馃攳
Is M1 Max worth $400 extra? - MacBook Pro 14" - YouTube
Channel: unknown
[0]
- Maxed out Macs or Max
Macs with M1 Max to the max.
[6]
The 14 inch MacBook Pro
represent double the GPU
[9]
and media horsepower the M1 Pro models.
[11]
But this raises a very
interesting question.
[15]
How can Apple fit that much more power
[18]
on this same architecture no less,
[20]
into the same chassis.
[22]
And for that matter, is
there any other advantage
[26]
to kitting out a MacBook Pro
[27]
with an M1 Max or is it only really useful
[29]
for video editors thanks to
its expanded media engine?
[32]
The answer may surprise
you, I know it surprised me.
[36]
But it's no surprise that
today's sponsor is Jackery.
[39]
Thanks to Jackery for
sponsoring this video.
[41]
Their Explorer 1500 power station
[42]
provides a huge 1500 watt hour capacity
[45]
with up to seven devices
charging simultaneously,
[47]
and it only takes four hours
to recharge from zero to 80%.
[52]
Get 10% off with code Linus
Tech Tips at the link below.
[55]
(upbeat music)
[64]
From the outside, there's
very little to distinguish
[66]
the M1 Pro from M1 Max.
[69]
So we (indistinct) space
gray for the MacBooks
[71]
and silver for the Pro
Books to tell them apart.
[74]
Jonathan Horst over at
Mac Address commented that
[76]
he liked silver better than
space gray this time around
[78]
and honestly, I kinda agree.
[81]
It's not that the space gray looks bad,
[83]
but there's too little contrast for me,
[85]
particularly considering
the blacked out keyboard
[87]
that Apple chose for the
new generation MacBooks.
[90]
But this is a matter of taste
[92]
and I'm sure there are plenty of folks
[94]
who'd like to see this thing
totally stealth instead.
[96]
There is one major external
difference between the M1 Pro
[100]
and M1 Max though, and
that's in the display support
[103]
where the M1 Pro can handle up to
[104]
two external Pro Display
XDR via Thunderbolt.
[107]
The M1 Max can handle up to three
[108]
while also using HDMI 2.0
port for up to 4K60 output.
[112]
That extra flexibility could
lead to some impressive
[115]
battlestation setups for
professional workloads.
[117]
You've got the integrated
mini led display,
[119]
a span of three big
displays worth of workspace
[122]
and a large TV or color
calibrated 4k display
[125]
for output premium.
[127]
One question that's been asked a lot
[128]
is whether the Thunderbolt ports
[129]
could handle an HDMI 2.1 adapter,
[132]
and unfortunately ours
didn't arrive in time
[134]
for testing these laptops.
[136]
I'll test them all when we
review the 16 inch models,
[138]
so make sure you get subscribed
for that, it's coming soon.
[141]
(upbeat music)
[144]
How about on the inside?
[145]
With these machines being
otherwise so similar,
[148]
could Apple also have given them
[150]
the exact same cooling setup?
[152]
Remember the M1 Pro
caps out at 16 GPU cores
[154]
while the M1 Max can double that number
[157]
and doubles the memory bandwidth,
[158]
which together are sure to
increase the heat output.
[161]
Let's pop the hood and compare the two.
[163]
Oh hey, Apple color matched
the screws, how thoughtful.
[166]
And they're the same picture.
[169]
If I hadn't told you that the
space gray was the M1 Max,
[172]
you wouldn't have been
able to tell the difference
[174]
just by looking at them,
[175]
unless you looked really
closely at that heat spreader.
[178]
It's just a little bit wider on the M1 Max
[180]
to cover that chips larger footprint.
[182]
Other than that, same band,
same heat pipe, same everything.
[186]
Given how hot the M1 Pro got,
[188]
this makes me worry that we'll see
[190]
some significant
throttling with the M1 Max.
[192]
We'll button everything
back up and give it a go,
[194]
this time with the
Zephyrus M16 repping the PC
[197]
with it's core i9 and RTX 3060 Mobile.
[200]
(upbeat music)
[203]
First, we'll see if anything
changed for the CPU cores
[206]
on the M1 Max.
