There seems to be only one public test of the Macmini9,1. Sometimes GeekBench is not accurate on the clock speeds. Take the clock speed with a little skepticism until more results are posted. I can't see why the system with the best cooling would be 3.04 GHz instead of 3.2 like both the MacBook Pro and Air.
Generally, yes. The benchmarks are calculating/computing the same things on both architectures, and the various sub-benchmarks are based on real-world non-trivial computational problems rather than microbenchmarks that are easily manipulated by instruction set differences.
Unless they are written in optimised Asm I would not compare across architectures, because there is considerable leeway in compilers and my experience has shown that a good human can often beat a compiler solidly --- at least on x86.
If anything, I feel like many tests are based on x86 special instructions like AES, which does not directly translate to other tasks. But for most of the tasks I think it wouldn't be possible that the processor are only better at those and performance of most program should correlate with geekbench score.
I wonder what the Macbook Pro 16 inch late 2019 is on the single core score. Not having the same devices in both single core and multi core makes me sceptical.
I really really hope that in real life scenarios this will be much faster than Intel CPUs, in case a wake up slap from AMD will not do. Maybe Intel will swallow its pride and design a new CPU for TSMC.
Geekbench is designed to measure CPU burst speeds, not to measure cooling systems. By design it does not run long enough to test the thermal situation.
Their line from the video about being the highest performance chip in single core appears to be true. This is of course a synthetic benchmark but the single core result is very promising. Note that the single core and multi core scores exceed the top-of-the-line 16” MacBook Pro (9th generation 8-core i9 2.4 ghz). I actually made the call to sell my 16” for the new Air yesterday. It’s looking like a good call. Glad I’m selling my 16” while it still has some value.
Replacing my 16" MBP (8 core i9 2.3ghz / 32GB RAM) with the new Air as well. All of that power (yes I know, it's not sustained) without a fan is incredible.
Why did you pay a fortune for 32GB RAM, a larger screen, and a dGPU if you don't need it? You could have bought a 13" MBP and saved enough cash to get the new Air and now have 2 laptops.
It is a general theme, people say how they are switching their specced out 2019 16inch MBP to these new M1 Macbooks, then getting all upset when people point out, they are probably not the professionals the top-end MBPs were aimed at.
A lot of people run around with way more powerful laptops than they actually need for whatever they are doing, because it's through a business or it's deductible, but news flash, buying a Macbook Pro doesn't make you a pro.
A question, if ALL pros were fine with 16GBs of RAM, why does Apple offer 4x as much? Answer, because a lot of people will actually need it.
I am happy for people who will get these new devices an be happy with it, I might get one too. But truth be told, most of us getting these devices could make it work with the latest iPad Pro + Magic Keyboard just as well. (OK, I do need to code occasionally, but even for that there is pretty ok apps for iPad I could use)
The expectations have to come the fuck down from where they are today, because the expectations put on these devices are just crazy. It's so overhyped that I think many will be disappointed, when compatibility issues surface and when people realise that the 3x, 5x, 7x performance digits are mainly down to Fixed Function Hardware and Accelerators and general performance increase is just slightly above the generational leap we are used to, with a bigger increase in efficiency.
I’ve always used 15” MBPs without external displays as my only computer for years. I like being mobile and don’t particularly like having a desktop + laptop. Since COVID I’ve been stuck at home so I decided to finally get an external monitor. The problem is that you can’t use the 16” with an external display and the lid open without the fans spinning at full speed. Some people don’t mind it but it drives me crazy. Now it’s hooked up to my display in clamshell mode. Do I need all of that power? No, but I did want the biggest screen at the time when I purchased it a year ago. I plan on selling it when the new Air comes in.
> The problem is that you can’t use the 16” with an external display and the lid open without the fans spinning at full speed
Say what? I have a LG 5K and two 27” Apple Thunderbolt displays (four screens total including laptop display) hooked up to my 16” MBP and fans definitely are no where near full speed, unless I’m compiling or in a Google Hangout that is...
Yeah I'm powering 4k external display for normal things (hangouts, photoshop) no fan, only when doing a-frame / three.js stuff in the browser will bring the constant fan and provide a nice finger warmer above the touch bar
Back in the day I had 2 Sun 20" GDM20E20 (1997) which was major $$ and after then I alway had 2 monitors, moving at some point to a single ultra wide LG (which are pretty neat). One day I looked at my setup and how I used it and realized I did not look at all of the screen. I swapped it for a small single Apple LG 4K and it turns out I am very happy. The dense nature of the 4K was a game changer. I plan on getting an 8K when it comes out.
Generally speaking, I’ve never found that to be genuine, but assume best intent and all as the site rules say, so here goes...
For me, I often am doing multiple things at once and juggling between unrelated tasks which actually need my attention sporadically. The LG 5K with it’s beautiful display gets my primary attention and is what I want to be focused on. Apps there are what I should ideally be working on. The two Apple TB displays then flank either side, and they get the “distractions”, but stuff important enough to allow distracting me when needed. What that is is variable from day (sometimes Slack makes the list, sometimes it doesn’t, as one example), but it’s intentionally in my peripheral vision so I only “look” for motion/changes in certain areas, not actually try to read. If I need to read, I context shift by rotating my chair slightly to the left or right (better for you than rotating head).
End of the day, do whatever works for you. Yes, there are folks who can legitimately take advantage of lots of screens like me. Some folks who have tried multiple don’t, and are happier when they switch back, but I’m not one of them and it’s something I routinely experiment with to ensure I’m still using the best “for me” setup. I’ve gone as high as nine screens attached (with eGPU) to my laptop (eGPU seems to keep laptop fans on elevated, but not full power btw, back to original thread purpose), but I found I was too easily distracted and hence am back to four. Ideally I’d like to do two 8K 32” or less monitors, but haven’t justified buying them yet.
Thanks, that makes sense. For me I have found having something on the other monitor catch my eye a distraction, so I have gone back to a single high DPI one.
