If it’s useful: It’s fine here including reasonable loads on the video card, like a lot of GPU-accelerated terminals and pretty heavy use of a tiling window manager and Exposé window swishing. Or, say, Sketchup in a browser? Plus a pretty serious IntelliJ setup with the JVM’s stops pulled out, running ReasonML and C++ source compilation, and use of Docker.
It is sensitive to setup, and I can get it to spin its fans up. Docker is especially prone to wasted cycles. And I did seem to have to do some SMC resetting at one point because the fans were a little eager to get going. I also think it makes a bit of difference if I charge through the right-side ports rather than the left.
But the main point I’m attempting to make: I do think your case is solvable.
That said, I have felt that Apple are losing their way in some respects. The fact that there are fan issues with some machines and that it’s not clear how to solve them is a problem with these machines.
I think they're referring to the thermal throttling, and not the MBP forcing the discrete card when an external monitor is connected. As far as I know, there is no way around using the discrete card for external monitors (on my 2015MBP at least, I'm pretty sure the HDMI port is wired direct to the M370X).
I've solved the thermal issues by using an eGPU (like GP suggests)
When it works, its great which is about 50% of the time. When it doesn't and for no apparent reason, it will often reboot a couple of times (just when plugging in the displays).
Sometimes the banding issue that the M1 Mac owners have described appears on one or both of the displays.
A pretty frustrating experience.
Already running discrete graphics all the time so no auto-switching.
Switching apps is often plagued with at least noticable up to utterly annoying lag. Starting even optimized apps sometimes feels surprisingly slow. If the CPU is even under the light load (20 % reported by Activity monitor), these issues happen more frequently and overall responsivness gets much worse. It is not due to the low memory which I try to spare on Mini.
Yeah, Safari is fast, even faster than on the iPad Pro. But I am developer, often use multiple browsers, many docker containers, might spin up IDE and I am certainly not the one who they made it for.
For reference, I just opened every application that's installed on my M1 MBP (16GB) and was able to switch between them with basically no lag.
I wonder why you believe that everyone’s ignoring the issues you experience? Could it be some local phenomenon, instead?
Conspiracy? Or...something up with your setup.
If you have an external SSD, you could potentially try a fresh install on it and see if the problem persists, then try to drill down to the offending driver/application.
I for one, have had no issues, but that's not to say they don't exist. Could be a process that is blocking or anything.
ssh-keygen -t rsa -b 4096
This was considerably slower on the M1 Mac Mini than my 2017 iMac.
My solution was to dump Apple laptops entirely (which I know a lot of people won't want to do). This Asus G14 has no issues powering 2x 4K displays without even breathing hard. No throttling in games or videos, and it supports 40GB of memory on top of a replaceable NVMe SSD.
I think several will trade their M1 laptops due to the memory, architecture limitations creeping up after long-term usage and realizing they're compromising their previous workflow's productivity benefits. The excellent performance (CPU/power) and surprisingly well built Rossetta 2 has overshadowed potential limitations with memory and IO in the reviews.
Lightning fast opening of apps is great, but it's not much use to people like me who don't close their apps at all! I keep at least 3 browsers(for different projects) with several dozen tabs(with hibernation), Note taking apps, IDE and a VM with a browser open in it. So what's important to me here is the memory, at least 32GB is bare minimum.
They'll trade them for new M1 Macs that don't have those limitations, which are estimated to be released this year. This obviously isn't a "forever" limitation. And Apple will release Apple Silicon Macs with more memory long before any other Laptop maker releases a laptop that can compete with all the advantages of the M1.
I am curious about the other "USB to PCIe cards" though. I know there are NVMe to USB (non TB) adapters these days but I think that's largely due to the NVMe protocol being transport agnostic not that PCIe is actually interacting over the USB connection.
I use an Intel MBP13 inch with two external monitors and it works pretty well. What am I missing?
And it's not terribly bad, just that I pretty much draw the full 92W the power adaptor supplies and get the mentioned thermal throttling much sooner than without.
Running a bunch of docker containers, slack, and general everyday workflow. Doing all the good stuff like ad-blocker and checking process monitor.
Our solution isn't the most lightweight, but still the end result is that plugging in an external monitor (1080), my productivity goes down dramatically which is the opposite of how multiple monitor setups are supposed to work. :-)
So whatever a “multi-monitor workload” is (video editing or something?) — it’s certainly not universal.
When I'm on a video call, the CPU fans are consistently on at what I believe is the lowest RPM.
I would classify that as 'not breaking a sweat'. I suspect you might as well. I'm not sure if OP expects to be able to use all of GPU memory using only passive cooling, or if they have something else is going on.
I've been using the preview beta of Docker on M1 for some time now. It works fine for me; some people have reported issues in the corresponding Docker Slack channel, but they usually stem from the particular images in use rather than Docker itself.
> It's got to the stage that I prefer to use my 13 inch M1 over a top of the line 16 inch provided by work.
I agree. It's really quite difficult to go back to x86 laptops. If they manage to fix the external monitor limitations with the next generation--and I'm sure they will--it will be really hard for me to justify spending money on anything else.
The problem is the vram using full power.
The only solution I’ve found under OSX is using an eGPU. I have both the Core X and the Core X Chroma, and the Chroma works fantastically both as a fully fledged docking station and keeping the output wattage / fans at bay. I’m almost exclusively using my 16” stationary with an external display.
Another solution, albeit expensive, is buying the 16” MBP with the 5600m graphics. That card uses HBM2 memory which uses around a 4th-5th of the wattage of the GDDR6 memory of the 5500m. Around 4w instead of 19w.
I don't understand why people insist on trying to convert laptops into desktops.
I finally bought a desktop again over Thanksgiving after a decade of having a laptop that functioned as a desktop. It's been a great experience.
I don't do anything taxing, so I just got a little Intel NUC i5 with a 24" monitor. Having an actual desktop setup is just head and shoulders better than the "laptop as desktop" I'd been living with.
The laptop only needs to drive the display(s) so anything made in last few years should suffice. Currently using a surface go.
They want to have just one computer without worrying about synching or other coordination.
I can imagine that when it came to configure the machine the decision was `Put more RAM in it, he's a web-dev, doesn't need a fancy graphic's card!`
Which I don't as I'm not going to gaming on my work PC, but if that 15W is pushing it above the thermal limits of the chassis, then that would have been a decent upgrade.
I'd prefer it without a heavy-lifting graphics card in that case, but that doesn't seem to be an option.
I also would be, if there were not some annoying bugs. Like external display flickering after waking up from sleep (interestingly, not when cold booted). Sometimes reconnecting the cable fixes it, but it is still an annoyance.
I have my problem with OWC TB3 dock and Dell P2715Q. I will try an older dock that I have in the office (Kanex TB2 Express; as the name says, it is TB2, thus extra dongle needed); but playing around with different TB docks can get expensive...
It also does not happen with Intel Macs, which use Intel TB implementation. With M1, the TB is brand new, so some issues are to be expected. I suspect that with sleep and wakeup, there is some state mixup with previous link training, so it doesn't talk to other side the way the other side expects. This might be - or might be not - fixed with firmware updates.
This might be the one: https://support.apple.com/en-us/HT204450
No, OWC Thunderbold 3 dock and Dell P2715Q.
I also have older, 2015 Macbook Pro, with Thunderbolt 2 and Apple TB2->TB3 adapter, connected to the same setup. This laptop does not flicker.
FWIW I use catalina on a 2009 core 2 duo mac mini hooked up to a 1440p monitor in native mode, and the fan remains at idle when just clicking around the desktop.
Does it suck? Yes. Are fans only $10 to save a $2000 machine? Yep. Apple simply doesn't design their kit to run hard and run often. It's like those internet services that sell higher speed than they can actually give every customer. They depend on you not using it all.
For docker I plan on using a pi 4 or another PC to just run stuff over the network.
2) free up memory on my mac. I’m working at home so using a networked computer isn’t a big deal.
Serious Q: Have laptops ever been decent for gaming?
I don't game much, but I think the form factor would seriously crush thermal capacity.
I have a 15" MBP and it's fine for most dev work save Node and Typescript occasionally cause the fans to spin up for extended periods of time. Getting rid of VSCode in favor of more memory/ CPU friendly editors helped a bunch.
Does your work involve a lot of 3D or graphics?
HP, Dell/Alienware, Lenovo, all of them have had overheating CPUs for over a decade. GPUs are fine half the time, surprisingly.
Their testing is garbage, what they call normal use seems to be completely different from what power users call normal use. And I'm talking about "workstations", not cheap consumer laptops.
Oh, but the design constraints, you'll say. OK, that's forgivable for Ultrabooks/thin-and-light laptops. Not workstations. And before the thinness fad, they were still overheating. And not because of design constraints.
As an example, I'll bring the Elitebook 8760/8770w/ZBook 17 G1/Zbook 17 G2 series, all of which have the same internal design and layout.
All were overheating after a while, due to thermal paste going bad (and being shit from the factory, but I digress).
With the ZBook G2 it got very, very bad. Those Haswell chips with FIVR were mini nuclear reactors. HP realized they weren't "just overheating", but "extremely overheating".
Their solution? The same heatsink with an extra heatpipe. Just like that, poof, problem solved. One extra heatpipe!
That can't cost more than $2 in the factory. I don't know any buyer who wouldn't have paid an extra $5-10 to have a proper working $4000 workstation.
Whoever is saving money on copper and screws in these companies needs to be fired and banned from ever working in companies making hardware.
I'm not even going to talk about Apple, everyone knows their stuff has a ton of problems. But they're selling people on design and software, not quality engineering.
And they actually solved the heat problems by going in the opposite direction with the M1 (ultra low power usage while maintaining performance instead of a good cooling system).
I do watch Louis Rossmann about Macs, and how terribly they are designed, but this is universal across many devices.
I had so many issues getting low latency audio to work on my 2 year old XPS, I just plain replaced it with a 5 year old mac and had better performance and stability out of the box.