[207]
And predictably, it absolutely did not.
[210]
None of the M1 Max's CPU scores
are appreciably different
[214]
in Cinebench R23
[215]
and this translates over to GeekBench
[217]
where we're effectively
looking at run to run variants.
[220]
Same deal with Blender,
[221]
where we're stuck with CPU rendering
[222]
until Blender 3.1 releases
[224]
with middle GPU rendering
and support sometime in 2022.
[227]
You don't have to wait that
long for our new shirt designs
[229]
and a new water bottle
from Lttstore.com though.
[232]
They're available right now.
[234]
You may be surprised by
these results though.
[236]
Given how much faster
the system memory is,
[238]
I mean Ryzen would sure love
to have memory that fast,
[242]
but it seems Apple CPU cores
[243]
just aren't taking full advantage of it.
[246]
Based on these results alone,
[247]
you might not be too impressed
by the M1 Max at all,
[251]
but the CPU's only half the picture.
[253]
No, we don't have more CPU
cores or even faster CPU cores
[258]
but what we do get are more GPU cores
[261]
and an expanded media engine
with double the encoder blocks
[264]
and double the ProRes blocks.
[266]
Nowhere would this be more
apparent than Final Cut Pro.
[269]
For this test,
[270]
we set up a five-by-five grid
of 4k plus ProRes RAW files
[273]
and set them to appear one
by one in the timeline.
[276]
Then we set Final Cut to stop
when it detects a drop frame.
[279]
With the original M1 MacBook Pro
[281]
stalled at just four streams
[284]
and the M1 Pro dropped at seven,
[285]
the M1 Max's managed 10 and 12
[288]
suggesting that the cutdown model either
[289]
has a slightly less capable ProRes engine
[291]
or the extra GPU cores
on the high end model
[294]
are picking up the slack.
[295]
Whatever the case,
[296]
this is a truly ridiculous
number of pixels being processed
[299]
all at once with no stuttering whatsoever,
[301]
like four or seven simultaneous
streams is already a lot.
[305]
But let's say you don't run final cut,
[307]
you're a Resolve shop.
[309]
Well, the good news for you too.
[311]
When rendering the ProRes 422 HQ
[313]
to take advantage of the ProRes blocks,
[314]
the higher end model gets scary close
[316]
to a sub 10 minute mark out
a timeline that is stacked
[319]
with de noising, Chroma
keying and filtering.
[323]
Not a bad result,
[324]
and our timelines frame rate keeps up to
[326]
where we observed as high as
16 FPS for the high end model
[329]
scaled by about two FPS
for each step down you take
[332]
from there.
[333]
The Zephyrus meanwhile
can only roughly match
[335]
the lower end M1 Pro.
[337]
That's a great result
[338]
but what if you need to
render some 3d assets.
[341]
Right now, the best
solution on Mac is Cinema 4d
[344]
with a Redshift renderer.
[345]
Surprisingly the jump in
performance from the M1 Pro
[348]
to the lower end M1 Max is substantial,
[350]
likely thanks to the
double memoried bandwidth
[352]
to go along with the extra GPU cores.
[354]
The higher end Max gives us
nearly a 10 minute render,
[357]
which seems impressive up
until the Zephyrus comes along
[360]
and smacks it down with a 539.
[363]
Of course, that is a 16 inch PC laptop.
[366]
But, spoiler for the 16 inch review,
[370]
the 16 inch M1 Max doesn't
do that much better.
[373]
We were told that the M1 Max
[375]
could outperform the RTX 3060
[377]
or even match the RTX 3080,
[379]
and Apple calls out Redshift
on their product page.
[383]
So, M1 gotta take the L in this one
[387]
GeekBench too, shows the
Zephyrus and its RTX 3060
[390]
absolutely pummeling the M1
Max in both render modes,
[394]
which places the max as GPU horsepower
[397]
further behind than Apple
had led us to believe.
[399]
Maybe there's more optimizations
[400]
that still need to be
done for non-video tasks
[402]
on these GPU cores,
[403]
or maybe it's just the media engine
[406]
was doing all the heavy lifting
[408]
in our video editing tests.
