OS Level Throttle Fix Is Evidence Apple May Not Be Done Resolving The Issue

[RumorConsumer]

Murmers have begun that Apple is not done working out the throttling issue.

“PCWorld’s Mah Ung speculates that even more tuning could be around the corner, and I suspect that’s because the performance uplift here, while welcome, isn’t quite where it should be yet.”

https://www.forbes.com/sites/jasone...cbook-pros-performance-problems/#2627b9359051

So here’s the thing. Right now the fix appears to take place at the OS level and did not change the firmware version # in System Profiler. This seems like an unlikely and somewhat uncharacteristicly nonpermanent way to resolve this. What seems likely/possible is that at some point soon there will be a firmware level update that resolves the issue permanently and more thoroughly under all OS conditions.

Reasons:
- this method, while helpful, is sloppy as a long term fix due to any OS needing to apply it and could possibly result in damaging the hardware which means bottom line is impacted in human service and repair hours and hardware costs. Also possible is that Apple’s strategy for issuing these types of fixes has shifted as the firmware (T2) has evolved and along with it its relationship to the OS.
- There are still some reports floating around that other methods of addressing the issue have been better performers than Apple’s fix. Yes, they might be hedging a bit on performance and heat dissipation but it’s possible they intend to hit this one more squarely than they have.
- There are people having other issues with the machine’s brain aka the T2 (kernel panics) and it’s likely this throttle issue caught Apple by surprise, the forthcoming T2 update wasn’t ready yet, this fix is now being incorporated into it and the OS level fix was the best stopgap available.
- The performance differences in benchmarks between the chips (benchmarks are a drag race, that topic has been beaten to death and still they aren’t nothing) is too small. More tuning is needed to eek out better differentiation between models. This one may be iffier than the others but I think still carries enough weight to bare mentioning.
- In every movie where the portal to hell is closed its a permanent fix not some temporary software patch.

Welcome all your thoughts either way.

 

[sennomulo]

I've heard from some Bootcamp users that there are still issues in Windows. If true, then obviously there is some deeper problem that must be rooted out, and this was Apple's quick fix to stop the bleeding on this throttling controversy. While the fix is welcome, if it is merely a band-aid solution, then I would expect (or at least hope for) firmware updates as Apple engineers dig deeper into it over time.

Though you'd think they would have had enough time already with these machines to understand how they work. One wonders if they hurried the release of these models to stop the bleeding on the previous controversy over the keyboard...

 

[DeanL]

This seems like an unlikely and somewhat uncharacteristicly nonpermanent way to resolve this. What seems likely/possible is that at some point soon there will be a firnware level update that resolves the issue permanently and more thoroughly under all OS conditions.

• Why does it seem like "an unlikely and somewhat uncharacteristically nonpermanent way to resolve this"?

• Are there any engineers on this thread that think that it's usually firmware thing? Do we know what's in the T2 firmware?
• How do we know that changes in other locations/chips weren't made and that this is a 100% OS level fix?

 

[RumorConsumer]

• Why does it seem like "an unlikely and somewhat uncharacteristically nonpermanent way to resolve this"?

• Are there any engineers on this thread that think that it's usually firmware thing? Do we know what's in the T2 firmware?
• How do we know that changes in other locations/chips weren't made and that this is a 100% OS level fix?

1. I think I laid out my reasoning.
2. We do know a bit. We have a lot of information indicating it’s the next step in the evolution from NVRAM to EFI and now T2 or “Controller” as its now categorized as in the System Profiler. There is no more “EFI version” which indicates this transition has indeed occurred. Additionally, there are some system attributes than can be changed in either firmware or with an overriding software change. Example - your display has a default brightness that even without an OS loaded it will demonstrate. The OS can change that once it’s loaded.
Based on this: https://www.macworld.com/article/32...e-imac-pro-the-start-of-a-mac-revolution.html
And numerous other articles describing the functions of the T2 it’s extremely low level. We don’t know if it operates at the level of a POST yet but I’d not be surprised as part of it’s purpose is to ensure security in the boot process and as it appears to be the full fledged successor to EFI.
3. Because the firmware on something detectable would almost certainly change. Apple may be secretive but they do reliably update versions even when they tell us nothing about why.

 

[johnalan]

Totally agree, as shown here: https://www.notebookcheck.net/MacBo...k-results-after-software-update.318234.0.html

The fall off after load is sustained is still significant.

Saying that, they fixed it enough to make it a good machine. Still I'd like to see more.

 

[Lennyvalentin]

There is no more “EFI version”

Maybe macOS doesn't report it, but it's certainly there. The CPU needs firmware, so it knows how to access hardware devices and bootstrap the OS. The T2 cannot be, and is not a replacement for that.

 

[RumorConsumer]

Maybe macOS doesn't report it, but it's certainly there. The CPU needs firmware, so it knows how to access hardware devices and bootstrap the OS. The T2 cannot be, and is not a replacement for that.

Genuinely curious why not? Or maybe better question - if the T2 controls all the aspects of the hardware that Snell's article (posted above) reports that it does, why would it not be fair to say its the replacement? Do you understand the relationship between the T2 and the EFI to be different?

Update:
This Forum QA seems to hold some answers:

https://www.reddit.com/r/apple/comments/7o6pch/the_t2_chip_makes_the_imac_pro_the_start_of_a_mac/

..with one user referring to the T2 as the "secure gateway to EFI". So is it possible we longer have an EFI version available to us? Seems odd.

 

[volumeWindow]

How do we know that changes in other locations/chips weren't made and that this is a 100% OS level fix?

The problem exists in its original form under windows. The system update Apple pushed to macOS is just doing the same thing that you can do manually in Intel's XTU. The difference is that it looks like the macOS fix takes into account a few more variables and is capable of changing the max power level dynamically.

 

[RumorConsumer]

Yeah it will be really interesting to see what the updates (if and when they emerge) actually change.

 

[Lennyvalentin]

Or maybe better question - if the T2 controls all the aspects of the hardware that Snell's article (posted above) reports that it does, why would it not be fair to say its the replacement?

It's not a replacement because EFI and T2 chip isn't even apples and pears; those are both fruit. EFI is software, which runs on the CPU, and the T2 is an embedded SoC (with its separate, walled-off software.) An external processor like the T2 can't boot up the CPU; only the CPU can boot up the CPU, and in order to do that it needs firmware to set everything up, initial parameters for chipset, memory controller, find a bunch of basic devices on various buses (disk controller, keyboard, video and so on), set things up to start loading the OS...and more.

 

[Stux]

The problem exists in its original form under windows. The system update Apple pushed to macOS is just doing the same thing that you can do manually in Intel's XTU. The difference is that it looks like the macOS fix takes into account a few more variables and is capable of changing the max power level dynamically.

Yes. When I was testing to find power limits I found having the LG5K monitor connected meant I had to reduce power limits by 4 watts, and I think disabling iGPU meant I could bump it up etc.

Quite possible the wattage is being modified based on what is active, which is something I couldn’t do with the basic XTU type tweaking, thus forcing me to assume worst case, which was less than 43W.

Thus, Apples current solution gets better performance for me, but also it appears they’re not pushing the CPU as hard as it could be.

 

[CrashTestWalrus]

I have no idea if they will actually do anymore improvements to fix the throttling. I am not sure what they can do that doesn't involve actual tweaks to macOS, and to be clear the fix is doing something to macOS you can see that by using bootcamp or just wipe your drive and use internet recovery to go back to shipping version the throttling will come back. I hope that they can and do fix it. Having the laptop perform in a way that it doesn't throttle, which in modern computer systems can be handled by a lot of systems power delivery and cooling to name a couple, in any OS you decide to run on it is important. I shouldn't need to run macOS on the machine so it runs at the actual base clock of the CPU. That seems absurd. There is a lot of theory out there on what exactly Apple messed up when making this laptop, unfortunately they alone know what exactly is causing the throttling with absolute certainty. Hopefully it is something they can fix at a lower level than macOS and do fix it at that level so the people using bootcamp and linux distros on this machine aren't hosed.

 

[DeanL]

The problem exists in its original form under windows. The system update Apple pushed to macOS is just doing the same thing that you can do manually in Intel's XTU. The difference is that it looks like the macOS fix takes into account a few more variables and is capable of changing the max power level dynamically.

So then you mean it should be fixed in macOS and not in the firmware? If it takes in account variables that can't be managed at the firmware level, then isn't the OS the proper level to fix it?
[doublepost=1532834170][/doublepost]

3. Because the firmware on something detectable would almost certainly change. Apple may be secretive but they do reliably update versions even when they tell us nothing about why.

Is it possible that the version number of the controller we're seeing is only one specific component of the T2 and that something else was indeed updated but we haven't noticed because we do not know how to check the version of that component?
[doublepost=1532834579][/doublepost]

Maybe macOS doesn't report it, but it's certainly there. The CPU needs firmware, so it knows how to access hardware devices and bootstrap the OS. The T2 cannot be, and is not a replacement for that.

It's not a replacement because EFI and T2 chip isn't even apples and pears; those are both fruit. EFI is software, which runs on the CPU, and the T2 is an embedded SoC (with its separate, walled-off software.) An external processor like the T2 can't boot up the CPU; only the CPU can boot up the CPU, and in order to do that it needs firmware to set everything up, initial parameters for chipset, memory controller, find a bunch of basic devices on various buses (disk controller, keyboard, video and so on), set things up to start loading the OS...and more.

That's where I get skeptical and also believe that the EFI still exists but Apple simply hides it or changed the way it reports in macOS.
[doublepost=1532834953][/doublepost]Also, Craig Federighi confirmed that Mojave β will be patched. If Apple really intended to fix it at the firmware level but couldn't because the fix for the T2 was not ready, why would they bother to include in in a release that's expected in September?

 

[Qaulity]

I've heard from some Bootcamp users that there are still issues in Windows. If true, then obviously there is some deeper problem that must be rooted out, and this was Apple's quick fix to stop the bleeding on this throttling controversy. While the fix is welcome, if it is merely a band-aid solution, then I would expect (or at least hope for) firmware updates as Apple engineers dig deeper into it over time.

Though you'd think they would have had enough time already with these machines to understand how they work. One wonders if they hurried the release of these models to stop the bleeding on the previous controversy over the keyboard...

I reinstalled macos 3 times in the past 2 days to get bootcamp to work and each time I had to reinstall the throttle update.

 

[Erasmus]

So then you mean it should be fixed in macOS and not in the firmware? If it takes in account variables that can't be managed at the firmware level, then isn't the OS the proper level to fix it?

Intel XTU is Software, not Firmware. It makes sense that Apple's OS fix would have more power than a software fix, because Apple writes the OS which would gate and moderate any 3rd party software fix.

However, a firmware fix would be better still, as it would be applied immediately on power-on, instead of being applied when the OS loads. The current MacOS fix is clumsy, as it is likely the CPU starts up with the original unattainable power limits (100/125W?) before being reduced once the OS is loaded. Obviously this means the fix does not exist when using BootCamp.

Other forum members are improving on post-fix performance by further adjustments to CPU voltages and power limits. Apple could do this too with further optimisation, and perhaps make other changes to optimise the performance/power/thermal balance which are not possible for us to change. They can also put it in firmware to make the fix clean and universal.

 

[vemac575]

Pretty certain there will be further measures taken to resolve any smaller issues and assure the larger issue to be resolved. Apple makes mistakes sometimes but they fix the hell out of them.

 

[Newtons Apple]

In the mean time my 2018 MBPto is running Win10 just fine. Apple constantly improves its devices with firmware updates.

 

[leman]

What in your opinion is a difference between a "temporary software patch" and a "permanent fix" and why do you think that this fix requires a new firmware version? Its a power controller configuration question, so it is a software issue. I very much doubt that Apple could create and test new OS-level power management code in those few days, you just do do this, its too risky. And they are not relying on the built-in CPU power management, which would be the easiest thing to do.
The explanation they gave so far makes most sense — that something was incorrectly signed and thus probably relevant parts of the power management code was not loading.

Also, is it true that "performance is not quite there yet"? It seems identical to me to other laptops of similar category, with comparable thermal headroom. During sustained Cinebench runs, my i9 consumes around 50Watt, I doubt that the laptop is designed to dissipate more.