I have two identical top-specced MBP16", bought them as soon as they were released. They BOTH do that with my LG 5K Ultrafine display, whether they're closed or open, without doing anything particularly heavy. Sent them both to Apple, they passed all checks with flying colours. And that's the end of more than a decade of giving Apple tons of money. I built a Ryzen desktop this year and, despite missing macOS terribly, I couldn't be happier with the speed and ergonomics.
edit: I've tried both sides of the laptop, I have iStat Menus and keep an eye on temps, etc.
edit2: they "only" spin to 3.5k-4k at idle, but go up as soon as I do anything with Chrome or am on a video call, which is most of my job
They're all set to native, I don't do scaling. I've since got a free top case replacement from Apple (keyboard was slightly busted because I spilled soda on it lol) and, after a PRAM reset, it's not going crazy at idle, only when being in WebRTC calls (which is 90% of my job).
It's not a process thing. The left-side TB3 controller freaks out under sustained load and can drive the system to a crawl. We have to unplug one of our LG 4/5Ks and run the other+power on the right when we want to be able to do back to back video conferencing.
It's been a running joke in corporate for years that Apple's "premium fan noise" is a brilliant branding move because you can identify the Mac users as soon as they unmute.
I regularly drive a 5k2k display on a 16" MBP, along with glowy keyboard and trackpad, and I only get the fan when I'm doing something like a full-screen streamed video: text editing and web browsing seldom trigger it.
I like the bigger screen, so I'll hold out on this platform for now. Pretty impressed with where the M series is going, though; might hold out two iterations instead of the four I had in mind before the M1 dropped.
> The problem is that you can’t use the 16” with an external display and the lid open without the fans spinning at full speed.
I had the same problem, when connected to a USB-C monitor I wanted to use the keyboard, but not the built in monitor. Even with the display backlight off the fan would still run. After a lot of searching I found that you can disable the built in monitor by:
- Booting into recovery mode
- Opening Terminal
- Entering `sudo nvram boot-args="niog=1"`
- Restarting
- Close the clamshell
- Plug in the external monitor
- Open the monitor
Anandtech on the memory-affinity of GeekBench vs SPEC:
> There’s been a lot of criticism about more common benchmark suites such as GeekBench, but frankly I've found these concerns or arguments to be quite unfounded. The only factual differences between workloads in SPEC and workloads in GB5 is that the latter has less outlier tests which are memory-heavy, meaning it’s more of a CPU benchmark whereas SPEC has more tendency towards CPU+DRAM.
The new MBA is a total beast. The comparison is almost unbelievable. Can't wait to see what they do with the iMac, Mac Pro and MBP 16". Just phenomenal!
Damnit I have to decide the same thing. I’m really happy with my 16” mbp for once and I’m not sure if I want to get a smaller screen and give up windows support (for now) I feel like Ms could be convinced to make a version for Apple silicone if it keeps its performance advantage.
Good news is Parallels announced a closer colab with Apple to bring x86 virt to M1, too. They demo'd Parallels running a linux VM at WWDC, but the upcoming release will also support seamless Windows virt again.
That PR doesn't actually say anything about running Windows. You can't just port the app. A VM on an ARM system is still ARM inside, and given that the PR specifically mentions "support of x64 applications in Windows on ARM", this is clearly for ARM VMs. You'd need actual Intel emulation in order to run the normal version of Windows.
> You'd need actual Intel emulation in order to run the normal version of Windows.
Microsoft is working on enabling x64 emulation on ARM, it should roll out in preview this month[1]. I can see Windows 10 ARM-edition working inside Parallels with its own x64 emulation inside. The issue right now is that MS does not sell Win 10 ARM, it is available for OEMs only.
x86 emulation on Windows 10 ARM was already done few years ago, when MS shipped their Surface ARM notebook.
That actually seems more logical than the alternatives.
One of the big hopes for Rosetta2 is the possibility of intercepting library calls and passing them to the native library where possible. So a well-behaved app using OS libraries for everything it can, and really only driving the business logic itself, would be running mostly-native with the business logic emulated/translated.
(This is hopes/dreams/speculation with no insider knowledge.)
If Windows could do the same, then letting windows-arm do the translation of windows-x86/64 binaries would allow it to leverage windows-arm libraries - so an app could be running in mostly-virt with some-emu. If we let parallels/qemu/etc do the emu, it can only ever be 100% emu.
You can emulate anything on anything (pretty much), but the real question is can you emulate it at a speed that’s sufficient. That takes host-system-specific optimizations.
Indeed. Might be misleading marketing. Docker meanwhile mentioned they will launch with ARM containers only, but are expecting QEMU to be able to run x86 (probably badly).
I bought the new Air, but I’m keeping my 16” as it may be the last laptop of that quality that can run programs without telling some stranger over the network that you’ve run a specific program.
Current (and presumably future) macOS does this and you can’t turn it off, except with Little Snitch. New APIs in macOS 11 means that Little Snitch will no longer be able to block OS processes, so it will require external network filtering hardware.
I’ll likely end up with Linux on the 16”, and use the new one for things that are not secret/private.
This is where I’m at too. Compromising Little Snitch and VPNs is just a bridge too far. It’s cool that they got this level of performance in a lightweight form factor, less so when it enables the worse of surveillance practices.
Looks like it will be impossible to use Apple Silicon (without external network hardware) without revealing your track log to the CIA. How cool is that?!
I'm seriously considering doing the same thing. My only hesitation is the screen size. But right now the laptop I bought in January is being outclassed by one that costs 1/3 the price...
Many don't even want to pay for the MacBook Pro's Touch Bar and many will probably see an Air's fanless design as an advantage over Pro, even if its CPU is throttled a little more often in sustained high CPU workloads. Complete silence is just that good. And it's going to be so much cheaper.
I think the star of the show yesterday was definitely the MacBook Air.