[410]
To be clear, it is scaling
relative to the M1 and M1 Pro,
[413]
but it's just not reaching
those dedicated GPU levels
[416]
of performance we kept hearing about.
[419]
Maybe they meant games, yeah no.
[422]
Again, we're seeing mostly linear
[424]
performance improvements across the lineup
[425]
in Shadow of the Tomb Raider
[426]
where (indistinct) makes
use of Apple's metal API
[430]
but it's not enough to
catch up to the Zephyrus,
[433]
at least not at 14 inches.
[434]
A few other games use metal,
[436]
so it's all downhill from here.
[437]
And when we switch over to CS GO
[439]
where there's no difference whatsoever
[441]
between the M1 Pro and Max.
[443]
Ironically this time, not
because it's using OpenGL
[445]
but because we've hit a
CPU bottleneck at 100 FPS
[448]
thanks to Rosetta.
[449]
Since we're looking for more ways
[451]
the M1 Max can differentiate itself,
[453]
there's one final ray of hope.
[456]
The venerable Dolphin emulator,
[458]
which is native Apple silicon support
[460]
via a universal binary.
[461]
Unfortunately, Dolphin doesn't
yet support the Metal API.
[465]
But in testing,
[466]
we did discover a number of
interesting points to cover.
[469]
The first head-scratcher is that
[471]
the high-level OpenGL renderer
[472]
is more performant than the
closer to middle Vulcan API.
[476]
This may be because Apple
never embraced Vulcan
[479]
necessitating the use
of a third-party library
[481]
called MoltenVK, if developers
want to support it on macOS.
[484]
Because Apple deprecated OpenGL
[486]
several macOS revisions ago,
[488]
the implementation while performant
[490]
is missing some new features
that makes Dolphin think
[493]
that the GPU might be underpowered.
[496]
Second, multi sample anti-aliasing
[498]
that can be enabled in most games
[500]
is a huge performance
killer and Apple Silicon.
[503]
It easily takes rendering speed
[505]
from an impressive double of
native speed to just half.
[508]
Third, some titles like Rogue Squadron III
[511]
depends so heavily on the CPU
[513]
that they never run at full speed,
[515]
even at native resolution
just like the original M1.
[518]
This makes sense though,
[519]
because the main differentiator between
[520]
the M1, M1 Pro and M1 Max CPU cores,
[523]
is that there are more
performance cores available
[525]
and Dolphin doesn't utilize
more than a couple at a time.
[529]
We did get performance data however.
[530]
B|y firing up Super Mario
Sunshine at 8X Native resolution
[533]
and an uncapped frame rate
[534]
to place the burden squarely on the GPU.
[537]
Bearing in mind that the
game is native to 30 FPS,
[540]
the starting area runs at about 67 FPS
[542]
on the lower end M1 Max
and 77 on the high end.
[546]
That's a pretty good
improvement from the M1 Pro
[548]
that topped out at 47 with the original M1
[551]
taking up the rear at the sub-native 21.
[554]
Except then we look at the Zephyrus again
[557]
and it's a slaughter.
[558]
That right there is over four times
[560]
the native frame rate for this game
[562]
and nearly double what the
best M1 Max can put out.
[565]
In laptop that costs a little
over half as much at retail.
[568]
(upbeat music)
[574]
So again, we're seeing
differences between the chips
[578]
and their performance is objectively good,
[580]
but we're just not coming
up to the performance claims
[583]
Apple made during their unleashed event
[585]
no matter what we do.
[586]
We're legitimately not sure
[588]
if this is an optimization problem
[589]
or if Apple over promised
with cherry picked data.
[592]
Because as we've seen,
it's quite difficult
[594]
to find anything that runs
Metal on Apple Silicon
[597]
to take full advantage of the M1 Max.
[599]
Now let's take a look at
thermals for a test we could run.
[602]
In our M1 Pro video, we focus on the CPU
[605]
so we'll be looking at
the GPU this time around
[607]
for the M1 Max by running Redshift.
[609]
The fact that it hovers around 90 degrees
[611]
isn't too surprising.
