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

[A]

And let's also mention that your PC (desktop or laptop) won't die from literal dust

Yeah, but models prior to 2021, so don't worry, you will be fine.

My Blender shenanigans?

Yeah because everyone knows x86 is faster in Blender (you can literally just google 'blender benchmark' and see it, and it was posted out of context.

I'm afraid that you will have to quote me on that, to show my post where I said that because I never even mentioned x86 all, searched even with Ctrl+F, until quoting your own words in my previous post.

You have responded to my message to RobertJasiek where we were discussing efficiency and that it's hard to make a direct comparison because x86 can't run big AI models.

I agree. Which is why the mentioned Strix Scar 17 for $2,000 is less is a much better choice as it's better than an M3 Max Mac no matter how high you spec that M3 Max - even with 128 GB you can freely upgrade that ROG to 128 or more and you can use an eGPU with 48 GB of VRAM - and still be $1,800 cheaper than that fully maxed M3 Max MacBook with higher performance.

Again price comparison shenanigans ), you compare laptop with 48GB VRAM able to run 34B models to 'maxed out' 128Gb laptop able to run 120B models. Please compare it to 64Gb base M3 Max for, i don't remember, $3500?

[A]

It is clearly visible that the whole "discussion" comes down to the maximum removal of the most valuable comments from the top next to the review text. If the editors change the arrangement of comments at the end of the review to normal - the old ones will be first, as it should be, and all the bots will immediately start wagging their tails in anticipation articles, because whoever writes that comment first and will read the majority. The topic is as old as forums and comments on the Internet.

Now someone is removing you, the most valuable member of the community aka clown everyone is laughing at? Man you need medical attention, like seriously, I'm not joking.

[A]

@Toortle please don't miss my other response to you above between this clown's insults.

So that laptop above (Strix Scar 17) can't perform equal number of tasks and productivities? Seriously? And you are dead-serious about you not trying to say that the M3 Max is an ultimate machine better than anything? Nice, early April 1st.

I never said 'M3 Max is an ultimate machine', you made it up. All I said was x86 can't run big LLMs. And yes, seriously, your Strix still can't run it and it's can't do large image Stable Diffusion inferences too.

[A]

P.S.

even with 128 GB you can freely upgrade that ROG to 128 or more and you can use an eGPU with 48 GB of VRAM

Also, maybe you haven't noticed but you again had to basically convert your laptop to desktop AND increase power consumption by 300-400W to make it competitive with 115W laptop.

Maybe it's time to admit that there's a reason for pricing and it's more efficient at comparable (let's meet in the middle here) performance? And that you can't do the same at 56W on x86... Or we're caring about power consumption and climate only on Twitter?..

[Toortle]

I agree. Which is why the mentioned Strix Scar 17 for $2,000 is less is a much better choice as it's better than an M3 Max Mac no matter how high you spec that M3 Max - even with 128 GB you can freely upgrade that ROG to 128 or more and you can use an eGPU with 48 GB of VRAM - and still be $1,800 cheaper than that fully maxed M3 Max MacBook with higher performance.

Again price comparison shenanigans ), you compare laptop with 48GB VRAM able to run 34B models to 'maxed out' 128Gb laptop able to run 120B models. Please compare it to 64Gb base M3 Max for, i don't remember, $3500?

No worries.

Base 14" M3 Max 64 GB is available for $3,899 -> apple.com/shop/buy-mac/macbook-pro/14-inch-space-black-apple-m3-max-with-14-core-cpu-and-30-core-gpu-36gb-memory-1tb

Base 16" M3 Max 64 is available for $4,199 -> apple.com/shop/buy-mac/macbook-pro/16-inch-space-black-apple-m3-max-with-16-core-cpu-and-40-core-gpu-48gb-memory-1tb

1 TB in both, nothing upgradable, 14" being $400 more expensive and 16" $700 more expensive than an upgradable laptop with 7945HX with 32 GB + RTX 4090 16 GB. Upgrade RAM to 128 GB and it is stil cheaper than both M3 Max 64 GB Macs above 😁

Or check this here - even faster CPU 7945HX3D, also RTX 4090 16 GB, 2 TB base storage which is replacable, 32 GB of upgradable RAM for still cheaper than any M3 Max newegg.com/p/2WC-000N-0E5M9

Superior GPU and 40%+ faster CPU than M3 Max 16C:

cpubenchmark.net/compare/5232vs5748/AMD-Ryzen-9-7945HX-vs-Apple-M3-Max-16-Core

notebookcheck.net/Apple-M3-Max-16-Core-vs-R9-7945HX3D_15113_15073.247596.0.html

[Toortle]

Also, maybe you haven't noticed but you again had to basically convert your laptop to desktop AND increase power consumption by 300-400W to make it competitive with 115W laptop.

Maybe it's time to admit that there's a reason for pricing and it's more efficient at comparable (let's meet in the middle here) performance? And that you can't do the same at 56W on x86... Or we're caring about power consumption and climate only on Twitter?..

Wow, Apple fan(atic) talking about climate and power consumption while Apple is the leading tech company in producing e-waste starting with all devices made to be nonrepairable and with soldered everything including the storage.

[A]

And you are talking like it's $400-500 cheaper? Or how much?

2 TB base storage

Replaceable SSD is a pain for M mac owners yeah.

Superior GPU

Superior for gaming and 3d workloads (maybe not so much when Blender devs will finally optimize their software), useless for actual AI productivity in 2023*

Actual data for Blender using their own benchmark looks like this:
NVIDIA GeForce RTX 4090 Laptop GPU    8203.46
NVIDIA GeForce RTX 3080    3172.08
Apple M3 Max (GPU - 40 cores)    3014.83
4090 surely destroys M3 because OPTIX optimizations are in place and Apple Silicon optimizations are sadly not present. (

40%+ faster CPU

As for CPU-only Blender rendering,
AMD Ryzen 9 7900X3D 12-Core Processor    412.81
Apple M3 Max    412.39
AMD Ryzen 9 7845HX with Radeon Graphics    370.47

And Geekbench
2695/15262 Ryzen
3070/21405 M3 Max
Or do I have to use some specific benchmark that's more convenient for Ryzen.

[A]

Wow, Apple fan(atic)

Please, don't go down to 'apple fan' rhetoric when you are running out of arguments, there's plenty of clowns here who will gladly do it for you. We're just having a friendly and argumented discussion, right?

I don't care about Apple itself and no idea if your data is even true, I'm caring about my own power consumption. I know that Apple can buy a working M1 Max laptop out for $1300 toward next laptop no questions asked to not create e-waste, they will probably refurbish it and sell to next user or reuse body or at least aluminum. Will ASUS buy their ROG Strix out? Nope. When you run out of CPU power it goes to landfill. So who knows, who knows, you might be right, but my eyes tell me the other e-waste story.

[A]

P.S. Also side note about gaming and comparison to 4090 - we actually have no idea how it performs because those games in review are not only NOT arm-optimized, they are not even MacOS-compatible games at all. It's _Windows games_ ran through three layers emulation, one of which is translating Windows API to MacOS and another translating D3D calls to Metal.

I kinda myself find it amusing M3 Max is able to run these Windows games @100fps at high settings, idk. I don't game myself though.

[Blendur]

They are always using BMW or Classroom for Blender benchmark cause there other benchmarks do not fit most windows laptop video memory like that big furry dino 25Gb.

[Plum]

@A also the few available games that run natively on M chips perform much better on x86.

[RobertJasiek]

Was off during the weekend so just starting to catch up discussion. But one point is worth closer consideration. Do huge language models when inferred really need much VRAM or would much RAM, say 128GB, with an ordindary amount of VRAM, say 20GB, be enough?

[A]

Was off during the weekend so just starting to catch up discussion. But one point is worth closer consideration. Do huge language models when inferred really need much VRAM or would much RAM, say 128GB, with an ordindary amount of VRAM, say 20GB, be enough?

youtube.com/watch?v=jaM02mb6JFM

Or you can just watch a video of how 4090 is faster on 7B model (noone actually uses those)
slightly faster on 13B model (probably around the best you can run on x86 laptop)
and M Max RAPES 4090 on 70B model (which is de-facto most balanced for running locally today) - and he explains that building x86 rig able to run it on GPU is worth more than M Max and uses way more power than M Max.

Fun starts at 5:47

[A]

@A also the few available games that run natively on M chips perform much better on x86.

I'm not much into gaming and never was looking at it, any links to check out?

[A]

@RobertJasiek replied to you on previous page