- There's no denying the MacBook Air 2020 is a pretty sleek little machine. Except for one small
flaw, instead of doing what every other laptop
manufacturer ever does and put a fan right next to the CPU where the majority of
the heat is generated, Apple opted to put the
fan over yonder somewhere, leaving the CPU with only about 10 watts of cooling capacity to work with. How much performance did
they leave on the table? To find out the answer to that we have collected this
bin of cooling apparatus with everything from
upside-down air things to water cooling pumps to this tray, which you guys might
recognize from our video from several years back, water cooling the original MacBook 2015 by actually submerging it in water. Yes, my friends today is
gonna be a lot of fun. And it's brought to you by Ting. Why pay for what you're not going to use? With Ting you only pay for
the mobile data that you use and you can learn more at the
link in the video description. (upbeat music) To tell if our wacky cooling
solutions are doing anything we will first need to establish
a performance baseline. Running our MacBook Air, which is a quad-core eight thread CPU with no aides whatsoever. I'm just gonna fire up
Intel Power Gadget here and you can also see we've got a quad core 8
thread MacBook Pro 13-inch running right next to it. Now, it does run a different CPU but it should give us some idea of what we might strive for
with a properly cooled chip. Looking at the power draw these CPUs, our MacBook Pro starts by
shooting up to about 40, 45 watts and then dips down closer to 30, 35 and finishes out the test there. So there's no way that a 25 watt peak CPU is going to be able to match
that in terms of performance. But you can see that over time we fall down as low as 10 watts. So I think we should be
able to close the gap a fair bit here. 100 degrees, that seems
perfectly reasonable. Actually, they both run at 100 degrees just this one doesn't
have a fan attached to it. 1,016 points, darn it,
about half the speed. Let's try and close that gap. Cooling method number one, a stand. - [Brandon] Well, and a fan. - Oh, and we're gonna add a fan to it. put that there. - [Brandon] Yeah that's why
the rubber chummy's are on it. - That's pretty cheesy. So between 40 and 45, we'll
consider that baseline before we start each test? - Yeah, sounds good.
- Is that reasonable, okay. So here we go 1,016
that's the score to beat. Conspiracy theory time,
tell me I'm wrong here. Apple has been sandbagging the performance of their Intel based
MacBooks for years now so that when they announce
the ARM based MacBooks they can be like, bam, it runs
it like half the temperature! No, no applause, nothing, all right. It is amazing how quickly
this thing shoots up to 100 degrees isn't it? - [Brandon] It's pretty much instant. - How can it not love
itself if itself is so hot? Ey, got 'em, lttstore.com. Okay, you told me this was gonna work, I don't think this is working. We're down to like 11.6 watts already. So tell me, Brandon, where
do you think the CPU is? - [Brandon] I don't know,
Linus, in the middle of the PC? - Could it be here?
- No. - Or here?
- No. - [Linus] 1,034, all right, well so much for this idea working. - [Brandon] Hey, that's a 2% increase. - 2%, all right, we're gonna
have to get more extreme. what's test number two? - Fan with no bottom, so this
is kind of the test where like if they just put a fan in there this is how good it would perform. - All right. With the unit open we've got a great look at what we were talking about before so here's the CPU right under this tiny little sort of a heatsink, so it's got some little
folded metal fins over it, it's a heatsink. And then here's the fan, that they went and they
they put it in the laptop, so like it has a fan. They just didn't bother to have the fan draw air over the heatsink to cool the CPU before exhausting it. I don't think anyone has yet offered up a clear and
believable explanation for why they would have done this. They went and they put the
heatsink and the fan in there, they just didn't bring
them together, you know. Run number three, definitely going slower, it's making its way to a
hundred degrees slower. Now we're having a look at
the heart of the system here, that is almost 10 degrees lower. Maybe not 10 but more than five for sure. Dang, MacBook, you fly. Our package power consumption
is higher than last time which means I'm expecting
better performance but our utilization is only about 87%. And of course we did hit 100 degrees here so it looks like, it looks like, anyway, that we are
thermally limited right now. Why is it asking for 3.4 gigahertz, it's got a know it's
not going to get that. There it is, 1,088. Woo, what's what, like another 4%? But we can do better. So far we've got about an
8% uplift in performance over our stock configuration. But that's without changing
out our thermal compound. So that's gonna be the
next leg of our journey. We're gonna go with Noctua
NT-H2 thermal compound. See, I'm pushing that down flat. It seems like it's not actually designed to sit right against the die. - [Brandon] That makes no sense. - I mean, it makes sense if you
want the thermal performance to not be as good as it could be. - [Brandon] Do you want to put it back on and like give it a wiggle? - [Linus] It's not gonna do anything 'cause it doesn't go down far enough. See? So I can wiggle it all I want but it can't go down any farther. See? Oh, shoot. - [Brandon] Are we taking a
trip to the milling machine? - I think we have to. - [Brandon] That's fun. - I knew you'd like that. Recess those a little bit. - Please don't ruin it.
