MacBook Pro M5 Max - Nuggets from the Reviews

[playtech1]

Overall it looks like the M5 Max is a solid upgrade over the M4 Max, albeit (as usual) probably not worth it to upgrade for M4 Max owners outside of some specific use cases.

The reviews are the usual mix of benchmarks and mostly lacking in real insights, but here are the things I picked up from the reviews that I thought worth mentioning:

- very roughly a 10% increase in CPU performance. IMHO this is impressive (Zen 4 to Zen 5 was about 5% over 2 years and Intel regressed in many areas with 14th Gen to Core Ultra(!)) and I think reviewers are downplaying this because they aren't looking at the stagnation on the x86 side.

- Notebookcheck was probably the most negative, also highlighting the wattage limits the 14 inch model has (72 watt peak / 44 watts sustained in high power mode, with 60/32 in automatic) and saying this is holding back the CPU. Might make the Mac Studio version very interesting if true.

- GPU performance overall improves but seems highly task dependant - 3D Mark sees 8% improvement, whereas other benchmarks show improvements up to 26%. Real world impacts are hard to discern. Tyler Stallman's comparisons are usually quite good for creative tasks but he is only comparing with M2 Max, so not the best for year on year comparisons.

- It doesn't seem like gaming is going to be massively better on M5 Max than M4 Max (but is definitely in the territory of equivalent to a mid-range Nvidia Blackwell discrete laptop GPU - so still definitely 'good enough'). Gaming definitely needs more testing - reviewers seem all over the place on this and I just don't think Mac reviewers are all that experienced in game benchmarking. I'm very interested in Andrew Tsai's take on this in due course as high end Mac gaming relies on Crossover / GPT and this is where I want to see what improvements there are. Also be keen to have Digital Foundry take a look.

- Some AI tasks seem hugely improved - Draw Things saw a massive gain (iPhonedo had it over 2x with Qwen Image 1.0 generation), whereas other LLM tests in LM Studio were a bit quicker, but not dramatically so. It doesn't seem like M5 Max is going to really move the needle versus Nvidia outside of the unified memory advantage.

- seems like at least one reviewer experienced coil whine under AI load (Hardware Canucks at 10:26). Alex Ziskind also mentioned coil whine but it wasn't clear to me if he was actually talking about fan noise.

- a couple of minor performance regressions (I would guess due to the different core configuration) - Tyler Stallman found one in a photo app (Reblum) where an M2 Max was significantly faster than M5 Max, there were other things some components of Geekbench, some AI tests too. Nothing too bad which suggests the move to the new CPU structure has been a success.

Did anyone else glean interesting nuggets?

 

[kasakka]

Display is again poorly, even non-existently tested. Notebookcheck has been historically about the only outlet measuring the display, and that section is absent from their article. I guess they imply that it's the exact same as M4 MBP?

Which means that it remains the display with the worst motion performance on the market. It's a blurry mess in motion, despite being pretty great on every other metric.

Otherwise it's about what I expected, though I thought the GPU bump would make a bigger difference. Instead it seems highly situation specific. Personally I don't give a damn about gaming on Macs as anything but a test point.

Another thing I'd like to see tested is external display support. Not with some Apple displays which "just work", but e.g 5Kx2K ultrawides, high refresh rate displays and others that are known to be problematic. I use a Samsung 8Kx2K which works on my M2 Max MBP 16", but is far from ideal.

For me it's atm between getting a M4 Max for work, or M5 Pro/Max depending on their pricing when my leasing vendor gets them. So far it looks like it'd be M5 Pro < M4 Max < M5 Max.

 

[playtech1]

I hear you - MacBook Pro reviews and Apple reviews in general are typically very superficial. Notebookcheck is I think the best, notwithstanding their omitting the display data (which, to be fair, they may cover later). I miss Anandtech and I also miss this forum being a bit more active with quality posts tbh.

External display support would be great to get data on - Apple's statements as to maximum supported resolutions and refresh rate don't make it easy to work out whether e.g. 4K 240Hz works or scales well.

 

[anticipate]

It’s not that big of a jump compared to last year I’m surprised. Even last year wasn’t a huge change. The really big jump is when they went to 3 nm which was the M2 max to M3 max so they could pack a lot more into the chip. There is one outlier and that’s AI workloads. It’s a much faster chip for AI workloads. I do a lot of video editing and use a lot of AI functionality initially in the edit to like categorize footage to do slow motion and things like that so the new laptop would be a lot faster at that. But typically that stuff runs in the background and is only really rendered once so I don’t see much of a major improvement otherwise.

 

[FlyingTexan]

People are looking at the wrong specs. Yes small bumps here and there, in those tests, but capability wise there's a completely better and different use case. I've never edited a video or really any photos on my mac. I've used my M1 Pro as a daily driver and mostly watching movies and tv while on the road for work. It's just been a great travel companion. However, LLMs are a tsunami of capability and we're still growing into what is even possible. There's an entirely new use case and I think it's going to put the video editing sales pitch of apple waaaaaaaay in the back. AI will be used by everyone. So for me I want to have it to learn and try. The real benefit with it here is the next gen of laptops are going to focus heavily on memory bandwidth. It's like Intel focusing on ghz forever but now everyone is pumping R&D into memory and bandwidth. So for me the Max is completely worth it even though I do zero editing and media because I've been spending every day jacking around with LLMs learning more.