Apple MacBook Pro 14 2023 M3 Max Review - The fastest CPU in a 14-inch laptop

[A]

As to why you are not surprised?

I'm not surprised because it's quite obvious his messages are machine translated and I've suspected them having some badly translated russian figures of speech. Untranslated word simply confirmed my assumption. Aaand this means he never really understood a single review he writes about. And explains why he never watches videos in english. And explains why he barely ever understands responses to him and just jumps to 1-2 line insults every time. And explains why he is always using higher prices.

And as a bonus his self-proclaimed multi-year IT career is a lie of course, because one can't be in IT and skip on english.

[Plum]

@A

I've suspected them having some badly translated russian figures of speech.

which ones for example?

[A]

which ones for example?

Tell the grandmothers on the benches

And many more, not going to spend my time digging through messages of this clown.

[Neenyah]

A $1200 laptop with a 7945HX+4060 is significantly faster than the $8,000 top-end Mac 2023. And that's it. This is a sentence. =)

You make me laugh, pathetic bot, this is L5Pro from Lenovo.

True, the 7945HX is certainly much faster than the M3 Max 16C, but there is one problem though - that L5 Pro is not available to buy anywhere (including the US) and a cheapest 7945HX is starting at €1,799. Just "a bit" over your $1,200.

No clue where do you even find such prices because they are not on Amazon, not on eBay and definitely not even on Lenovo's website where even 7645HX is $58.55 above your price (and that's without taxes).

I mean yeah, Lenovo is going to offer their top spec'd L5P AMD model (4060 and 4070 are the possible pairings) for cheaper than they sell their LOQ line, yep - all legit there. Not. It's like Mercedes pricing their C class below their A class.

[A]

True, the 7945HX is certainly much faster than the M3 Max 16C[/url],

False
browser.geekbench.com/v6/cpu/compare/3766635?baseline=3764696

[Neenyah]

True, the 7945HX is certainly much faster than the M3 Max 16C[/url],

False
browser.geekbench.com/v6/cpu/compare/3766635?baseline=3764696

Passmark is a complete CPU benchmark, Geekbench is great but very specific and it runs very short so many devices can't even hit their thermal limits there. To put it into car terms - Geekbench measures only top speed, Passmark will measure lap times. So you have a hypercar capable to go 420 kph (Geekbench) but a Formula 1 car with its 350-360 kph top speed will do faster lap times (Passmark). M3 Max 16C is certainly a speedy demon in single thread loads, there is no argues about that tho.

-

Edit: Btw, benchmarks are just meaningless numbers in the end. You can't drive a submarine on the road, you can't drive a car underwater. If one's purpose is not met then no benchmark score is going to help there. Buy hardware accordingly to your needs, do not buy hardware after looking at silly numbers in benchmarks and then try to find its purpose. In PC terms - if you need macOS you go with Apple, obviously. If you need Windows you don't go with Apple, obviously again. Neither OS is right or wrong.

[A]

Passmark is a complete CPU benchmark

No. Passmark is a sh*tmark. Either you will have to believe that 'stupid Apple' put in M3 Max 2x weaker integer modules but 5% faster floating point modules than 7945HX or test was optimized for x86 SIMD instructions to run simultaneously 2 or 4 integers. Choose yourself.

Kinda also sus CPU allegedly 2x slower at integer math is 2x faster at prime numbers and physics )

Geekbench is great but very specific

False. Geekbench runs a wide range of real CPU workloads, including a mix of Cinebench/Blender benchmark as "Ray tracer", where 7945HX wins just like in real Cinebench/Blender. So can't say Geekbench is biased, right?
geekbench.com/doc/geekbench6-cpu-workloads.pdf

After they've removed widely criticized workloads (including critics by Linus Torvalds himself) and updated the rest, today Geekbench is by far the most objective illustration of hardware peak performance.

it runs very short so many devices can't even hit their thermal limits there

Because it's not about thermal limits, but about hardware performance. M3 Max doesn't hit them anyway, it's a 115W laptop.

[A]

CPUMark was caught before on manipulating results using SIMD instructions unavailable on other platforms too, around 2020, that time they did it in favor of Intel against your lovely AMD:
guru3d.com/data/publish/205/b8a5fd4fff06dbbc4be758026d311f9759a035/35512_untitled-1.png

They simply added AVX512 optimizations to everything, and, as AMD didn't have that instruction subset, they were immediately pushed out of top charts.

Basically the same situation with the same sh*tmark again.

[A]

Aaaand coming from Linus Torvalds, creator of Linux:

I hope AVX512 dies a painful death, and that Intel starts fixing real problems instead of trying to create magic instructions to then create benchmarks that they can look good on.
...
Because absolutely nobody cares outside of benchmarks.

[Neenyah]

So only relevant benchmarks are those where Apple wins but those where they lose, such as Cinebench R23, are invalid and worthless? 🤔 Even here on Notebookcheck's comparison across multiple different benchmarks it's the 7945HX that wins again.

Because it's interesting how Geekbench was also "Shitbench" prior to version 6 considering that M1 and M2 weren't topping any charts in Geekbench 5 and now suddenly hop - everyone's workflow is running nothing but Geekbench 24/7, hm? What if you need 3D modelling in Cinema 4D because that's what you do for a living? Would you value Geekbench 6 or Cinebench R23/2024 as more relevant for your usage?


Actually here is one good and objective comment from Reddit, regarding Geekbench:

Geekbench is a benchmark that is testing something that Apple's chips happen to be particularly good at. It's not "bias", it's just... what the test is testing. Geekbench tests short, bursty workloads that are common for regular consumer use of their devices. Apple knows their target audience very well, and knows that targeting that kind of workload is what is going to give their users the best experience. So their stuff is obviously going to be designed to excel at consumer tasks. Which Geekbench results verify. That's not to say that they're only good at one specific test/benchmark, just that it's a key performance area for their designers. Of course they're going to be good at it.

As far as whether geekbench is 'biased' or not, consider this analogy. If you are comparing a dragster to a semi truck, A 0-60mph acceleration test isn't inherently biased towards presenting the dragster as a "better" vehicle. Likewise, a towing capacity test isn't "biased" as showing the semi truck as better. They're just data points. Being better in one doesn't necessarily mean the vehicle is better overall. And if I, the purchaser, really just need a minivan to drag around 4 kids to soccer practice, then both vehicles are poor choices and neither test tells me anything definitive towards my decision.

But how do you design a "performance as a minivan" test objectively? Well... you can't. You can test fuel efficiency, cargo space, passenger space, horsepower, acceleration, cost, safety, and a slew of other considerations individually and provide hard measurements of them. And then compile and weight those results into some kind of "overall" score. But there is no objectively correct weighing of those factors, because not everybody needs or wants the same balance. Weighted "performance as a minivan" results are pretty irrelevant if what I actually do need is a semi truck, or a dragster.

There is no one universal benchmark of performance. There are many kinds of tasks and individual tests that need to be weighed based on use-case. That weighing and balancing of different scores is where nuance (and thus, necessary bias) comes in.

In the end, as I said before:

Edit: Btw, benchmarks are just meaningless numbers in the end. You can't drive a submarine on the road, you can't drive a car underwater. If one's purpose is not met then no benchmark score is going to help there. Buy hardware accordingly to your needs, do not buy hardware after looking at silly numbers in benchmarks and then try to find its purpose. In PC terms - if you need macOS you go with Apple, obviously. If you need Windows you don't go with Apple, obviously again. Neither OS is right or wrong.

...AKA different horses for different courses.

-

Edit: Not really sure what does this mean:

favor of Intel against your lovely AMD

Which is why I said that benchmarks, including those 7945HX vs M3 Max 16C, are just meaningless numbers and nothing else. In many benchmarks AMD is wiping the floor with Intel yet I use Intel for my work (to earn living) because despite lower benchmark numbers it is considerably faster for the same money in the same type of work (I do animaton/motion graphics, 3D, video editing and graphic design).  That PC is strictly for work so it's in my workplace. At home for gaming I have a nice AMD build and that AMD, despite weaker benchmark numbers (including Geekbench) is faster in Counter Strike 2 than my i9 14900KF build in the office. I don't play other games (semi)competitively nor I earn any money from them so I don't care about them in terms of performance, only CS2. I also have an inexpensive Beelink mini-PC and my trusted ThinkPad X1 Carbon with 1.1 kg of "weight" to carry it on the go. I don't have "my lovely" anything in tech except my wallet where I value performance per $ in my everyday usage; I couldn't care less if the name is AMD, Intel or Apple, I care about getting the most out for my money and I won't take one isolated benchmark as a determining factor to make my purchase-decision.

[NikoB]

And as a bonus his self-proclaimed multi-year IT career is a lie of course, because one can't be in IT and skip on english.

Most Americans are so stupid that you are a clown, only confirming this honorary title.

True, the 7945HX is certainly much faster than the M3 Max 16C, but there is one problem though - that L5 Pro is not available to buy anywhere (including the US) and a cheapest 7945HX is starting at €1,799. Just "a bit" over your $1,200.

Right now I can order L5Pro 7945HX+4060 from many sellers from China for $1200-1300, with free delivery within 2 weeks maximum.

Prices in China for Chinese Lenovo models can be easily checked even by a child.

You can, of course, expect manna from heaven in the EU and the USA, but whoever needs it (they suited them according to their characteristics) bought them a long time ago.

Also, yes, a top-end Apple laptop for $9,000 completely drains a $1,200 laptop in terms of processor performance from the power supply, which was originally intended, but clown "A" keeps talking nonsense about running on battery power, which in such machines is essentially useless to anyone . Delivering 64GB of memory will cost less than $250, i.e. the price, taking into account the installation of 64GB, as in the M3 Max configuration from the reviews here, will be no more than $1500. If desired, it's easy to add 2 SSDs of 4TB each for $220-250. Total sum of 7945HX+4060+64GB/8TB ~ $2000 vs Mac 16 Pro with M3 Max 8000-9000$.

This is a complete failure for Apple's fans. The Chinese just laugh at them.

[Neenyah]

Right now I can order L5Pro 7945HX+4060 from many sellers from China for $1200-1300, with free delivery within 2 weeks maximum.

Prices in China for Chinese Lenovo models can be easily checked even by a child.

Mmm, "prices in China for Chinese Lenovo models" are being listed in USD. In China. Sure.
Btw, you live in Somalia or something similar where import taxes are not a thing? I mean you can literally be in China and order something from the US or the EU - and you will pay hefty imports. So...

-

Edit:

I went to check your claims, NikoB, and, surprise, surprise, you are lying again.

The mentioned model is ¥10148 and that is $1,427.06, not $1,200 as you claim and that's the cheapest Lenovo with 7945HX altogether. Screenshot 😀😀

[A]

Because it's interesting how Geekbench was also "Shitbench" prior to version 6 considering that M1 and M2

Are you arguing with Linus Torvalds? ) He was criticizing Geekbench 5.5 for having en encryption test and basing benchmark around it specifically. x86 chips simply added hw acceleration of AES to cheat in it. ) Also 5.5 was bit outdated and too small of a load for modern PCs (compilation test was only 1024 lines of code, sic). So it was simply updated to remove x86 cheats, made workloads more lifelike and longer, etc.

I am always objective on hardware cheats, much like we yesterday discussed Apple cheating in Video playback test by hardware codecs implementation. Your 4K@60 HDR Costa Rica video uses 0.9W package power on BASE ENTRY M1 MBA to play.

So only relevant benchmarks are those where Apple wins but those where they lose, such as Cinebench R23, are invalid and worthless?

Again, Cinebench benchmark is also part of Geekbench testing and Ryzen wins there:
"Ray Tracer
This workload renders the Blender BMW scene using a custom ray tracer built with the Intel
Embree ray tracing library."
That's EXACTLY what Cinebench does, with the same Intel Embree library, and your Ryzen wins in it no questions asked.

Geekbench is a benchmark that is testing something that Apple's chips happen to be particularly good at

There's a TON of chips having higher scores than Apple. 7945HX is just worse. 7950X is better, shall we say 'Geekbench is a benchmark that is testing something that 7950X happen to be particularly good at' now? ) It's not serious, their test methodology is known and widely accepted.

Which is why I said that benchmarks, including those 7945HX vs M3 Max 16C, are just meaningless numbers and nothing else.

Geekbench uses real workloads from everyday use.
geekbench.com/doc/geekbench6-cpu-workloads.pdf

performance per $

I'm more into 'comfort for $' in 2023.

[A]

Total sum of 7945HX+4060+64GB/8TB ~ $2000 vs Mac 16 Pro with M3 Max 8000-9000$.

And you'll just get a laptop that is worse in everything.

Will you show us this price already somewhere or it's your usual bs and it will be more expensive and you'll have to pay how much, +30%? +50%? When you import it.

[A]

Most Americans are so stupid that you are a clown, only confirming this honorary title.

Despite you are a local clown, i know you are learning something from my posts (you will never admit of course) and becoming smarter. I'm satisfied.

Btw what are you even doing here if you have to machine translate back and forth? Did they kick you out of all russian tech forums and you've found an unmoderated place? )) Such a great loss for russian tech community (no).