- That's the plan. - [Linus] Stressed out, Alex. - [Alex] You and me both. - [Linus] They don't exactly
sell this replacement part on apple.com. - [Alex] Have you tried
getting the Apple card? Maybe they have it then. Now it's touching. We're going to take this nut right here that controls the depth. It's nice and snug there, gonna
back it off just that much. - I'm uncomfortable. - [Brandon] Yeah, I was not thinking we'd be doing this here today. Well I guess--
(mechanical whirring) - [Linus] I can't look. But I can't look away. Okay, that's lots, that's enough. - [Alex] Aw, I think
that's more than we needed. - I hate this.
- Why did I buy you this? - You know what sucks though, Alex, is well we might be getting better contact we now have what I would consider to be a significant amount less
copper on the heatsink. Like we didn't remove much,
but compared to what's there. - [Brandon] The worst part is this was supposed to be the
reasonable do at a home option. - We had no way of knowing that they were gonna
have such a boneheaded, I mean, I don't want to call it a flaw 'cause it's clearly intentional,
but design decision? We use thermal pads because they allow for greater tolerances in things like surface mount
components like a memory chip. To do it with a die like this where the manufacturing
tolerances are so tight, there's no reason for it. That I'm aware of, other than just not wanting it to perform
well, I don't get it. - [Brandon] Well I'd say it's
just cheaper for them to make. - All they had to do was put a spring on each one of those screws
and do a spring mount like everyone else ever. Got a little wiggle wobble,
let's have a look here. Oh yeah, that is way better. Now we just need to be careful not to accidentally over tighten these, 'cause that is a definite possibility. - [Brandon] If it's not
immediately at 100 degrees then it's way better. - [Linus] We're still at
17 watts package power. - Huh?
- And only 90 degrees. - [Brandon] Really? Okay, so that actually is
worse than before with the fan but the temps were definitely better so I feel like it might
just be the turbo profile? - [Linus] But that's better
than our first run, isn't it? - [Brandon] Yeah, that's 3% better. - Even if I'm not sure
exactly what to make of that the show must go on so let's
move on to our next idea which is expanding the amount
of cooling surface area that the machine has by
putting a thermal pad in between this heatsink
here and the chassis. It actually seems like Apple's
using a thermal insulator in order to keep the bottom
of the computer cooler, so I guess we're gonna have
to pull that off first. - [Brandon] Yep, that goes rip. - To be clear, the inclusion
of this insulator here is something that I actually oppose. The difference between something being comfortable
against your skin and being literally burned by it is actually a surprisingly small amount. - [Brandon] Yeah, like 43
degrees is what you shower at and 46.7 will burn you in 35 minutes. 50 degrees will burn you in one minute. - Okay, how many layers thick
should we go here, boys? Sure, why not, if we're gonna wing it we might as well wing it hard. These kinds of pads are hardly the most
thermally conductive material but they're obviously
better than an insulator if we're trying to cool the
CPU at the cost of the user. - [Brandon] So if the
bottom of the chassis gets above 46 degrees then we're bad, if it stays below that
then we're all good. - [Linus] Ooh, 40 degrees. Now that's a hot bottom! - [Brandon] No, that's actually fine. - [Linus] Well, it's also not done yet. - [Brandon] Yeah, but 40
degrees is totally comfortable. - [Linus] Well, let's see how comfortable. - [Brandon] 43 degrees is
like a bit uncomfortable, 46 is you get burned.
