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

[NikoB]

I'm also waiting for an optical universal link at distances up to 50M with bandwidth of 500 Gbit/s at least as, as standard port, for PCs and laptops. But for some reason it is still not there. And it's not clear to me what the obstacle is...

[A]

I, out of the corner of my eye, monitor the used market for x86 and have never seen really balanced x86 models on this market at a low price relative to the initial price of the models. Moreover, they are practically not available for sale on the used market in good configurations. Guess why?

Sure, because you have no clue.

For any laptop battery, consuming more than 50W is a bad load. But we are not talking about PL1, but about PL2. It should always be maximum for the fastest possible response of the system and software. But it's difficult to understand this from the review.

What does this even mean

The author specifically indicated in the review that the tests were carried out in the maximum performance profile, and from the power supply.

Again, there's no maximum performance profile

The fact that Apple has reduced the memory speed in the M3 indicates obvious problems with overheating and consumption (including from the battery) even with the new technical process.

Bs

It's more about torturing a high load with an external monitor and, for example, an eGPU, if it works on Macs with Arm. The screen definitely won't like 50C temperatures next to it. And this must be taken into account in such scenarios.

Bs

The resting temperature is clearly abnormal - you have problems with the cooling system.

Lol no, 50C on E-cores is totally normal working temp. Laptop is cold to the touch.

Dell G5 5587 with i5 8300H

Your 8300H warmed up +20C just to play 4K video )))

Apple equipment in general is more difficult to resell

Haha what? Maybe true for Russia or whenever you are from. Macbooks sell like hot cakes.

[A]

Dell G5 5587 has TB3 and eGPU support, so it won't be difficult for me to connect for example a future GTX5060 with DP2.1+ to support lossless 8k@60fps. And no matter how funny it is,

Your ability to connect something doesn't mean your laptop will manage it. Try connecting a mouse with 1000Hz polling rate first, it'll load up your Intel more than 4K@60 video. )

[Neenyah]

Try connecting a mouse with 1000Hz polling rate first, it'll load up your Intel more than 4K@60 video. )

No? This is my X1 Carbon's CPU usage with a 1000 Hz mouse + Half Life 1 running on external screen. Barely any. With eGPU (3070 Ti) I'm playing Counter Strike 2 (my main game) in 233+ fps at 1440p (capped at 233 actually because it's a 240 Hz monitor), no sweat.

Edit: 4K 60 fps vid on YT (COSTA RICA IN 4K 60fps HDR (ULTRA HD)) gives 13-19% CPU usage, tested now.

[RobertJasiek]

Macbooks sell like hot cakes.

You wanted me to point out to you examples when users write Apple PR. Yours is the first;)

[A]

No? . Barely any. With eGPU (3070 Ti) I'm playing Counter Strike 2 (my main game) in 233+ fps at 1440p, no sweat.

Move it quickly and look at CPU usage when it's idle/moving. Also make sure it's really doing 1000Hz by googling "mouse polling rate checker" and doing stuff on that website.

[A]

You wanted me to point out to you examples when users write Apple PR. Yours is the first;)

Nah, can't prove it, it's subjective personal experience. Macbooks don't really lose price and sell withing one-two weeks. The only actual price drop is Intel macs after platform switch. Feelsbad for guys who got their i9 2019 MBPs just a year before switch for $6500 and now they are $1000-1500 give or take because no one needs Intel except 3.5 guys running Bootcamp for their productivity needs.

[Neenyah]

Move it quickly and look at CPU usage when it's idle/moving. Also make sure it's really doing 1000Hz by googling "mouse polling rate checker" and doing stuff on that website.

devicetests.com/mouse-rate-test shows 1060 Hz maximum rate and 960-980 average rate, CPU usage 2-4% with fast consecutive movements around for about one full minute without pause.

I mean I am literally gaming with that laptop (despite having a desktop too) and have not a single issue performance-wise, high CPU usage would definitely be noticeable in CS2 as it's a fairly CPU demanding game. HWiNFO is always open to monitor stuff, nothing unusual is happening on graphs.

Same with other usage (working in almost whole Adobe CC for living). Using desktop (i7 13700KF + RX 6800 XT) basically only to render vids in Premiere Pro and for After Effects.

[A]

gives 13-19% CPU usage, tested now

And that's with GPU hardware acc?

CPU usage 2-4% with fast consecutive movements around for about one full minute without pause

Well, lucky you ) Probably depends on mouse, my Razer Pro Click heats up Intel machines at 1000Hz )))

[A]

4K 60 fps vid on YT (COSTA RICA IN 4K 60fps HDR (ULTRA HD)) gives 13-19% CPU usage, tested now.

11.1% of SINGLE core, 2% overall, power consumption 13.24W while playing it, give or take (idle avg 11.94W)

[Neenyah]

gives 13-19% CPU usage, tested now

And that's with GPU hardware acc?

Yes, UHD 620. Microsoft Edge browser (so Chromium engine).

CPU usage 2-4% with fast consecutive movements around for about one full minute without pause

Well, lucky you ) Probably depends on mouse, my Razer Pro Click heats up Intel machines at 1000Hz )))

Could be just some bad luck in pairing, different factors but almost certainly not an issue with a CPU or a mouse, something else is there in question 🤔 1000 Hz shouldn't be an issue for any CPU from the past 10-12 years even for weak Celerons from the third gen Intel.

4K 60 fps vid on YT (COSTA RICA IN 4K 60fps HDR (ULTRA HD)) gives 13-19% CPU usage, tested now.

11.1% of SINGLE core, 2% overall, power consumption 13.24W while playing it, give or take (idle avg 11.94W)

Screenshot of both screens at once, both 1440p (external on top, laptop below), USB-C connection (DP); Core Effective Clocks all but one under 500 MHz during the video playback with Core Usage 11.43%, Thread Usage 16.4%, Total CPU Usage 11.4%, CPU Package Power (power consumption) 8.759W to 15.1W.

[A]

CPU Package Power

Mine is not package power, it's overall system power from socket, hard drives, screens, external SSD, etc.
Package power (CPU+GPU+DRAM) while playing it is 0.45W, according to Mx Power Gadget, including 0.09W on GPU.

[Neenyah]

CPU Package Power

Mine is not package power, it's overall system power from socket, hard drives, screens, external SSD, etc.
Package power (CPU+GPU+DRAM) while playing it is 0.45W, according to Mx Power Gadget, including 0.09W on GPU.

Ah, ok then, my bad. Very good numbers there! 😳 I checked mine now and logged these numbers; with 17.591W being about the upper end of Total System Power (only one spike above 20W, everything else 12-18). iGPU 0.551W

[A]

We were promised RAM speeds of 300GB/s for M3 Max. The result of the test is almost 3 times less, because the real efficiency of the memory controller is only 40%. Despite the fact that 400GB/s was declared in the M2 Max - but we did not see tests, only declarations.

Some is reserved for GPU and hardware media engines. You can't get all of it unless you properly and simultaneously stress every module of SoC.
Higher bandwidth isn't there to mindlessly dump RAM at full speed on any request. It's there to have a guaranteed bandwidth to unified RAM for every SoC module at any time.

You became 2x smarter after reading this message, NikoB. No need to thank me.

[Neenyah]

Higher bandwidth isn't there to mindlessly dump RAM at full speed on any request. It's there to have a guaranteed bandwidth to unified RAM for every SoC module at any time.

Tbh here, many people have noticed slower performance because of reduced memory bandwidth where M3 Pro is easily losing to older M1 and M2 variants, two of plenty examples:

• M3 Macbook Pro - Unfortunately not what I was hoping for (Oliur / UltraLinx)
• MacBook Pro M3 (14 & 16): You're Being Misled (Just Josh)

Apple is basically forcing people to upgrade to the most expensive M3 models just to get faster performance than with previous M1 and M2 devices. Just Josh's comment under the first linked video:

Hi Oliur, I was referred your video by a viewer. Yes, as others have said, this operation is heavily limited by Memory Bandwidth. That is the primary issue. When it comes to why the M1 Pro amd M1 Max did not exhibit this behavior, here is my initial hypothesis. When the M1/Pro/Max were released, most programs ran via Rosetta 2, an emulation software. This is due to the switch in Architecture from Intel to Apple M series ARM. The emulation layer would choose what hardware execution path to run, and in your case, may not have been using the right one. Over time applications switched to running native, and therefore using the hardware acceleration available to them. By the time M2 was out, this Memory Bandwidth limitation for video editors was on full display. You can see me demonstrating this issue on the older M2 Pro vs M2 Max here: https://youtu.be/OezOluKHyVw?t=271 (sorry to link to my channel, feel free to remove if you are uncomfortable). With regard to GPU cores. These are used to accelerate vector operations used by 3D rendering (and even crypto mining). Not those doing 2D rendering, which you are doing. Not all rendering is the same type of rendering - Hope this helps.