[612]
But given how much surface
area the GPU occupies,
[616]
the rest of the SOC
[617]
is also brought up to those temperatures,
[619]
which indicates that we're
probably at the threshold
[621]
of what the cooler can manage
without maxing out the fans.
[625]
Those only ever got to around 50% speed.
[627]
So there's definitely still some headroom
[629]
if you want to max the fans out manually.
[631]
We'll be doing a more strenuous
[633]
combined CPU plus GPU torture tests
[635]
when we test out the 16 inch
form factor in our next review.
[637]
So, stay tuned to find out by how much.
[640]
When we look at the FLIR, it
becomes clear that the M1 Max
[643]
is really pushing the upper limits
[645]
of the 14 inch chassis cooling capacity.
[647]
Easily hotter than even the CPU test
[649]
we did on the M1 Pro chips
in our previous review,
[652]
we're seeing over 50
degrees in the hotspot
[654]
and well above that
exhausting out of the display.
[657]
The bottom is still within
the relevant comfort
[658]
and if it's sitting in your lap,
[660]
chances are the undersides hotspot
[662]
won't be touching your skin,
so points to Apple there.
[664]
Now let's talk battery.
[666]
Last time we showed that
the M1 Pro's endurance
[669]
was nothing short of amazing
[671]
while doing YouTube video playback.
[673]
Easily beating out contemporaries
[675]
like the Surface Laptop Studio
[676]
with a literal all day
battery life result.
[678]
For the M1 Max, we matched all the screens
[681]
at four ticks of
brightness and let them go.
[683]
This time around, both
of the M1 Pro models
[685]
lasted a little over two hours longer
[687]
than both of the M1 Max models.
[689]
So while battery life is still phenomenal
[691]
at about 20 hours in this test,
[693]
you do give up a little bit of endurance
[695]
for the extra horsepower
[697]
even when you're not making use of it.
[698]
Depending on your priorities though,
[700]
this could be a side grade,
a net benefit or a net loss.
[704]
A fun fact to this test, failed once.
[707]
And these endurance
figures being what they are
[708]
made that a huge pain.
[710]
Overall for an extra $200
for the 24 core variant,
[713]
M1 Max is not a bad upgrade.
[716]
It's just difficult to
recommend in general,
[718]
unless you've got a need for
more than 32 gigs of memory.
[722]
The extra GPU, horsepower
and memory bandwidth
[724]
seem very situational,
[725]
particularly in light of the dearth
[727]
of Apple Silicon native
applications with Metal support.
[730]
This may change moving forward,
[731]
but it's a teething pain
that's going to hamper
[733]
these higher end SOC for the coming years.
[735]
Rosetta can only do so much as we've seen.
[740]
As such, M1 Max only really shines
[743]
in high-end video editing as of right now.
[745]
For everything else,
[747]
we're looking at something much more akin
[749]
to the RTX 3050 TI, much less
the RTX 3080 we were promised.
[754]
Apple's leaning heavily on
special purpose Silicon here.
[756]
And while that's an approach
that may work out for them,
[760]
they run a very real risk of
having purpose-built hardware
[762]
that becomes dead weight once
consumers inevitably move on,
[766]
just like you should move on
from untidy nether regions
[771]
with Manscaped.
[773]
Manscaped provides an
all-in-one grooming kit
[775]
that has you covered from head to toe.
[778]
Their performance package
4.0 features their awesome
[780]
Lawnmower 4.0 waterproof body trimmer,
[783]
their Weed Whacker ear and nose trimmer
[785]
plus a whole lot of other goodies as well.
[788]
For a limited time, you get all
of this plus two free gifts,
[791]
The Shed travel bag
[792]
and the Manscaped
anti-chafing boxer briefs.
[795]
Visit manscaped.com/TECH,
[797]
or click the link below for
20% off and free shipping.
[800]
Thanks for watching guys.
[801]
The 16 inch review is going
to include both processors.
[804]
So go check out a review
of the M1 Pro MacBooks
[807]
for a little more on the whole
platform before the finale.
Most Recent Videos:
You can go back to the homepage right here: Homepage