- That's not to bad yet. We are still under 75 degrees.
- What the hell? - Well, you got so little heatsink area and then you go and spread
all that heat out over this giant piece of aluminum, yeah, it makes sense it
would do a little better. - [Brandon] It did so much better when you just had the
heatsink mounted properly and it probably has a tiny bit
more mounting pressure now. - [Linus] 43.9. - It's getting hot.
- It's getting hot, it's a hot bottom! Results coming in, Alex, you ready? 1,155, we never cracked 80 degrees. - [Brandon] That's a 14%
improvement above stock. - That is freaking nuts, hold on, I'm gonna start it running again just so that it doesn't
really get a chance to cool down too much, I want to see what our hot
bottom temperatures ended up at. 46. - [Brandon] Yeah, you're in
getting burned territory now. - Yep, that's too hot,
that's not comfortable. - [Brandon] Now like
lots of gaming laptops will be hotter than that on the bottom. - Well, I mean, it doesn't
have to be like that, they could have put a fan on it. That was an option. So much performance left
on the table though. It's clear that to reach
our next performance tier we need to do something about the cooling of the chassis itself. That is where this comes in. Yes, my friends, I would
like to do a recreation of of our 2015 experiment,
immersing the laptop in water. - [Brandon] I don't think
we should do that at all. - Well, what's your brilliant idea? - [Brandon] We water cool it. So I got this little thing
right here off of Amazon, it was $17.
- Amazon, lies. You've been shopping on AliExpress again. - [Brandon] it was $17 and
it says laptop water cooling. That's it, no documentation, this is all that came in the box. I think it can do the job way better than that
cookie sheet right there. I'm gonna go get water for this. - Look, it's not like
the thing died last time. - [Brandon] Yeah, but like you
can fall off your motorcycle on the highway and not die, that doesn't mean you want to do it again. - That is a false equivalency,
it wasn't even damaged. - [Brandon] You could just
you know slide along the road and be like, oh, it's fine. Next time I definitely
won't turn into jam. - [Linus] So there's that. Really?
- Yeah. - Huh, you know what, I
think you might be right, this might actually perform better than putting it in non frozen water. But we could also put it
directly in frozen water. - And maybe kill it.
- Big brain time. That's what the kids
say, right, big brain? - [Brandon] Let's just, you
know, have some safety here. - Safety, I see. You didn't even catch my pun. Safety, I see. Oh, that's chilly boys! - [Brandon] Is it zero
degrees, give or take? - [Linus] It's hard to tell because of like emissivity
issues or whatever. - [Brandon] Are you able to see a cold spot on the MacBook now? - [Linus] It's chilly. Nope, it's still hot. I mean, I'm sure the bottom
is a bit of a different, ooh yeah, the bottom is a different story. - [Brandon] What's the
CPU temperature right now? 1,716, we're sub ambient. - [Linus] Here we go, boys. - [Brandon] What's it at? - [Linus] It still hit 70 degrees but 16, oh, well, it also spiked to 35 watts. - [Brandon] Damn, that's
way faster than before. - Temps are 51 and our package
power is over 16 watts still. It really is a shame how aggressively this
thing gets power-limited. 'Cause even though we're at 41
and a half degrees right now it's still all the way down to
about 13 watts package power. I'm disappointed. I thought keeping the
thermals under control would net us significantly
better performance, this is just power limited. - [Brandon] Yeah. - Did our pump just stop?