It'll definitely be a different chip either way. To have a 2x or 4x M1s would require interconnects between the CPUs for sharing RAM and general communication, which isn't there on the M1 currently. So at a minimum you're talking a rev of M1 to add such interconnects. But that all seems very unlikely. If they do go that route it'd almost certainly be something like AMD's chiplets instead of just straight "multi-socket"
I've been holding onto my late 2013 15" MacBook Pro for almost 7 years now and it's wild to see it get absolutely out-spec'd by the MacBook Air. It's been a long time, but I've always convinced myself that I'm not missing out on that big of a processor improvement. I do a lot of audio work with it, but even then it doesn't always feel like a slow experience.
That MacBook has to be one of the best computers Apple has ever created. Mine has lasted this long with non-stop daily usage for years and travel around different countries.
Throwing in a comment in this thread to say same. Incredible how I've used mine for 7 years and it seems new still. Only issue is 8 gigs of ram makes things like docker or ides difficult, but I get hesitant to buy a replacement because of the year to year changes. I mean, why buy a new one if you have a machine that's been going strong for so long?
Yeah I’ve had two new models since but my 2013 keeps on ticking. Never had any issues and still running just fine. Night and day compared to the piece of garbage 2017 model I got.
The Dell XPS with a i7-1165G7 processor has a base price $1,499. This includes a 512GB SSD so compares to the Macbook Air's $1,199 price with the same SSD. With that Macbook Air configuration, you also get a higher resolution screen.
The one you shared looked crazy overclocked or something(it says 4.7 ghz). Almost all other tests of 1165 is in the range of single core performance of 1500. This is more representative of the chip of the same laptop: https://browser.geekbench.com/v5/cpu/compare/4648891?baselin...
I guess for those who desire more RAM the call is hold off on M1 based MacBooks until this ceiling is raised. I’m guessing an intel based MacBook is a better call, except resale price will suffer.
I currently have the same MacBook but everything seems to be breaking (keyboard, battery, screen, major OS upgrades fail, etc) and it needs to be replaced. But 16gigs seems not very future proof. I’m already having issues with some video codecs I can’t work with, example .360 GoPro files. (I bought an iPhone SE to handle that!)
editing programs are loading a ton into memory though. edits are done in memory before their final output is saved to disk. if your viewing it on the screen, it is almost certainly in memory, and if your working with high res video you very well could exceed 16gb. more memory means larger (longer clips or higher res) video files loaded to memory means faster and more efficient working
Kinda my thought exactly. Although most any video editing software uses proxy files, so you don't have to load in the original hi-res footage.
It's just that... swapping data in and out of ram is far slower than clock speed, right? So if the bottleneck is RAM, would it make sense to get the most RAM? (ie: Intel Mac)
The question is which professional software for video editing is available for this new platform? Probably not much at the moment. So these machines will be perfect for those who do their work with browser, email and not much more.
I'm using Tableau Desktop in my daily work and until it is available on this new platform - these Macs are not an option for me even if they are 10x more performant. I guess there are a lot of professionals that are constrained in a similar manner. So we will see if M1 is adopted in such scenarios at all.
Anyone have thoughts on where are these benchmarks might be coming from? Part of me wants to put in my order today, and part of me wants to wait until after the YouTube reviews.
Geekbench is designed to test CPU's in short bursts to not get them into thermal throttling (this is by design, it is supposed to be CPU test, not cooling test), so I would be wary of Air's results - at least until other benchmarks results are available. MacBook Pro's results on the other hand should be more representative as it has active cooling and can sustain heavy load for longer.
I know, and they show the same numbers as Air's CPU.
I'm just making sure everyone understands that it does not mean Air will have the same real life performance as Pro or Mini. If you compare top MacBook Air 2020 (Intel)[1] with lowest MacBook Pro 2020 (Intel)[2] their results are almost identical, but their real life performance was not even close - Air starts to throttle after just few minutes of work. At the moment there is no reason to believe that Apple's CPU will behave differently (after all, the fan is in Pro and Mini for a reason).
It's a mobile CPU with many silicon advantages (widest decoder in industry, memory closer, deepest re-order buffer of any CPU and much more) plus a sane ISA and optimized OS. So yeah, you're seeing the benefit of Apples integration. That's why even the Anandtech page calls that graph "absurd", because it seems unreal, but it's real.
Gotta be frank because it's not getting through: you're jumping way ahead here. Every time one of these threads has happened, there's an ever-increasing # of people who vaguely remember reading a story about Apple GeekBench numbers, so therefore this one is credible too - I used to be one of those people. This has been going regularly for 3-4 years now, and your interlocutor as well as other comments on this article are correct - comparing X86 versus ARM on GeekBench is nonsensical due to the way GeekBench eliminates thermal concerns and the impact of sustained load. Your iPhone can't magically do video editing or compile code faster than an i5.
My comment, and this specific thread, isn't even related to GeekBench. The graph I linked used SPEC instead of GB5. The gigantic architectural deep dive over on Anandtech even includes a discussion on the strengths and limits of both benchmarks, and how they make sense based on further micro-architecture testing.
The reason that graph doesn't include the A14 Firestorm -> M1 jump was simply timing. We know the thermal envelopes of the M1 and the cooling designs. We now have clock info thanks to GB5. So yes, the data is pretty solid. No one's saying that the iPhone beats the Mac (or a PC) at performance when you consider the whole system. Just that the CPU architecture can and will deliver higher performance given the M1 clock, thermals and cooling. Remember that The A14/M1 CPUs are faster at lower clock speeds.
that's comparing a hardware encoder to a software one, unfortunately, as the replies note.
it's unfortunately drowned out by the cpu throttling scandal on google, but, its well-known in ar dev (and if you get to talk to an apple engineer away from stage lights at wwdc) that you have to proactively choose to tune performance, or you'll get killed after a minute or two due to thermal throttling.
This raises the question of just why the Mac is doing software rendering—I think the hardware it’s running on should have two compatible hardware encoders, that on the CPU and that on the GPU. Is the software being used incapable of using hardware encoding? Does it default to software rendering because of its higher quality per bit? Was it configured to use software encoding (whether ignorantly or deliberately)?
