2020 Air is thermal throttling - yes, really. (CS:GO & bootcamp & other tests inside)

[thermals]

TLDR: 2020 Air is good for using Chrome and super lightweight tasks, but not at all for ANYTHING that stresses the cpu longer than 4-5 sec.

I was lurking here in the forums for almost a month now. I was waiting to upgrade my 2015 Air for years, and ordered the new Air right when I realized the keyboard seems to be reliable, finally. And the i5 was the icing on the cake - even if it's far from the Pros, I thought that would be good enough for years.

Oh boy how wrong I was.

Here's what I found out within 24 hours of having the new Air. Some of the charts are photos not screenshots, I'm sorry about that.

My config is i5/16gb/512gb SG.

I intended to use the Air as a work machine: this means 20-30 Chrome tabs, a few spreadsheets and Word docs open, Spotify is running in the background and occasionally an other tool that's specific to what I do (Screaming Frog if you know it).
I also wanted to be able to play CS:GO occasionally on the lowest possible settings. For the context, it's almost a 10 year old game, and the 2020 Air is a laptop with an above-average price, especially in the EU.

So no 4K video watching or video editing or Geekbench or anything hypothetical.

In the past 5 years I've used a 2015 13" MBP, the 2015 Air I mentioned, and a 2017 touch bar MBP. ALL OF THESE were able to handle my usual workload without the fans being too loud. The Air was noticeably slower, but quiet at least. CS:GO was running fine on the MBPs, but too slow on my old Air.

Now onto the 2020 Air. First I have to admit, the machine is fast until I do nothing else just open browser tabs. But in my opinion, the "not planned to use for heavy workload" does not mean that you shouldn't be able to do ANYTHING CPU-demanding at all.

Here's what I experienced:

First I set up bootcamp to be able to try out CS:GO in better circumstances. Again, this is something that my old 2015 Macbook Pro was able to handle (and it wasn't much more expensive back then than this Air now!) so I don't think I ask too much.

Note how the CPU drops below the base frequency. And not even on a constant level, but it does it like that. In-game this meant that I had 10-15 fps and 70-80 for a few seconds.

Here's what the result looks like with a better benchmark program:

Yup, 400 mhz. "But this surely must be some other error, not thermal throttling!"

Well, this is what the sensors said about it:

"Ok, but this is because of bootcamp, it must be better under MacOS" - that's what I thought at first.

By the way the bootcamp install was a clean install with fresh setup, bootcamp drivers installed as intended, so no idea what could be this wrong.

But I gave it a shot under MacOS as well:

No thermal throttling indeed. But do I want a CPU that "boosts" to 1.2 ghz from 1.1?

(I have the log files saved with more precise data, let me know if you want it)

To the common argument that says "bUt dis Is a LiGHt uSE lapTOP" -> light use does not mean that it should perform worse under load than my 5 years old MBP - for about the same price.

But I thought that this must be just a CS:GO bug or something similar, so I did another test. Here's the performance under my regular workload, under MacOS:

Yep, it drops under the base frequency until the CPU cools down a bit. So it IS thermal throttling.

Geekbench tests are a lie. It can reach maximum performance for a few seconds, but it makes no sense if the constant performance is just about the bare minimum.

To sum up, I spent more money than the original price of my old 2015 13" MBP to get worse performance. I hate the touch bar but I'll send the Air back and wait for the 14" MBP.

I completely understand that I shouldn't expect Pro performance in an Air, and I would be fine with that. But I don't think I should be satisfied with bad performance under load. Just worse than Pro should be enough.

If you want me to test anything else you haven't found in Youtube reviews, let me know! This was just what I could experience within the first 24h. And of course, all the tests were done after updates and indexing finished.
[automerge]1587464905[/automerge]
This is by no means a professional test, this is just what I wanted to see in reviews but no one tested it like that. I hope some of you will find it useful.

 

[lllc]

TBH the machine should never thermally throttle unless its malfunctioning or you are using it in sauna.

 

[RiaKoobcam]

Thanks for doing all that testing, I think that info will probably help a lot of people inform whether they want this laptop or not.

0.4ghz in Windows when gaming... would not have expected that. When I tried gaming on the i5 using Bootcamp it was only with Civ 5 where the FPS doesn't really matter at all, but I didn't have any kind of power monitoring tool running.

Prepare for the influx of people telling you you're using it wrong and there's nothing wrong with a new computer performing worse than a five year old one. I got that when I mentioned my 2015 MBA ran better than the 2020 I owned for a week, even when I tried to qualify with 'my use case is literally just browsing and watching the occasional video'.

Apparently I'm supposed to be thankful that paid $1500 for a computer manufactured in 2020 that is hotter, louder, and has worse battery life than one from the same series made five years ago.

 

[thermals]

Thanks for doing all that testing, I think that info will probably help a lot of people inform whether they want this laptop or not.

0.4ghz in Windows when gaming... would not have expected that. When I tried gaming on the i5 using Bootcamp it was only with Civ 5 where the FPS doesn't really matter at all, but I didn't have any kind of power monitoring tool running.

Prepare for the influx of people telling you you're using it wrong and there's nothing wrong with a new computer performing worse than a five year old one. I got that when I mentioned my 2015 MBA ran better than the 2020 I owned for a week, even when I tried to qualify with 'my use case is literally just browsing and watching the occasional video'.

Apparently I'm supposed to be thankful that paid $1500 for a computer manufactured in 2020 that is hotter, louder, and has worse battery life than one from the same series made five years ago.

Thanks a lot for your time replying to this. I was afraid that despite the numbers people would tell me "it should be like that" but I thought even if I help just 1 person to make the decision, it's worth it.

I get why Apple made these choices, they are in a strong position where they can upsell freely.
Like many others I'm addicted to MacOS, I love how it works (aside from a few weird things) so my best pick for a work laptop will always be a Mac.

But since the Air is not sufficient anymore for anything other than basic stuff, I have to spend more on the Pro even though I won't need that level of performance all the time. But at least that will be consistently good, and not only for a few secs when the thermals allow the turbo boost to turn on.
[automerge]1587468572[/automerge]

TBH the machine should never thermally throttle unless its malfunctioning or you are using it in sauna.

100% agree.

Apple was well aware that the benchmarks will suggest top performance even though real-life usage will convince some users to upgrade to a Pro. And except Notebookcheck, not a single review site pointed this out, just copied the Geekbench screenshots from each other.

 

[RiaKoobcam]

Thanks a lot for your time replying to this. I was afraid that despite the numbers people would tell me "it should be like that"

They're on they're way man, trust me 😂

But since the Air is not sufficient anymore for anything other than basic stuff...

My review's in the User Reviews thread. My needs are very basic and the 2020 i5 wasn't sufficient for those either. It did the tasks, but it did them hotter and louder than the 2015 and 2019 sitting on the desk beside it. For some people that's fine, but it wasn't for me.

 

[Frosties]

The MacBook air was constructed for a 5w processor clocked to 7w, even then it thermo throttled. Now with a 10w that goes in peak over that with the same design it is a spec sheet computer.

 

[esphil]

Sounds like I'll be sending mine back when I receive it. I guess I'll wait for the new mbp or get a refurb 16, this has been extremely disappointing overall reading all of these reports about this laptop after finally deciding to order a machine after using a iMac for the last 10 years. The fact that base clock speeds in 2020 are at 1.1ghz seems like a huge step back.

 

[Malus120]

Wow that is just embarrassingly bad. I get
it it’s a thin and light notebook, sacrifices have to be made yada yada but...
can we be honest for a minute and then method it’s not that thin or light? Seriously the difference between it and a 13 inch MacBook Pro is not all that big.

If we were talking about something more along the lines of the old Retina MacBook form factor this kind of performance would be more acceptable but at this size and weight you don’t have to have performance this bad at this size.

While I’m personally holding out hope for 13” MBPs based on Ryzen 4000 (yes I know Apple is probably going to saddle us with the mediocrity that is Ice Lake, in which case they just don’t get my money), I was planning to recommend the new MacBook Air to some friends of mine who need a new lightweight computer. Unfortunately it looks like I may have to rethink that recommendation.

 

[thermals]

Wow that is just embarrassingly bad. I get
it it’s a thin and light notebook, sacrifices have to be made yada yada but...
can we be honest for a minute and then method it’s not that thin or light? Seriously the difference between it and a 13 inch MacBook Pro is not all that big.

I really wish that more people would realize this. It's not an upgraded 12" Macbook, its a little bit lighter 13" MBP.

 

[bill-p]

To the common argument that says "bUt dis Is a LiGHt uSE lapTOP" -> light use does not mean that it should perform worse under load than my 5 years old MBP - for about the same price.

At risk of beating a dead horse, I also got a 2020 MacBook Air a short while ago. Read extensively on the heating problem thread here, then joined it, discussed for a while...

Here's what I think:
1. Apple apologists will always apologize. Period. Apple can do no wrong. You will not be able to convince them.
2. I got a 16" MacBook and now I'm perfectly happy. I can push it all the way up to the limits of the power brick (100W) and the power brick is actually the thing that gives out before any significant thermal throttling occurs. This is what a "Pro" laptop should be.
3. I made a thread a while ago after getting the 2020 Air that it's clear to me Apple is just giving up on the Air and perhaps the Mac line altogether. They want to push the iPad Pro as a "computer" and it's starting to eat into the Air and the Mac line. But again... Apple apologists will always apologize.

