbp16 with turbo boost on reaches 90C easily (5k rpm fans) with a not even complex Ableton project, with turbo boost off the same project makes the laptop sit at 65C. I think apple is boosting clock speed too much, but as it is right now for me at least the only way to run this laptop in a decent way is with TB off and SwitchResX. This is the opposite of a "plug and play" device, let's pray for the buyers that are not as nerd as us to find this weird fixes for a broken 3000$ laptop
This does seem promising as I've notice a bit of lag with turbo booster switched off. However, multicore scores are even lower than with turbo boost switched off. Assume that would be noticed as well?
I don't want to take a noticeable performance but, just want to fans to stay under 3000, or so doing everyday stuff.
[automerge]1594561145[/automerge]
Folks that have tried both the i9 and i7, is it clear that the i9 runs hotter? What about with lower lower via volta, or with turbo boost switcher? I'm thinking of downgrading to an i7 and maybe spending the savings in 64gb of ram or just pocketing the difference. Also will downgrade the video card.
I think the 2.4ghz i9 is the coolest "for a given workload".
But it doesn't matter either way which processor you get because any efficiency gain will be used to execute tasks faster, not cooler. But if you disable turbo or use Volta or voltage shift to keep fans under control, the 2.4ghz i9 will be fastest at a given fan speed.
I think the 5500m produces slightly more heat and/or power but both use similar power and both are equally problematic.
If you don't need GPU performance, get the 5300m and bump CPU to i9. If you do need GPU performance, get the 5600m or an egpu and bump the CPU to i9.
The eGPU option, if you can live with its quirks (quit some apps on disconnect) will be cheaper than the 5600m and the fastest CPU performance because dGPU only uses 0-5W leaving max power and thermal headroom for CPU. And 2.4ghz i9 can then run sustained at 3.9-4ghz. Very fast.
The 2.4ghz i9 is the only processor I didn't consider, since it seemed like overkill.
Given the pricing scheme, if you want to upgrade to 1tb, it makes sense to just spec up the higher end model. That's how I ended up with i9 2.3ghz 32gb of ram and an 8gb 5500m. The higher end model starts with the 5500m 4g and it's only 100 usd to upgrade to 8gb. As I do some light video editing in bursts throughout the day I figured why not.
I also ordered a refurbished i7 2.6ghz model with the 5300m, but am sending it back since the battery is not great and I need more storage.
Does the 5300m draw less power in clamshell/with an external monitor plugged in? From what I could tell it does not, but if the 5500m is going to give me problems with thermals I can look at playing around with specing options some more.
I was going to possibly return the i9 2.3ghz with 5500m also because of heat issues discussed here, but if it's a wash with the i7 in most use cases maybe I won't.
If I can underpowered the i9 (2.3) to 27 watts (or use the dynamic power limiting referenced in the video above) and it will outperform the i7 while drawing the same power, then I may stick with the i9 model I have. But then there's the question of the graphics card... I don't need the 5500m but figured it might have a few minutes off my work every day.
Does the 2.4ghz i9 have better thermals than the 2.3ghz?
Im aiming for the coolest system that isn't crippled... like everyone here. The 5600m is beyond my price point... and would prefer to upgrade to 64gb of ram if I was spending more cash.
Nice thesaurus. Pro tip: you'll actually sound smarter when writing normal, concise english. Unintuitive, but it works.
Anyway, I've read more than 5% of this thread and also contributed a workaround to the issue, as have others. Yet again I'm consistently reminded how toxic this community is
If you configure to the same specs there is no price benefit in starting with one or the other. But it is impossible to get i9 2.3ghz without 5500m and 1tb, which add up to the same upgrade cost as going from i7 to i9 2.4ghz.
Unless you NEED fast GPU, don't waste money on the 5500m. Reviews have said the 5300m even performs better for gaming (more stable, the 5500m peaks faster but then stutters in some games likely due to heat and/or higher power consumption starving the CPU some).
And if you do NEED fast GPU, just get the 5600m or an eGPU instead.
The 2.4ghz will have better thermals in that it is faster at a given temp (unless you happen to get a "good" 2.3 in the silicon lottery). But it will not run cooler. Even light to medium workloads that don't need to max out all cores to 100% will still boost as fast as possible to execute their tasks. They will just do it quicker.
With an external display the additional 18W dGPU means that even light tasks push up the overall temp to noisy levels.
I think starting with the i7 model and bumping the CPU is the way to go. Sure, you are forced to pay for 2.4ghz which is "only" 100mhz faster, but it's a better binned chip that can run cooler at a given temp/voltage and therefore can run faster.
Whereas starting with the i9 model means you are forced to buy 1tb you might not need and forced to buy a 5500m you almost certainly don't want even if you do need fast GPU.
What do you get if you set it to, say, 45 W, which is the TDP of the chip? For some of us, curtailing the (extreme) short bursts above this amount might be a good compromise. The i9 can temporarily draw 90 W or more I think.
I think TBS is adequate to extend battery life on battery, but as you noted it comes with a hefty hit to single-core performance; even if you only need one core. This makes it a poor choice (as you noted) to manage heat and noise in an on-charge scenario.
At the end of the day, these chips produce good numbers for marketing but come with significant issues for mobile users, and both Intel and Apple haven’t done much to optimise them for mobile use. This leaves it up to us to use various apps (some of which compromise the security of our devices) to build specific power policies and profiles that best suit our preferences and needs. This might always be the case, but it seems particularly pertinent and necessary for the i9s.
At this point I wouldn’t expect Apple to do any more either, given that they’re moving to ARM. It’s perhaps even in their interest for us to have battery life, heat and noise issues with Intel chips. But so do all the other laptop makers that use them as well. As users, we’ve opted for the slimmest possible laptop with these high-performance processors.
Edit: sorry it’s a bit off-topic regarding the specific GPU issue discussed in this thread.
Last edited: