Yeah, my reaction the whole time to reading OP is, “now do performance per Watt…”

The chips it is being compared to are a year older. If there wasn't an advantage somewhere, it'd be an absolute failure.

Intel and AMD sell laptop chips with comparable performance per watt to M2 laptops. If you're buying an M2 Ultra desktop computer, your primary concern is performance period, not performance per watt.

But many of us don't want that performance to come in a large tower that requires fans to be constantly running.

I own a Mac Studio and PC 10980xe with 4090 and the latter has to live in a separate room because of the noise.

I don’t know what tower case you are using, or what cooling system you have set up, but in modern times, if you hear much noise from your cooling solution, you have chosen components poorly. From fans to pumps to dampened cases, you could have build a relatively silent system.

With a Mac Studio it isn't just relatively silent in almost cases it is silent.

With a PC you still have case fans running because even with modern motherboards they don't support switching them off at low temperature. You need to be running special Windows software.

And to cool the CPU you really need to be running AIO or dedicated water coolers which themselves will have 3 fans almost always running.

The fans in a max studio never actually turn off… they’re just really really quiet, and I’ve never managed to actually make the fans come off idle, even maxing out all cores.

It takes sustained usage for the heatsink to saturate, and then the fans ramp up quickly. They do stay next to silent otherwise.

> cool the CPU you really need to be running AIO or dedicated water coolers

There is no shortage of silent air coolers, such as the NH-D15, that are 20-30 dBA and undetectable unless the only noise in the room. It's usually stock case fans that are the culprit.

If you're running a PC with a dedicated GPU then you will have minimum 6 extra fans.

3 for the GPU and 3 for the case. The latter are pretty noisy and constantly run.

Sure, loud fans are loud but anyone that cares about noise is not using OEM fans, and picking up something from Noctua/Arctic. It's also common to upgrade to 140mm fans which further reduces noise.

You seems to live in the 90s or early 00s, sorry. Consumer fan technology has come a long way since those days. Case sound dampening has also come a long way, so the little noise that fans do make, it gets pretty well dampened inside the case.

> With a PC you still have case fans running because even with modern motherboards they don't support switching them off at low temperature

What do you mean? Most motherboards support sending 0W to fans at given temps. You just install the motherboard software.

The motherboard software that is Windows only.

Or bios, or whatever has, although non-Windows support is generally bad in hardware because it requires open sourcing drivers.

I am not familiar with any motherboard that does not expose all its capabilities through BIOS or UEFI. Yes, it might be less convenient to have to reboot to set a fan curve, but it’s certainly possible. And the Windows software is so terrible anyway, going through BIOS is always my first choice.

> even with modern motherboards they don't support switching them off at low temperature

I have a ASUS TUF GAMING B550M-PLUS WIFI II motherboard in my desktop set to do exactly that, under I think 48 °C it stops the case fans. No Windows software required, it’s all in the MB’s BIOS.

The Mac Studio in particular is a terrible example to use here. At least the first iteration of it. It has a fan that is constantly spinning, and it’s a cheap fan, so it creates a very annoying whine sound at low RPMs:

https://www.macrumors.com/2022/05/09/mysterious-noise-sparks...

Hopefully, Apple has solved this issue with the second iteration.

How many hours of research and people on forums yelling at you do you have to go through before you do that?

I built a fairly stock PC, it is not silent and needs BIOS undervolting and custom fan curves to get there, and most recently I found updating the BIOS erases all of that.

Also built a NAS (which is too loud so I'd like to rebuild it) and the one time I reported a bug to FreeNAS a graybeard yelled at me on Bugzilla for not building a 1U server with fans that sound like an airplane taking off.

Really should be no yelling! If the place you are visiting has assholes yelling, just leave for another. But like any major purchase, it takes research. I do this research every 4-5 years or so, when I will be undergoing a desktop replacement. I’d choose the CPU and GPU first, then investigate cooling solutions. Liquid cooling is not always optimal with regard to noise, so another decision to make before committing on a solution is if you’ll be overclocking or not, as there are air-based cooling solutions that could be silent, and give you all the TDP cooling your CPU needs. After you’ve chosen your cooler (liquid or air-based), investigate low-noise, high-airflow fans at the size your cooler accepts. Going the largest fans is not always the best choice for optimal airflow or noise, so factor that in the investigation. I have less experience with custom GPU coolers, as I’ve only used stock coolers—those are pretty silent and well optimized these days.

The ultra isn’t in laptops. The ultra in the Mac Pro starts at 7 grand, if you’re going to pay a 4 grand premium, is 200$ of extra power over its lifetime really that important?

I have a m2 Max, the battery life is awe inspiring. M2 ultra doesn’t best the competition for its use case.