No company is perfect, but Apple does get consistently close. I'm glad they don't focus on the specs and focus on experience. Who really needs a 4k screen on a laptop?
Ok, this is more or less what I thought.
This does seem solvable, but also seems like the sort of thing the bean counters would cut early.
From what I recall, pre-Dell Alienware did a decent job at it.
I certainly didn't fancy a gaming laptop that's all compromises and no upsides just to play mindless AAA titles.
But the problem with the XPS is they jammed as much processing power as they could. It's great in theory having a Nvidea card as well as an i7, but it just can't handle both running at above half capacity.
I guess it would be better to wait for M2 or M3 rather than jump on a new system that has a work-in-progress developer ecosystem or un-optimised software (Even if it is faster on M1 than Intel Macs).
By the time that Apple Silicon software support is optimised, the M1 Mac would already be out of date and superseded by M2 or M3 Macs.
Docker and Virtualbox (this one is not coming) are the only two apps that do not work for me, and worked on Intel Mac. Sure, there are many apps that run in Intel mode, but the user won't notice, unless interested in that detail.
Intel at the time was shipping P4 Netburst CPUs. They had all the features, like multiple cores and 64-bit support, but also one huge disadvantage: they were power hungry, and it dissipated all that heat. Definitely not something you could put into laptops.
So the next generation, Core Solo/Duo, was not continuation of Netburst, but Pentium M, their mobile line which itself forked off P6 (Pentium Pro). They were released in January 2006, and were indeed 32-bit. The reason was time to market, they were basically throwing away their current arch and forking from an older branch; for the majority it was not a problem, most people were running 32-bit XP anyway, with PAE if you happened to have more than 3,5 GB of RAM (pre-SP2 XP supported that). Then, the 64-bit Core 2 Duo was released in July 2006.
The unfortunate part is, that Apple announced transition to Intel in mid-2005, and shipped the products across all lines in January-August 2006, i.e. hitting the window when only the 32-bit version was available. That means, that they had to define and maintain i386 ABI for their OS just because they ever shipped Core Solo or Duo machines. If they were able to ship C2D from the beginning, they could have skipped i386 and just use amd64 ABI from the start.
I can't recall, but I never could finish Persona 4 on my XPS. :( 9350.
Agree 100% w/ 16" MBP.
I think Dell must hijack those TurboBoost power limits as a way to massively underclock the processor and try for better advertised battery life.
I'm excited that the M1 is good, also- that it seems fast with everything, even the old Intel-specific code via Rosetta. I don't subscribe to Apple NIH[1] philosophy, but what they have made historically, barring the problem I mentioned above and a few macOS and iOS issues over the years, has been awesome. The M1 seems to follow Jobs' philosophy of making the best product.
I've had the following in my life
* Pentium 4
* 21.5" iMac
* 27" iMac
* Another 27" iMac
* Self built 2700K Intel
* Dell XPS 9560
* Self built 2700X Ryzen
* 2015 15" Macbook Pro
* M1 Macbook Air
For my work PCs, I had a Dell Desktop, a Asus Laptop and 2 Macbook Pros.
If selecting a PC for work, I would never go for a Windows PC nowadays. You'll get the cheapest option available, filled with all the corporate crap you don't want, and locked down to oblivion.
Macs have a minimum quality bar, and don't tend to get hamstrung quite as easily by company policy. It's just not worth it unless you know for a fact you'll get to choose the machine and software yourself.
As for at home, I'm flexible, but I wouldn't go for Mac if I wanted to game, and now in 2020, I wouldn't go x86 if I needed battery life.
Well there's your problem. Don't define your choices by what you're coerced into using at work where you only go because they pay you. Define yourself by what you choose freely as a person.
It sounds like you've self built before, so why not again? It'll be far more compatible and with fewer proprietary bits than using something from Apple.
They are completely unsuited for multi-monitor workloads without an eGPU.
More specifically, the draw 30W of additional power that has to go somewhere. You can turn down the refresh rate, to say 10Hz or something, but it still has to use the discrete GPU. The cooling solution simply isn't handled to deal with it, as they have crammed as much processing power they can into the ultra-light chassis.
Meanwhile you can play World of Warcraft whilst drawing just 25W of power TOTAL on the M1 (estimate, I don't play WoW anymore). This is while the Intel Mac is drawing 70-100 Watts from the socket for just rendering a window to the secondary display.
It's got to the stage that I prefer to use my 13 inch M1 over a top of the line 16 inch provided by work. When M1 finally supports docker, I might even consider not using the 16 inch at all. It crashes due to thermals about twice a week due to the constant high temperature and eats through batteries!
If anyone has a solution to this, I'm all ears, but it feels like every nay-sayer of the M1 is dismissing just how many compromises Intel based laptops (not just Macs) have. I had just as many issues with the Dell XPS, where even basic video games would become unplayable after 15 minutes due to CPU throttling down to 0.8 Ghz.
The M1 has been the first laptop I have been truly happy with as a desktop user.