Finally, my speculation is that it doesn't work in Windows because Windows lacks the OS driver for interacting with the T2 chip, so power management features cannot be properly utilised. If this is the case, I doubt that we will ever see a fix. As an easy remedy, Apple should set CPU power limits when EFI booting to other OS.

 

[CreeptoLoser]

That logic board and the T2 chip is new and we won't see final optimal drivers immediately. By the time Mojave comes out we'll have it better but you have to hit the bug report button. Apple internally won't have all the wide range of configurations you have.

 

[maflynn]

That logic board and the T2 chip is new and we won't see final optimal drivers immediately.

I have to disagree, first off the T2 is not new, its been out since 2017 in the iMac Pro. I don't think we should give apple a pass in terms of saying its not their fault they need more time to optimize the drivers for a new logic board, especially since they developed that logic board themselves.

 

[CreeptoLoser]

I have to disagree, first off the T2 is not new, its been out since 2017 in the iMac Pro. I don't think we should give apple a pass in terms of saying its not their fault they need more time to optimize the drivers for a new logic board, especially since they developed that logic board themselves.

If you have a PC you will notice that new firmware is always coming out for motherboards and SSDs even up to 2 years after release.

 

[maflynn]

If you have a PC you will notice that new firmware is always coming out for motherboards and SSDs even up to 2 years after release.

Yes, and apple continues to release firmware as well, but to say its not optimized is quite different imo. The issues we're facing with throttling and heat are not because the drivers need to be tweaked. Apple screwed up, plain and simple, they're shoe-horning a hot CPU into a tiny enclosure that cannot cool it as effective. Additionally they've not did the testing needed to ensure that the temps are more reasonable imo.

 

[CreeptoLoser]

Yes, and apple continues to release firmware as well, but to say its not optimized is quite different imo. The issues we're facing with throttling and heat are not because the drivers need to be tweaked. Apple screwed up, plain and simple, they're shoe-horning a hot CPU into a tiny enclosure that cannot cool it as effective. Additionally they've not did the testing needed to ensure that the temps are more reasonable imo.

Dude. My PC has had eight BIOS updates from Asus this year for performance and stability issues with Coffee Lake. Those updates are issued by Intel to the OEMs so all PC motherboards are affected.

https://www.asus.com/uk/Motherboards/TUF-Z370-PLUS-GAMING/HelpDesk_BIOS/

When was the last time Apple had to release 8 firmware updates in 7 months?

 

[RumorConsumer]

Maybe macOS doesn't report it, but it's certainly there. The CPU needs firmware, so it knows how to access hardware devices and bootstrap the OS. The T2 cannot be, and is not a replacement for that.

Dude. My PC has had eight BIOS updates from Asus this year for performance and stability issues with Coffee Lake. Those updates are issued by Intel to the OEMs so all PC motherboards are affected.

https://www.asus.com/uk/Motherboards/TUF-Z370-PLUS-GAMING/HelpDesk_BIOS/

When was the last time Apple had to release 8 firmware updates in 7 months?

Do you really think it’s fair to compare the two? Apple, who seamlessly guides or builds all the components on their boards or ASUS, who builds mobos for customization and a zillion different possible component possibilities? And... ASUS has competition for what they do so every little bit is scrutinized and if they miss a play that’s sales. Very different markets.

 

[leman]

Yes, and apple continues to release firmware as well, but to say its not optimized is quite different imo. The issues we're facing with throttling and heat are not because the drivers need to be tweaked. Apple screwed up, plain and simple, they're shoe-horning a hot CPU into a tiny enclosure that cannot cool it as effective. Additionally they've not did the testing needed to ensure that the temps are more reasonable imo.

There are very few laptops on the market that can dissipate significantly more heat from the CPU, and those are large and heavy gaming + workstation machines. The MBP is able to dissipate 45-50W of CPU heat in sustained runs, which is absolutely adequate. So while these CPUs can perform better in a larger enclosure (equally true for any other CPU in the last few years), they perform within their spec.

The issues we were facing are because of power management algorithms. The throttling problem was not because the laptop can't dissipate the heat but because the power manager was misconfigured to let the CPU draw more power than the system can handle. With the firmware update, this has been fixed and the MBP now performs in line with other comparable laptops such as Razer Blade and XPS 15".