Don't waste your money on the Pro, you can get an M2 Ultra in a Mac Studio for $4000. The only reason to buy the Pro is if you have a bunch of really weird PCIe cards[0].

As for the desktop use case... sure, you aren't going to care about the $200 of saved power draw, but not having a very loud and hot machine at your desk has to count for something, right?

[0] GPUs won't work, their memory access is locked out in hardware and Apple removed the MPX connectors that powered the Intel GPU modules.

GPUs won't work at all? Can't you even use them for ML/compute stuff?

They don't work at all, there are no discrete GPU drivers for ARM macOS.

> The cooling solution and 1250W(!) power supply they carried over from the 2019 model are kind of hilariously overkill now

I wonder if that's a small hint or giveaway of something coming in the future that will make use of it.... One can hope.

If they were going to make a dGPU or other big accelerator in the future then I think they would have maintained support for the MPX form factor cards. The new Mac Pro doesn't support those.

GPUs can work provided the drivers exist.

And those really weird PCI cards are what underpins many industries for which the Mac Pro is designed for e.g. video and audio production.

> provided the drivers exist.

Provided Apple signs 'em :p

Microsoft has implemented driver signing with Windows so not sure what the issue is.

And users can still use unsigned drivers provided they disable the option.

Apple has beef with the worlds largest GPU maker and refuses to sign their drivers because they got screwed by that company over a laptop GPU a decade ago and still hold a grudge.

Microsoft on the other hand, is run by adults.

Where can you get these drivers that exist but just aren't signed?

You can’t because Nvidia gave up and stopped developing drivers for macOS. The OS won’t let you install unsigned drivers.

I know I'm a crazy person, but the idea of having a computer that actually won't let me run whatever I want on it is so nuts.

If that was the only barrier you could sign them yourself. They stopped a long time before that though.

The issue is that Apple explicitly doesn’t want to.

If I understand correctly, accesses to address ranges mapped to the GPU have to be aligned on Apple Silicon due to the PCie controller requiring nGnRE mapping mode for external devices. There's a proof-of-concept AMD GPU driver patch for Linux that kinda makes this work but it's an incredible hack, the performance isn't great and applications still need to be reviewed and patched.

> The only reason to buy the Pro is if you have a bunch of really weird PCIe cards

They’re telegraphing the use case for future SOCs.

> They’re telegraphing the use case for future SOCs.

Can someone explain what that means?

Telegraphing means to give away away a message, inadvertently or not.

While the M2 doesn’t include external GPU support, I think the M3 Mac Pro will in order to continue justifying the astronomical price.

Otherwise, Apple would have gone through an awful lot of trouble to design a new motherboard with PCIe slots for an otherwise limited-use case.

In releasing an M2 Mac Pro, Apple is trying to balance a consistent message regarding the ARM transition and customer expectations, but also the hiccups and limited experience of developing desktop CPUs. There may be some silicon bug or silicon-level design oversight in the M2 preventing use of an external GPU.

Discrete GPU support is supposed to be there, but not quite ready.

I’ll take the under on that.

Thanks. So by "for future SOCs" they meant future iterations of Apple Silicon. And you think the use case the gp means is GPU support? I'm not sure I agree thats what they meant.

I was the GP. I don’t know what else to call the M-series chips :)

A proper cooler won't get loud or hot.

And as far as heating the room, okay that's an extra $50 of air conditioning to remove the $200 of heat.

It is not the dollars per watt that matters to many users.

What matters more than people may realize is battery runtime. The Apple hardware just spanks x86 hardware, unless that hardware is fitted with a big battery, and or a secondary one.

I typically run Lenovo hardware, and on my faster machine, I got a very large battery and could get 6 solid hours. And that is at a nice 3Ghz speed too.

My current machine (i5) is far slower than the M1 and that hot running Lenovo. (i7) It has two batteries that can yield 5 to 7 hours.

Those machines are heavier, slower, bigger, and just feel crappy compared to the M1 Air I have been using.

And that thing is crazy good. Battery life is longer, sometimes by a considerable amount. It is great hardware, fast, easy to use, light, you name it.

If we forget about servers.

Which we should because no one is using a Mac as a server.

They should (or equivalent product). ARMv8 processors are much more secure than x86 and will continue to improve in ARMv9.

People used to run Mac OS X servers, but Apple does not care for that market.

If we only consider web servers.

Mac servers are pretty much required for any kind of CICD for Apple development.

Just about any company with an iOS or Mac app will have or rent a Mac server.

Sure. When talking about desktop, personal computing, servers are not really a discussion.

It is when overreacting the extent of Apple achievements in the industry.

Most people in 2nd and 3rd tier countries don't have the necessary means to buy those equipments, and additionally they aren't a presence on servers and embedded markets.