I'll still continue to enjoy my MacBook Pro, at work and at home, for light and heavy tasks, but I'm under no illusion to cut Apple any slack.

Also as an aside, Catalina is a buggy mess on my 16" and AMD graphics still suck.

 

[esphil]

At risk of beating a dead horse, I also got a 2020 MacBook Air a short while ago. Read extensively on the heating problem thread here, then joined it, discussed for a while...

Here's what I think:
1. Apple apologists will always apologize. Period. Apple can do no wrong. You will not be able to convince them.
2. I got a 16" MacBook and now I'm perfectly happy. I can push it all the way up to the limits of the power brick (100W) and the power brick is actually the thing that gives out before any significant thermal throttling occurs. This is what a "Pro" laptop should be.
3. I made a thread a while ago after getting the 2020 Air that it's clear to me Apple is just giving up on the Air and perhaps the Mac line altogether. They want to push the iPad Pro as a "computer" and it's starting to eat into the Air and the Mac line. But again... Apple apologists will always apologize.

I'll still continue to enjoy my MacBook Pro, at work and at home, for light and heavy tasks, but I'm under no illusion to cut Apple any slack.

Also as an aside, Catalina is a buggy mess on my 16" and AMD graphics still suck.

Initially I wanted to retire my iMac and use the Air with my external screen I have with my windows PC. After reading all of this it sounds like the Air isn't going to cut it for me. I've had my iMac 10 years and I wanted a decent upgrade if I'm spending the money. Sounds like I should go with the refurb i7. How do you view the weight difference between the 2? I wanted to be able to use the laptop on the couch when Im watching tv and switch back and forth. I've been getting by with this old imac but I want something more beefy, and not have to worry about how much something throttles or blasting fan speeds if I have 1 too many tabs open.

 

[bill-p]

Honestly, the 16" is a fair bit heavier. I'm not going to diminish the weight difference since I know there are some who think the extra 1.8lbs is a significant number. I don't have any problem lifting and carting the 16" with only one hand, but that's just me.

But the 16" is definitely a lot cooler and quieter than the 2020 Air for me. It's not a small difference there. The new 16" seems to have really good cooling, and it shows.

Also battery life is excellent when I'm doing light tasks like browsing the web, watching some Youtube videos, writing code, etc...

 

[vyruzreaper]

It’s been mostly fine for me under light use like web browsing, sketch, some Dev, office etc BUT when I tried playin hearthstone (card based game - not a fps or anything taxing) on the lowest setting - I got maxed out fans, hot shell, and started to see lags in performance.

 

[RiaKoobcam]

At risk of beating a dead horse, I also got a 2020 MacBook Air a short while ago. Read extensively on the heating problem thread here, then joined it, discussed for a while...

Here's what I think:
1. Apple apologists will always apologize. Period. Apple can do no wrong. You will not be able to convince them.
2. I got a 16" MacBook and now I'm perfectly happy. I can push it all the way up to the limits of the power brick (100W) and the power brick is actually the thing that gives out before any significant thermal throttling occurs. This is what a "Pro" laptop should be.
3. I made a thread a while ago after getting the 2020 Air that it's clear to me Apple is just giving up on the Air and perhaps the Mac line altogether. They want to push the iPad Pro as a "computer" and it's starting to eat into the Air and the Mac line. But again... Apple apologists will always apologize.

I'll still continue to enjoy my MacBook Pro, at work and at home, for light and heavy tasks, but I'm under no illusion to cut Apple any slack.

Also as an aside, Catalina is a buggy mess on my 16" and AMD graphics still suck.

When my 2020 Air started being noticeably hotter in my lap than my 2019 Air, after the indexing was done, and when the fan was on... all of the time... I thought 'huh, must be a MacOS thing.'

I'm really surprised that the performance on Windows 10 was as bad as it was. In my experience, definitely better than it was on Catalina, especially from a cooling perspective. The CPU ran cooler and Windows was able to keep it from hitting 100 every time an action was performed.

But still, using Windows and still eventually having the Air get reeeal warm and have terrible battery life made me wonder.

Caveats - I'm not an engineer, fluid dynamics expert, designer, etc. Everything I learned about electronics and hardware was by taking stuff apart when it was broken and fixing it.