Video encoding is generally done on CPUs because they can run more complicated video encoding algorithms with multiple passes. This generally results in smaller video files with the same quality. As you increase the compute insensitivity of the video encoder you get diminishing returns. 30% lower bitrate might need 10x as much CPU time. That tweet says more about the type of encoder and chosen encoder settings than anything about the hardware.
Imagine going on a hike and climbing an exponential slope like 2^x. You go up to 2^4 and then go down again and repeat this three times so you have hiked 12km (43) in total. Then there is a athlete who is going up to 2^8. He says he has hiked 8km and you laugh at him because of how sweaty he is despite having walked a shorter distance than you. In reality 32^4 (48) is nowhere near 2^8 (256). The athlete put in a lot more effort than you.
It's also the most wrong explanation. The actual performance efficiency between those processes isn't that drastic. The power efficiency of the M1 would come from better IPC such that it just doesn't have to clock as high to be competitive.
That's why A14 only runs at 1.8ghz base, 3ghz boost. That's how it has low power consumption. And similarly Intel pushing 5ghz is why it has high power consumption.
TSMC's 5nm will have a raw transistor performance/watt advantage, but it's not huge
Compare for example the 1 core power numbers between the chips. The 5600X 1 core result is 11w @ 4.6ghz, whereas the other two chips boost higher and hit 4.8-4.9ghz 1 core turbos, but it costs 17-18w to do it. Huge increase in power for that last 1-2% performance. So you really can't or shouldn't compare more power-concious configurations with the top end desktop where power is infinite and well worth spending for even single digit percentage gains.
And then of course you should also note that the single-core power draw in all of those is vastly lower than their TDP numbers (65w for the 5600x, and 125w for the 5800x/5900x).
Yeah comparing TDP is meaningless even within the same processor. The 4 core workload in this table uses 94W and the 16 core workload uses 98W. There is also an anomaly at 5 cores where the CPU uses less power than if it only used 4 cores.
If you tried to derive conclusions about the power efficiency of the CPU you would end up making statements like "This CPU is 3-4 times more power efficient than itself"
This is pretty crazy to see, even if the full story isn't clear yet. A base level MacBook Air is taking the crown of the best MacBook Pro. Wow. SVP Johny Srouji and all of the Apple hardware + silicon team have been smashing it for the past many years.
For what it's worth, I have a fully specced out 16 inch MacBook Pro with the AMD Radeon Pro 5600m and even with that I'm regularly hitting 100% usage of the card, and not to mention the fan noise.
Looking forward to a version from Apple that is made for actual professionals, but I imagine these introductory M1 based devices are going to be great for the vast majority of people.
It still has a lot of limitations that matter to many pros - max 16 GB of RAM, max 2 displays (only 1 external for laptops), only enough PCIe lanes to support 2 thunderbolt ports. eGPUs aren't supported either, but hopefully that is a software thing that will be fixed.
It will be very interesting to see what the performance will be of the more "pro" chip that overcomes those limitations that they'd put in the 16" and iMacs
why would it be only 1 external display? Can you not plug into a thunderbolt dock and use multiple external displays? I do exactly that with my 2012 MacBook Pro.
This may work. On a 2014 macbook air, which officially only supports only one external display, I've been able to use two external screens at the same time by sharing one through airplay. However, you're always limited in resolution and latency that way.
It’s a first generation chip in a thin-and-light laptop.
I suspect all those problems will be fixed as they scale it up. Not to mention their unique position of leverage dedicated acceleration silicon in their software now.
eGPU support will depend on if they have a way to work around the need for a PCIe I/O BAR. Many GPUs require that to initialize and as far as I know no ARM cpus support it since it's a legacy-ish x86 thing. It'll be the same problem that prevents gpu use on raspberry pi 4s still. I bet you can make a controller that'll provide a mapping for that to allow it but that'll mean needing a new enclosure (probably not a huge deal) and new silicon and drivers.
AMD GPUs do not require I/O BARs, and highly doubt Nvidia ones do either. The VBIOS will probably assume it can use it, but most modern cards can be initialized without actually running the VBIOS (because people use them on headless servers, for virtualization, and things like that). The I/O BAR is only required for legacy VGA compatibility, you can ignore it.
Dunno, seems like most professionals would want to use docker, virtual machines, or enough video/data to want more than 16GB ram. Or maybe even plug in more than a charging cable and one more device. Or run more than one external monitor.
Doesn't seem very "pro" to me. The MBP16" intel has 4 x USB-c ports, can drive two monitors, and can have >= 32GB ram.
Most professionals aren't software developers and don't need to run virtual machines.
But regardless of professional requirements, "Pro" in Apple's product line just means "the more expensive slightly better version." Nobody's arguing that AirPods Pro are the wireless earbuds of choice for people who make money by listening to wireless earbuds.
Hm, a good VM is one you don't even notice. Say you run Qubes. Not necessarily for development (I would argue Nix is the best OS for DevOps). If all goes well, such an OS becomes very adequate for an average user. For example, a hardware VM could allow you to run a browser more secure.
Yep. My new work laptop only has 16GB of RAM and it’s never been an issue. I’m usually running half a dozen containers, VS Code, Slack, Brave/Chrome, and a few other things. Maybe our work loads are just computationally lighter than some?
I ordered a 16GB Pro the other day to be my personal dev machine. I’m sure it’ll be more than fine. I’m upgrading from a 2013 8GB Pro which was only just starting to slow me down.
The code and compilation for me is the light part, and for the most part an hour's worth of essentially text editing for about a moment of compilation anyways.
My resource hogs are slack, mainly the browser, and Zoom calls are apparently the most computationally intensive thing in the world, especially if you screen share while you have an external plugged in.
Memory wise the reason I had to go from 8Gb to 16GB on my personal laptop was literally just for TravisCI.
Honestly, adding external monitors cripples MacBooks pretty quick, even unscaled 2 2k monitors will slow a 2015 15 down significantly (don't try and leave YouTube on either), and it gets worse from there once you start upgrading to 4k monitors. a 2017 15 is good for a 4k and a 2k, and gets a bit slow if you try and go dual 4k.
I planned on looking into eGPU solutions until IT offered me a new Macbook, and I convinced them I needed a 16" Pro.
tldr: External monitors or badly optimized applications (Zoom, YouTube, or browser based CI) will make most MacBooks feel sluggish pretty quick.
Are those displays in scaled mode? Scaled displays tend to perform badly on integrated graphics, and suck up memory, because it has to render a 2x or 3x size internally and then scale down for every frame. Running something that updates the screen constantly, like zoom, probably exacerbates that issue.
Gaming laptops are huge for a reason: they need to get rid of a lot of heat.
Thin and light is great for short bursts of activity, but, when you need sustainable heavy usage, you'll need a bigger computer, even if it's just to have a bigger heatsink.
It was Surface that started the trend, then MacBook followed and became a feature.
By the way, “throttling” refers to CPU _slowing down_ despite cooling working at full capacity, so loud fans in itself isn’t one.
e: Another way to explain thermal throttling would be “thermal fading”, like brake fading on a car. Whether brake fading is considered a design fault or a feature that allow bursts of stronger brakes is semantic.
Sorry, yes, I realise that’s what throttling is but wasn’t clear. I meant to say in the original comment the fans normally come on followed by throttling.
If we're doing analogies, let's do them right :) Therefore I'd argue it is more like supercharger overheating on a car. As a supercharger gets hotter from prolonged load it gets hot and warms the air entering the engine which reduces cold-air intake thus reducing horse power compared to a cooler supercharger. There is a way to solve this: by fitting a chargecooler - which is basically a cooling system for the supercharger.
My XPS15's GPU runs at 100C (on benchmark loads), even after I replaced the heat pads and cooling paste. It took me weeks to get used to the idea. I guess that's just the new normal. I don't find mine to be loud though.
But did the test run long enough to need the fan, and what was the ambient temperature?
The fanless Intel Core-M CPUs could post excellent benchmark scores (for its time). But if you give it a lengthy compile task, it'll slow down dramatically.
> Or are you saying that the test needs to run for way longer to be fair?
Yes, the main computing constraint of mobile devices is heat management (This doesn't really reflect the CPU but the complete device. Putting the CPU in a more ideal setup like a traditional desktop or water cooling will improve the CPU's performance in longer tasks)
> And it's doing this while using more than an order of magnitude less power (10W vs. a TDP of 125W for that Intel part).
That's the wrong conclusion to make. For instance, the Lenovo ThinkBook 14s (with a Ryzen 4800u) with a 15W TDP posts the same Geekbench multicore scores [1] as the M1 Macbook. But the ThinkBook isn't in any way faster than the top-end iMac for real world compute intensive tasks.
The M1 certainly looks efficient, but there's little you can conclude from a single benchmark running for a very short period of time.
That makes sense, AMD is selling to OEMs for a profit (over cost) while Apple is its own OEM, if any charging is done it's purely internal and for accounting purposes.
This comparison looks at different segments of the fab<>manufacturer<>OEM relationship. Add the user in there and you might say that you can buy an AMD CPU for $100 but an Apple CPU will cost you $1000. Not very meaningful as a comparison.
That price is a meaningless comparison, you can't buy the Apple processor in retail. What's the cost to procedure the AMD part? Something similar I'd guess.
It’s how companies price their products. Is it perfect? No, but that’s why I used the word “around”.
You can argue that this particular Ryzen has a higher gross margin, say 50%, and lower ASP than $300, but that only gets your cost down to what, $140? And with RAM costing extra.
>There’s been a lot of criticism about more common benchmark suites such as GeekBench, but frankly I've found these concerns or arguments to be quite unfounded. The only factual differences between workloads in SPEC and workloads in GB5 is that the latter has less outlier tests which are memory-heavy, meaning it’s more of a CPU benchmark whereas SPEC has more tendency towards CPU+DRAM.
Based on a sample size of 1 or 2, most likely.
It could be due to something as stupid as Spotlight running in the background on the MBP but not the Air.
It's thought that the MBP score is due to it being ran possibly during indexing on setup. The score difference is too big, it's a single sample, and the pro still fries it at single core.
It's not measuring sustained performance. The fanless MacBook Air is going to throttle much sooner than a desktop iMac with proper cooling and unlimited power.
The data we don’t have is for sustained use over time. An Intel iMac Pro can sustain max performance far longer than an Intel laptop, as it has a far higher thermal exhaust capacity.
Does the M1 performance have to be ramped down during sustained use due to exceeding thermal envelope of the fanless MBA? Of the fan’d MBP?
While it sounds promising, I'm going to wait for some additional benchmarks and real world usage scenarios; factors like cooling, multi-process work, and of course suboptimal applications (browsers, Electron apps, stuff compiled for Intel) will be a big factor as well.
That said, it's promising and I'm really curious to see where this development will lead to in a few years' time.
Me too - let's see what the sustained performance is like. That said, with this much headroom, I'm cautiously optimistic that even with some throttling going on, it'll still be plenty fast for anything I'm likely to throw at it.
iOS/iPhones/iPads already smoke every other device in running javascript. Apple Silicon may end up being the best thing to ever happen for Electron apps.
Without doubt, the Apple M1 has the highest single-threaded performance of any non-overclocked CPU, being a little faster than AMD Zen 3 and Intel Tiger Lake.
Nevertheless, because Apple has chosen to not increase their manufacturing costs by including more big cores, the multi-threaded performance is not at all impressive, being lower than that of many much cheaper laptops using AMD Ryzen 7 4800U CPUs.
So for any professional applications, like software development, these new Apple computers will certainly not blow away their competition performance-wise, and that before taking into account their severe limitations in memory capacity and peripheral ports.