So in those markets it doesn't matter if M2 Ultra smashes AMD and Intel, no one will ever use one.

I has never mattered. I do not understand your point.

M1 mini will make a great home server, would be nice to see someone do a comparison with similar products like Intel NUC for a home server of that size. I'm using an old NUC as a home server and the only drawback is that it only has the capacity for 2 harddrives.

With 8gb of ram and 256gb of storage? Otherwise it is like $1000. So, yeah that buys you a decent home server, but at that price it is kinda underwhelming considering the limited OS compatibility, ports, size (much larger than a NUC), ...

Also, consider how much a PC laptop has to throttle performance if you unplug it from the wall.

Yes! That is often significant. Sometimes worse than half performance, or it will suck the battery dry fast enough to seriously increase battery life decay.

And don’t forget cooling! All that extra power draw generates heat.

  > And don’t forget cooling! All that extra power draw generates heat.

in turn can cause more throttling if its not cooled (loudly) enough

TFA is about the M2 Ultra chip - it is not available in any system that is battery powered. They are in the Max Studio and Mac Pro only.

Power isn’t just the stuff coming out of the wall, it’s also the heat you have to dissipate.

If the chips are designed to sacrifice performance for power usage, then they have no place in a computer without a battery.

Apple should assume that desktop machines have an unlimited power source, and that users want a maximum speed. That the users time is more expensive than the electricity being fed into the computer.

Laptops, sure, sacrifice power to extend the battery. Apple does an excellent job here.

The only reason you would buy one of these desktop machines is because you want to be in the Apple ecosystem.

Sure they do, same as mobile chipsets end up Intel NUCs. You're ignoring a whole category of devices.

I always though NUC's are designed to be cheap, quiet, and small, not power efficient. I don't think they run on battery power.

It's easier to make something small and quiet if it's power efficient. It will need to dissipate less heat.

I ran one on battery power, IIRC some models run on voltages in the 12V-20V range, so it is pretty easy. Fun project. I got a portable monitor and keyboard, and… realized I’d reinvented the laptop, except it was really hard to balance all that stuff on my lap. Next time I just got a laptop, but I can still use all the old NUC peripherals so that is nice.

99% of the time I'm hooked up to a power source so battery life isn't nearly as important to me as the $$/performance ratio of the laptop.

> it's becoming a lot more difficult to justify the power consumption of the current Intel (and, to a lesser degree, AMD) offerings, especially for sustained personal productivity and burst compilation tasks.

I don't get it.

The M2 is 15W. There are mobile Ryzen APUs with the same TDP. They're about the same speed; maybe the M2 is a little faster for single thread and the PC is a little faster for multi-thread, it's not that different. You can get a PC laptop which is under 3 lbs and has 14+ hours of battery life.

You can also get a PC laptop which is heavier and has a shorter battery life because they have no qualms about selling you what amounts to a desktop CPU which is then something like 30% faster for multi-thread. But nobody forces you to buy that one.

can someone recommend a concrete laptop then, please? I have been putting off the buying of a new one for years now. (I have an old XPS 13 and a P50 with an M1000M, so ... yeah, old dogs!)

I want to put Linux on it. So far I have been looking at the legion 5 pro, because a friend got one recently, and seems amazing, though haven't yet got my hands on it long enough to try it with an USB stick with a recent Ubuntu (with a recent kernel) to test the power management feature support.

From what I’ve read and seen, Linux is the worst option if you’re concerned about battery life.

This is basically down to some hardware vendors writing crappy Linux drivers that don't include power saving features, while also not documenting the hardware so no one else can do it either. But that's only true of specific devices. For example, this is claimed to have 14 hours of battery life, on Linux:

https://system76.com/laptops/lemur

My x270 could pull about 20 hours on Linux when it was new (it did degrade a bit over time, as batteries do). I don't recall if I fiddled with any settings as this was five years ago and the laptop since died in an accident, but I don't think I did other than setting it to "power saving" instead of "balanced".

Before that I had a Dell XPS, which lasted about 7 hours IIRC, which was pretty good for the time.

My Acer could pull about 15 hours when new, but it is a fairly underwhelming Celeron CPU so that probably helps.

Maybe other systems are worse; I don't know, but overall things always seem to work well for me.

Try the starlabs starfighter.

Something without a second GPU. Linux never did quite get those working properly.

Lol, indeed, the Intel chip they use for comparaison is 320W: https://www.tomshardware.com/reviews/intel-core-i9-13900ks-c...

Your use case is watching a movie? Wouldn't this just hit the hardware decoder on any of these chips?

I guess I don't understand why you would want the ultra if you still prioritize TDP over performance.