Windows is designed to work with... a stupidly large array of software. Like Android, I don't envy MS the task of making an OS that works well with almost every computer part combination under the sun. But they manage it - Windows 10 allows you to set up the OS so that it works on almost any configuration you have. Even with Apple's subpar drivers, Windows 10 managed heat and power better than Catalina.

Apple's value proposition lately has been 'we design the hardware and the software together so they work, no tweaking required.' If I had a dollar for every time someone suggested I turn turbo boost off in my week with the i5, well, I'd have about five dollars.

Still, when the hardware works better with another OS, that's when Apple loses its value proposition to me. And by better I don't necessarily mean faster (look at the original post), I mean what I buy a Macbook Air to do - cool, quiet messing around with non-taxing stuff.

So even though I'm not an engineer, even though I'm not a fluid dynamics expert, even though I'm not a designer, maybe the hardware just isn't right. It seems like they bit off more than they could chew with the Ice Lake chip, but really wanted to save on the R&D costs of replacing the Macbook Air's chassis. That's their prerogative, they're a business, and if most people are happy with that, cool.

But I'm not crazy for wanting a computer that doesn't have to actively make itself slower to stop itself damaging its internal components, especially when previous generations of this computer have managed everything so well.

I think this is one of the problems that happens when people get really, really attached to a brand. I'm not saying the Air is terrible, but I experienced similar stuff to the OP - I just didn't have the knowledge to test it, quantify it and post it. It's possible for something to be somewhere between perfect and absolute trash, and the 2020 Air sits somewhere on that line.

Trying to explain that has gotten me told either i) I'm using it wrong or ii) 'Yeah, it's the worst computer ever'. No no, there are way worse computers out there, I'm just trying to understand how they got from 'very good ultraportable in 2019' to 'why don't you just buy a Macbook Pro then, if you want to have more than TWO TABS running?!' in a year.

I think the answer might be as Bill suggests - maybe they want to push people towards the Pro, maybe they want people to switch to iPads for this kind of use, maybe this is the last generation of Intel Macs and they just don't care about the subset of users who say 'this thing sucks for what I bought it for'. I don't know.

 

[Falcon80]

Hopefully we can see 14" MacBook Pro soon. The price for a refurbished 16" MacBook Pro looks attractive!

 

[ARJR84]

Thanks for your results, OP. Can you disable 1 or 2 logical cores and see if it decreases temps? Of course, it won't be a solution as who buys a 4 core only to use 2 or 3, but just for testing purposes.

The brave souls in here, utilizing Apple good return policy and showing their results have helped dozens of us from the hassle.

 

[RiaKoobcam]

Thanks for your results, OP. Can you disable 1 or 2 logical cores and see if it decreases temps? Of course, it won't be a solution as who buys a 4 core only to use 2 or 3, but just for testing purposes.

The brave souls in here, utilizing Apple good return policy and showing their results have helped dozens of us from the hassle.

Anecdotally, when I still had the i5, I disabled two of the cores using msconfig.exe. It was so slow it was pretty unusable (mouse input lag etc), also no reduced power draw when viewed in Throttlestop.

I didn't try in MacOS though, so don't have any info for that sorry

 

[CEaton]

You are far from alone my friend, I just received my MBA i5/16/512 yesterday, just during set up, photos app parsing 22k photos, mail downloading 110k emails, iCloud syncing, onedrive, you know, normal stuff, this thing was sounding like a get plane at 100 C, throttling as well. I am going to give it the week, but I am most likely going to be returning this on Monday next week, 1 week from the return time frame. I will wait for the new MBP 13. This is replacing a 2017 MBP 13 i5/8gb/256, and there is no doubt, the old one handles the load better than this one does. It‘s a shame, I was super pumped for this, I have a super powerful workstation and two work laptops, this is basically just for play, and some VMware Fusion, but I haven’t even gotten around to putting fusion on it. Also, when hooked up to a 4K monitor, where this will spend most of it‘s life, matters get even worse.

ugh! Now....when is the new MBP coming out???

 

[bill-p]

I think the answer might be as Bill suggests - maybe they want to push people towards the Pro, maybe they want people to switch to iPads for this kind of use, maybe this is the last generation of Intel Macs and they just don't care about the subset of users who say 'this thing sucks for what I bought it for'. I don't know.

I honestly don't even just "think" this is the case. It is obviously the "case" to me, as Apple just came out with a new keyboard "case" specifically just so the iPad Pro can turn into a computer.

And I don't think this is the last generation of "Intel Macs". The iPad technically is already running Mac OS (iPad OS runs the same Darwin core as Mac OS). All Apple has to do is gradually tweak the desktop environment (i.e.: user interface) so it'll slowly turn into a computer, and voila, they will have their new portable computer for the masses. Meanwhile, Mac OS will probably turn into the "Pro OS" for developers, designers, movie makers, etc... and Intel Macs will essentially just be MacBook Pros or Mac Pro.