But given that M1 is clearly the basic CPU for low cost, thin-and-light devices, we can strongly infer that Apple’s next M chip will be significantly more capable. Chips with eight or more performance cores would be a certainty for the upper tier of laptops and iMacs.
Given that the M1 is a full node ahead of Zen 3 and two nodes ahead of whatever Intel has to offer, one would think that when on same node, Intel and AMD will be just as capable.
But the truth is comparing to future offerings is bullshit, and we have to stick to what's available today. Impressive power/performance and all that, I have to say. We will see how sustained load looks like and how it runs non-optimized software. But to put in perspective 1 CCX of zen 3 performs better on 7nm (but draws up to 65W). With approximately the same die size (although w/o GPU and other things, the M1 has).
M1 is not magic and can't break the laws of physics. SMT makes better use of silicon and will probably push speeds closer. OTOH, M1 has a fast memory that the i9 can't match.
I still bet on the i9, but it'd be interesting to run a test.
x86 needs to use more complicated logic to deal with the instruction stream than ARM, freeing more of the silicon for things like better reordering and more execution units. OTOH, the SMT somewhat mitigates the delays caused in reordering by working on more than one instruction stream at once. I'd say the 16-thread chip will end up being overall faster than the 8-core one, if cache misses don't create a huge penalty for the slower memory bus of the x86. The i9-9980HK is also two generations behind, which doesn't help it much.
When I said there is no magic, I was warning that we shouldn't expect huge speedups or a crushing advantage, at least not for long. The edge M1 has is due to a simpler ISA (which is less demanding to run efficiently, freeing more resources for optimization and execution) and a faster memory interface (which makes an L3 miss less of a punishment). This fast memory interface also limits it to, for now, 16GB of memory. If the dataset has 17GB, it'll suffer. Another difference is that all of the i9 cores are designed to be fast, whereas only 4 cores of the M1 are. This added flexibility can be put to good use by moving CPU-bound processes to the big cores and IO-bound and low-priority ones to the little ones.
In the end, they are very different chips (in design and TDP). It'd be interesting to compare them with actual measurements, as well as newer Intel ones.
>M1 is not magic and can't break the laws of physics.
Anandtech's deep dive provides several examples of advances in Apple's core design that didn't involve magic or breaking the laws of physics. For example...
Instruction Decode:
>What really defines Apple’s Firestorm CPU core from other designs in the industry is just the sheer width of the microarchitecture. Featuring an 8-wide decode block, Apple’s Firestorm is by far the current widest commercialized design in the industry. Other contemporary designs such as AMD’s Zen(1 through 3) and Intel’s µarch’s, x86 CPUs today still only feature a 4-wide decoder designs
Instruction Re-order Buffer Size:
>A +-630 deep ROB is an immensely huge out-of-order window for Apple’s new core, as it vastly outclasses any other design in the industry. Intel’s Sunny Cove and Willow Cove cores are the second-most “deep” OOO designs out there with a 352 ROB structure, while AMD’s newest Zen3 core makes due with 256 entries, and recent Arm designs such as the Cortex-X1 feature a 224 structure.
Number of Execution Units:
>On the Integer side, we find at least 7 execution ports for actual arithmetic operations. These include 4 simple ALUs capable of ADD instructions, 2 complex units which feature also MUL (multiply) capabilities, and what appears to be a dedicated integer division unit.
On the floating point and vector execution side of things, the new Firestorm cores are actually more impressive as they a 33% increase in capabilities, enabled by Apple’s addition of a fourth execution pipeline.
> Featuring an 8-wide decode block, Apple’s Firestorm is by far the current widest commercialized design in the industry. Other contemporary designs such as AMD’s Zen(1 through 3) and Intel’s µarch’s, x86 CPUs today still only feature a 4-wide decoder designs
This is one place where the 64-bit ARM ISA design shines: since all instructions are exactly 4 bytes wide and always aligned to 4 bytes, it's easy to make a very wide decoder, since there's no need to compute the instruction length and align the instruction stream before decoding.
> advances in Apple's core design that didn't involve magic or breaking the laws of physics.
That's exactly what I said. It's faster, but not an order of magnitude faster and different workloads will perform differently depending on a multitude of factors (even if benchmarks don't). Do not expect it to outperform a not-too-old top-of-the-line mobile CPU by a large margin.
Remember the intel part has 8 fast cores while M1 has 4 (and 4 puny ones which really doesn't count). The Intel part also uses SMT to squeeze some extra parallelism that the reordering plumbing can't.
It's also funny that Johny Srouji and probably others in his team come from the team at Intel in Israel that "saved" Intel in the early 2000s by designing the Intel Core architecture which is still used by Intel today.
Did not know that, and indeed the pentium m / core / core 2 series microarches have done incredibly well.
I've become something of CPU collector in recent years, and I have a nice line of p6 cpus from thePentium Pro -> Pentium 2 -> Pentium 3 -> Pentium M -> Core 2 that conveniently sidesteps those awful Netburst p4 CPUs.
It feels like this (p6+) microarch has finally run out of road and needs a rethink. What 'saved' intel was a change in philosophy rather than chasing MHz they chased power savings. And with Apple's new chips that history is repeating itself (and appears to be with a similar outcome).
It's an exciting time for hardware again because Intel and AMD are going to have to react to this and I think there's still legs to x86, it's survived everything thats been thrown at it so far...
It's fascinating to see history repeated - the architecture and design with the most power efficiency generally also turns out to be the one that can be pushed furthest for performance when you want to go that route.
That's an interesting observation. It holds true in other areas as well. For example, we have lots of high horsepower cars as the result of R&D effort into high efficiency engines.
It is sad that engineers like him are not multi billionaires, but it all gets pocketed by investors, who don't even pay the same tax as people who actually do the work...
It's also worth pointing out that Apple is also benefitting from TSMC's latest and best fab processes. Intel is not only behind architecturally but in manufacturing too:
Agreed that the vast majority of Hacker News comments about the M1 Macbook Air are very glass half empty.
This seems like a really cool piece of technology, and I'm kind of bummed that everyone is so cynical and pessimistic about everything these days (albeit understandably so).
That's true, but some people need the larger screen and discrete GPU. I personally run bootcamp Windows all the time (for development as well as gaming with an eGPU) and I just run macOS for compiling my iOS apps. ARM Macs won't be able to use an eGPU it seems.
I'm assuming (for now) that discrete GPUs will still be featured in higher-end machines, so the Radeon drivers will appear over the next couple of years.
There is still PCIe bus and Thunderbolt, technically it is still possible to connect NVIDIA/AMD GPU to Apple Silicon, but we need enough bus (x8 or x16) and drivers.
And this test is very likely not the native test, but ran with translation from Rosetta. Kudos to Apple. This has the highest single core performance of all processors including highest end desktop processors from Intel and AMD: https://browser.geekbench.com/processor-benchmarks
I have a burning question: Can I use gcc, python, etc. on the new M1-based Macbooks? I am thinking about getting one but not sure if I can use it as a development machine. How does rosetta translation layer work with brew apps and other binaries?
They showed this during the previous transition announcement. They flashed up a screen of various open source tools that they’ve tested. It even included Blender.
I don’t have an M1 yet, but the things you mentioned work perfectly fine on my Developer Preview hardware, so I don’t see any reason they wouldn’t on the production hardware.
Apple is hardly the first laptop manufacturer that built a arm based laptop. Microsoft did it before them with the Surface back in 2012. However it's not going to be as popular as Apple's laptop.
Apple is rarely "first" with anything (they were nowhere near the first with mp3 players, for example). But they are often first-in-class once they commit to an idea. And then we got the iPod after many other companies tried to also do that in the preceding years. ARM is not new, but Apple committing to this is going to be very disruptive, and ultimately a more impressive implementation, in ways the Surface never was.
But it seems that everyone else had the idea that ARM is cheap material and only Apple put in enough effort to make the full potential of both the instruction set and the chip design.
That said, I am definitely waiting for at least one generation to pass before I jump on the train.
I’ve been through this before. It will be great, but Apple is a master at smoke & mirrors. Things will not go as smoothly as the sizzle reels make it seem.
Like the grandparent comment, I would avoid this M1 Macbook since the developer ecosystem is still unsupported by it. Take this for example: https://github.com/docker/for-mac/issues/4733
Docker still doesn't run on Apple Silicon Macs, so the migration path is already disrupted here.
We can include the A12Z developer transition kit, which is an 'Apple Silicon Mac' which was available since July.
The problem here is that in WWDC, Apple showed Docker Desktop running on an Apple Silicon system (maybe suggesting that it is at least running) and here we are in November it is still not running or not known if Apple Silicon is supported.
I don't see any patches in the Docker repositories on such support, thus maybe Apple has a private fork for Apple Silicon. In general it is not available to us and we don't know when it will be.
I think you’re probably right, but it’s still a massive change. Basically, repaving the highway, while there’s cars driving on it. Apple has done this a few times before (but we don’t talk about Copland in polite company), so I know it will end well.
I think their new architecture will be awesome. Personally, I look forward to being able to debug iOS Bluetooth apps in the simulator, and I think some of the new form factors will be pretty cool.
But I can’t help but notice the current dearth of AAA apps that are already universal.
A transition like this is a big deal; especially if you have an app with thousands of function points, as it requires a complete, top-to-bottom re-test of every one.
In the unlikely case that you won’t find issues, it will still take a long time. Also, most companies won’t bet the farm on prerelease hardware; instead, using it to solve issues. They will still need to test against the release hardware before signing off for general distribution.
Also, I have a couple of eGPUs that I use. I don’t think the new architecture plays well with them.
I’m hoping that the next gen will obviate the need for them. They are a pain.
All those thinking of buying the new MBA, are you not concerned about availability of applications? Not all x86 applications will have an ARM alternative, yet.
Had access to a DTK, and know even more devs who have one. Even that machine could run essentially all x86 software just fine with Rosetta (including things like dynamically loaded plugins). Only thing missing is virtualization, but that's coming soon (Docker & Parallels are working with Apple).
Nope. This transition will be fast. Not like PPC -> Intel at all.
Everyone is on Xcode these days and using much higher level frameworks. Adobe and Microsoft already announced early 2021 availability of native binaries.
Besides, Rosetta 2 is even more impressive than the original was. I’m betting it will be a breeze. 6 months in and almost everything will be native.
I won't be replacing my workstation this weekend but then, I won't be updating it to Big Sur just yet either. I am getting a new personal machine sometime in the next year though I don't imagine I'll be buying an Intel Mac.
I haven't encountered any problems running x86 programs on the DTK. Performance is fine (with a slower CPU than the MBA) and everything I tried running worked. I'm sure someone will hit problems once a few orders of magnitude more people are using it, but most people won't.
Several benchmarks have shown that popular x86 apps running in Rosetta on the M1 still run faster than running on native x86 on other machines, so it's probably not going to be an issue.
This is a pretty good single thread number. It's essentially the same as Zen 3. On the other hand, it's using 5nm rather than 7nm to get there, so 5nm Ryzen is likely to pull ahead in the not too distant future.
The more interesting thing is the power efficiency, which doesn't have that much impact on single thread performance because higher power CPUs don't actually use their entire power budget for a single thread. But that's an impressive multi-threaded score for that TDP. It gets stomped by actual desktop CPUs for the obvious reason, but it has better multi-threaded performance than anything with the same TDP. Though that's also partially because the low-TDP Zen 3 CPUs aren't out yet.
What I'd really like to see is some benchmarks that aren't geekbench.
1,171 comments
[ 3.3 ms ] story [ 489 ms ] threadhttps://www.realworldtech.com/forum/?threadid=185109&curpost...
IMHO it is reliable across different Arch/CPU/OS.