- Nope. - It ain't flowin' fast though. I was really hoping it would stay up in the like 15 to 18 Watts neighborhood and we were gonna see like a
really big performance jump. - Yeah, 'cause its up
what, like 40 degrees now? Definitely not thermally limited. - Oh yeah, like the top of
this thing is 24 degrees. It's cool to the touch. So you can see this look, Alex's hand is actually warmer than the surface of the laptop. Hey, do you have less
blood flow in your pinky? I mean, we learned a lot of things today, just I didn't expect this. - I think that none of the things that we learned today we expected. - The score is 1,155. So there was no real benefit from going from just putting a thermal pad on the CPU that helps the bottom of
the chassis dissipate heat to putting an ice-cold water
block on the bottom of it, that's disappointing. We didn't get the result
that we expected or wanted but we still learned a lot today. One, is that the MacBook Air is gonna be in a completely different
performance class from the MacBook Pro, no matter what kind of
cooling you put on it because A, even though they've both got four cores and eight threads, it is a different model of CPU and B, because the MacBook Air just doesn't have the
power delivery robustness to keep up with allowing that CPU to really stretch its legs. Number two is that the cooling
solution on the MacBook Air is even worse than we thought, to the point where even just removing a little bit of material from the already miniscule heat sink gave us a 20 plus degree
improvement in temperatures while under sustained load. Unfortunately, like I said, we didn't get the performance improvement that we wanted from that,
but that's how she goes. Just like she goes into her sponsor, Ting. There's no contracts, no overage fees, and no other carrier tricks. You just pay a fair
price for the talk, text, and data that you use each month. It's especially great if you
happen to be at home a lot or on an irregular
schedule because that way you're not paying for mobile
data that you're not gonna use 'cause you're gonna be
on Wi-Fi all the time. Ting gives you complete
control over your account, you can set alerts and caps for
every device on your account and they've got nationwide LTE coverage on T-mobile, Sprint, and Verizon, so they've gotten great
coverage from coast to coast. Almost any phone will work with Ting but you can check your
phone's compatibility at Linus.ting.com, we're
gonna have that link below, and if you use our link you'll
also get $25 in free credit. Hopefully you guys enjoyed this video, if you like these kinds of
wacky laptop cooling experiments maybe you'd enjoy our five gigahertz laptop with blowimatrons
hooked up to a shroud blown through the bottom of it, it's a gaming notebook, pretty crazy, it's definitely worth the watch. Man, I really wanted to post some crazy performance numbers today but we just couldn't
do it, but it is cool. - [Brandon] Yeah, it's
a 10 degrees right now. - Get it, it's cool.
- Ugh.
TL;DW: While water-cooling the MacBook Air, the CPU only ran at 13W of power under load despite being at ~40C. The Air is not only thermally constrained, but power constrained too
Here are their final testing results
The lack of clamping between the heatsink and the CPU die was actually pretty surprising. I wonder what the rationale for that was.
I didnโt put the video title because itโs most likely gonna be changed within a few hours and I didnโt want this post to become outdated that fast
My theory goes along with what Linus said in the end of the video: that the air is in a different class than the pro. I think the air is constrained in order to not compete with the pro.
Max Tech did this when it first came out
https://youtu.be/VLKVvpwgGW0
I donโt care if it performs โas advertisedโ, they botched the cooling and it could have performed considerably better.
Whatโs the old and new title of the video?
Anyone have any good suggestions for thermal pads? Itโs seem normal thermal pound wonโt really help :(
My theory about the MBA cooling is a little less tinfoil-y than it is practical: they only had so many engineering resources to go into a MacBook Air, and they all went into the new keyboard. The thermal design (which isn't great) was really designed around the heat and performance of the 2018 model (which most reviewers noted ran silently). The latest intel chips Apple could get didn't't fit the thermal design of the MBA, but there was no compelling business reason to delay the release just to improve thermal performance in a laptop that isn't targeted at performance or enthusiast markets.
I'm not saying that choice is correct, but it might explain a notable lapse in engineering.
Would it be a good idea to put a few layers of graphite pad to give better heatsink contact so the fans fun less loud?