The ARM MacBook does not exist and the MacBook Air's days are numbered.

 

[kdm0505]

had the same feelings about my 2020 MBA i5 / 256 / 8gb and now back to my 2013 MBA i5 / 128 / 8gb. I desperately need a new laptop and Apple will (most likely) get what it wants and get me to upgrade to either the new 14" or 16" MBP.

 

[high heaven]

Obviously, MBA 2020's cooling system is poorly designed and yet people kept saying this: MBA is for light uses! So what? How come MBA throttles for watching 4K video and conferencing with Zoom? MBA 2020 is already around $1000 which is very far from a cheap laptop.

 

[magbarn]

Hopefully we can see 14" MacBook Pro soon. The price for a refurbished 16" MacBook Pro looks attractive!

There's some really good deals on NEW mbp 16's right now if you know where to look. Can get $500 off Apple's MSRP
I suspect there will be a MBP 16 refresh with the MBP 14's release. 10th gen 6/8 core CPU's though are another 14nm++++++ Intel retread and not 10nm like the MBP/MBA will/are using...

 

[nylon]

You certainly have strange results. While the other reviewers have pointed out that the computer cannot sustain turbo boost clocks for pegged cpu loads under synthetic benchmark conditions such as Cinebench, none, to my knowledge, have indicated that the machine thermal throttles (i.e to below base clock) under sustained load. Has bootcamp been updated with drivers for the 2020 Air yet? Something seems very wrong here.

P.S. Also in your first screenshot the PkgPwrLimit0 is reading 100W and only running at 5.75w. The MBA cpu is not rated or capable of 100W power draw limit.

 

[thermals]

You certainly have strange results. While the other reviewers have pointed out that the computer cannot sustain turbo boost clocks for pegged cpu loads under synthetic benchmark conditions such as Cinebench, none, to my knowledge, have indicated that the machine thermal throttles (i.e to below base clock) under sustained load. Has bootcamp been updated with drivers for the 2020 Air yet? Something seems very wrong here.

P.S. Also in your first screenshot the PkgPwrLimit0 is reading 100W and only running at 5.75w. The MBA cpu is not rated or capable of 100W power draw limit.

I can't wrap my head around these results either. I think this should deserve more attention, so other people might try it too and see if they can get the same results, because at this moment just like you said, not one reviewer admitted that the machine actually thermal throttles below the base clock.

Drivers: I tried running the tests right after bootcamp finished installing (drivers included), tried reinstalling the drivers, running Windows Update, and even tried updating the most important drivers manually, one by one in device manager. I'm not sure what else is there to try...

Draw limit: yeah I noticed that too, but that might be just the Intel Gadget's weird behaviour, that's why I tried HWINFO as a fallback.

 

[nylon]

I can't wrap my head around these results either. I think this should deserve more attention, so other people might try it too and see if they can get the same results, because at this moment just like you said, not one reviewer admitted that the machine actually thermal throttles below the base clock.

Draw limit: yeah I noticed that too, but that might be just the Intel Gadget's weird behaviour, that's why I tried HWINFO as a fallback.

Well they have been trying it. There has been lots of hoopla about inadequate cooling and lots of tests have been run. It's not that they have not been 'admitting' (hiding) anything. None of those who have run tests under multiple sustained cpu load scenarios, both synthetic or real world (like final cut rendering), have shown throttling under the base clock. The consensus is basically that the 'expectations' for the quad core CPU don't match reality. While the MBA is on par and even surpasses the base 13"MBP for single core burst performance its still slower on the multicore side both in synthetic and real world tests. The machine simply does not have the thermal envelope/cooling to sustain turbo boost for sustained cpu workloads (but it does not throttle under base clock). For that you need to go to a MacBook Pro, as you always have had to in the past.

That aside, there seems to be something seriously wrong with your computer and I would suggest returning it or exchanging it. Even under the MacOS power gadget your CPU utilization never even approaches 100% in either picture. i.e not even pegged. And in one of your 'attachments' of the MacOS power gadget the scale under Power graph is not the same as in the other photos with a power spike at 260 watts at the end with CPU utilization at 5.6% (which is not physically possible). And the UI scaling is different. In fact in none of the MacOS Power Gadget photos you've posted does the CPU ever approach anything close to 100% utilization. It's all totally irrational. The only consistent result is in your first picture under windows 10 (apart from PkgPwrLimit0 indicating 100w) where CPU utilization indicates 100%. This only indicates that the power management driver has not been updated for the MBA under bootcamp.