Hacker News Reader

Comment by rowanG077 | Hacker News Reader

A shame that thermal/power limitations aren't investigated. That is the most deciding factor for me getting a Pro or Max. And something Apple has historically had a lot of trouble with.

View on HN

If I'm understanding you correctly, you're thinking of previous issues with thermals and throttling, but this has been an issue over the past several years due to Intel falling behind AMD and TSMC, and thus driving power through their chips in order to stay competitive, but that generates heat, and ultimately ends up triggering throttling.

If you read about these particular chips, it should be startlingly clear that they are much more efficient than the Intel chips they replace.

In this article:

> Apple doesn’t advertise any TDP for the chips of the devices – it’s our understanding that simply doesn’t exist, and the only limitation to the power draw of the chips and laptops are simply thermals. As long as temperature is kept in check, the silicon will not throttle or not limit itself in terms of power draw.

> The perf/W differences here are 4-6x in favour of the M1 Max, all whilst posting significantly better performance

Read page 3 of this article. They really do cover a lot of this.

It is much more efficient. It's also more powerful. You can see in one of their benchmarks they hit 90+W of package power. I doubt it can sustain this. The "turbo-mode" Apple has announced for the 16-inch version also indicates that it will be significantly thermally limited.

In my Lenovo Legion 5, in performance mode, the CPU is configured to draw up to 70W, and the GPU is configured to draw up to 115W. It's able to do this just fine for gaming sessions. Yes, the fans are quite audible while doing this. I think in contrast, having to handle about half that power draw overall should be attainable. For sure, it's now a very large SoC, so the heat might be a bit more concentrated and require some engineering to cool. But it doesn't seem like it should be a showstopping concern. Of course, you can wait for additional reviews and see if anyone addresses longer, more sustained load testing.

I don't doubt it's physically possible. I doubt Apple implemented it. So I rather wait till there is some hard data. Apple could also have implemented proper cooling for their Intel laptops but they didn't.

That's perfectly understandable. It's a major purchase decision.

But also, I'm not aware of anyone that implemented "proper cooling" sufficient to handle the last few generations of Intel chips (at least at the high end.) I read reviews of a variety of machines. All of them had issues with throttling.

I was so happy when the Ryzen 4000 mobile chips were reviewed and did not require elaborate cooling systems just to perform their regular duties. I would be shocked if the 2021 Macbook Pro 14/16" have issues with thermal throttling.

I haven't seen too many benchmarks yet, but Dave2D ran a "Cinebench R23" test in a both a single benchmark and thirty minute loop. [1] He saw that the score remained the same after the 30 minutes.

He also reported that the loudest fan noise he could get was 38dB, with typical loads under 30dB. [2]

-----

[1] - https://youtu.be/IhqCC70ZfDM?t=360

[2] - https://youtu.be/IhqCC70ZfDM?t=438

> As long as temperature is kept in check, the silicon will not throttle or not limit itself in terms of power draw.

Apple laptops has, for as long as I know, had issues with just thermals. Sometimes they get so hot you can't even have them in your lap, so they are just "tops" at that point.

Has this issue been solved with these new models?

Yes

It was solved with M1 last year

The m1 macs do not have this problem, they draw a fraction of the power of the old laptops even when boosting.

Since M1 is designed to low power (but still impressive performance), it's unsure that M1 Max machine getting very hotter like Intel era or not. It uses 145W AC adapter so I suspect they getting very hotter if it utilized extremely.

This wasn't really a _huge_ problem if you didn't need fast graphics until the 2016 15"es. With the switch to Ice Lake, they had to add a discrete GPU even to the base model of the 15" (due to lack of appropriate GPUs in Intel's bigger mobile chips in that generation), and things got very bad.

I've got a personal 13" 2016 MBP, and a work 15" one. The 13" is _fine_; it doesn't show significant thermal throttling in normal use. The 15"... not so much.

I had every model of the MBP that was released between 2012-2016 and they consistently got too warm to have in my lap, every single one of them. Some of them even when just browsing the web. Fresh install every time I got a new model. After that I simply got tired of not being able to use my laptop as a laptop and jumped ship.

(Er, Skylake, not Ice Lake)

Did you even bother to read the article? They have an entire page on power consumption under various workloads.

https://www.anandtech.com/show/17024/apple-m1-max-performanc...

Can you please quote the part which discusses power and thermal limitations backed up with tests because I don't see any of it on that page.

As far as I read it these test only report package power and wall power used under certain loads. It doesn't say anything about any limitations. No long term tests or temperature graphs. No information at what temperature throttling kicks in. Is CPU temperature truly the only limiting factor or are the VRMs also a pain point? I could go on but I think this illustrates enough of what I want to see.

Their stories span multiple pages, but for some reason people on mobile frequently seem to miss the other pages.

Here's the whole story on one page.

https://www.anandtech.com/print/17024/apple-m1-max-performan...

Yes I read everything. There is no concrete information or tests about thermal or power limit throttling.

Here you go:

>https://www.anandtech.com/comments/17024/apple-m1-max-perfor...

Someone saying without any data to back it up doesn't exactly inspire confidence. Like I said nothing concrete.

> Any pure CPU or GPU workload doesn't come close to the thermal limits of the machine.

So does that mean it is thermally limited on a CPU + GPU workload? What about a CPU + GPU + Media engine workload. What about using the NPU? Does SSD load have an impact? SSDs nowadays can consume 15+W of power. dozens of questions, unanswered by such a short sentence. Please just test it out and give us the data.

>Someone saying without any data to back it up doesn't exactly inspire confidence.

That "someone" is the editor in chief for Anandtech.

He says you need a workload that stresses both the CPU and GPU as much as possible and let that run overnight before the ability to crank the fans higher than normal is handy.

> That "someone" is the editor in chief for Anandtech.

I don't see how that is relevant.

> He says you need a workload that stresses both the CPU and GPU as much as possible and let that run overnight before the ability to crank the fans higher than normal is handy.

Then why is there no data? This is much more interesting then what some of the tests done in the report.

What you are looking for is a Laptop Review, how M1 Pro / Max under the MacBook Pro 14 / 16 cooling works with maximum possible TDP and cooling.

But this isn't a Laptop Review, it is an SoC review. My guess Dave2D will look into these sort of thing since he cares about it before anyone on the internet actually test these sort of things for Laptop. ( Possibly due to various PR restrictions )

>A shame that thermal/power limitations aren't investigated.

It's covered in the comments, along with when the "crank up the fans" mode would be useful.

>Any pure CPU or GPU workload doesn't come close to the thermal limits of the machine. And even a moderate mixed workload like Premiere Pro didn't benefit from High Power mode.

It has a reason to exist, but that reason is close to rendering a video overnight - as in a very hard and very sustained total system workload.

https://www.anandtech.com/comments/17024/apple-m1-max-perfor...

Reply on news.ycombinator.com