[1] https://browser.geekbench.com/v5/cpu/search?q=MacBook+pro+16
On my 10980xe it will cause the CPU to reach peak temperatures for 30s or so but only during specific workloads. Typically those utilising AVX-512.
You can see all Air results so far here: https://browser.geekbench.com/v5/cpu/search?q=MacBookAir10%2...
2. Even if he didn’t need it, why assume he didn’t want it?
3. Why are you assuming that money is an issue for him?
A lot of people run around with way more powerful laptops than they actually need for whatever they are doing, because it's through a business or it's deductible, but news flash, buying a Macbook Pro doesn't make you a pro.
A question, if ALL pros were fine with 16GBs of RAM, why does Apple offer 4x as much? Answer, because a lot of people will actually need it.
I am happy for people who will get these new devices an be happy with it, I might get one too. But truth be told, most of us getting these devices could make it work with the latest iPad Pro + Magic Keyboard just as well. (OK, I do need to code occasionally, but even for that there is pretty ok apps for iPad I could use)
The expectations have to come the fuck down from where they are today, because the expectations put on these devices are just crazy. It's so overhyped that I think many will be disappointed, when compatibility issues surface and when people realise that the 3x, 5x, 7x performance digits are mainly down to Fixed Function Hardware and Accelerators and general performance increase is just slightly above the generational leap we are used to, with a bigger increase in efficiency.
Say what? I have a LG 5K and two 27” Apple Thunderbolt displays (four screens total including laptop display) hooked up to my 16” MBP and fans definitely are no where near full speed, unless I’m compiling or in a Google Hangout that is...
Back in the day I had 2 Sun 20" GDM20E20 (1997) which was major $$ and after then I alway had 2 monitors, moving at some point to a single ultra wide LG (which are pretty neat). One day I looked at my setup and how I used it and realized I did not look at all of the screen. I swapped it for a small single Apple LG 4K and it turns out I am very happy. The dense nature of the 4K was a game changer. I plan on getting an 8K when it comes out.
Generally speaking, I’ve never found that to be genuine, but assume best intent and all as the site rules say, so here goes...
For me, I often am doing multiple things at once and juggling between unrelated tasks which actually need my attention sporadically. The LG 5K with it’s beautiful display gets my primary attention and is what I want to be focused on. Apps there are what I should ideally be working on. The two Apple TB displays then flank either side, and they get the “distractions”, but stuff important enough to allow distracting me when needed. What that is is variable from day (sometimes Slack makes the list, sometimes it doesn’t, as one example), but it’s intentionally in my peripheral vision so I only “look” for motion/changes in certain areas, not actually try to read. If I need to read, I context shift by rotating my chair slightly to the left or right (better for you than rotating head).
End of the day, do whatever works for you. Yes, there are folks who can legitimately take advantage of lots of screens like me. Some folks who have tried multiple don’t, and are happier when they switch back, but I’m not one of them and it’s something I routinely experiment with to ensure I’m still using the best “for me” setup. I’ve gone as high as nine screens attached (with eGPU) to my laptop (eGPU seems to keep laptop fans on elevated, but not full power btw, back to original thread purpose), but I found I was too easily distracted and hence am back to four. Ideally I’d like to do two 8K 32” or less monitors, but haven’t justified buying them yet.
Here's a 173 page thread on MacRumors about it: https://forums.macrumors.com/threads/16-is-hot-noisy-with-an...
As crazy as this sounds, using the left hand side ports for charging causes the fans to kick in more often[0].
[0]https://apple.stackexchange.com/questions/363337/how-to-find...
edit: I've tried both sides of the laptop, I have iStat Menus and keep an eye on temps, etc.
edit2: they "only" spin to 3.5k-4k at idle, but go up as soon as I do anything with Chrome or am on a video call, which is most of my job
Os-query was going bonkers on me.
It's been a running joke in corporate for years that Apple's "premium fan noise" is a brilliant branding move because you can identify the Mac users as soon as they unmute.
The monitor gets plugged in and the fans just start taking off to the moon - no workload or anything. It's extremely annoying.
I like the bigger screen, so I'll hold out on this platform for now. Pretty impressed with where the M series is going, though; might hold out two iterations instead of the four I had in mind before the M1 dropped.
I had the same problem, when connected to a USB-C monitor I wanted to use the keyboard, but not the built in monitor. Even with the display backlight off the fan would still run. After a lot of searching I found that you can disable the built in monitor by:
- Booting into recovery mode - Opening Terminal - Entering `sudo nvram boot-args="niog=1"` - Restarting - Close the clamshell - Plug in the external monitor - Open the monitor
I hope that helps.
1.5x single-core perf.
M1 MacBook Pro vs Intel MBP (top specs) show same performance: https://browser.geekbench.com/v5/cpu/compare/4652718?baselin...
Likely because GB5 doesn't run long enough to trigger thermal throttling on the M1 MBA.
M1 is beating all CPUs on the market in single-core scores: https://browser.geekbench.com/processor-benchmarks (M1 at 1719, vs AMD Ryzen 9 5950X at 1628).
Anandtech on the memory-affinity of GeekBench vs SPEC:
> There’s been a lot of criticism about more common benchmark suites such as GeekBench, but frankly I've found these concerns or arguments to be quite unfounded. The only factual differences between workloads in SPEC and workloads in GB5 is that the latter has less outlier tests which are memory-heavy, meaning it’s more of a CPU benchmark whereas SPEC has more tendency towards CPU+DRAM.
Microsoft is working on enabling x64 emulation on ARM, it should roll out in preview this month[1]. I can see Windows 10 ARM-edition working inside Parallels with its own x64 emulation inside. The issue right now is that MS does not sell Win 10 ARM, it is available for OEMs only.
x86 emulation on Windows 10 ARM was already done few years ago, when MS shipped their Surface ARM notebook.
[1] https://blogs.windows.com/windowsexperience/2020/09/30/now-m...
One of the big hopes for Rosetta2 is the possibility of intercepting library calls and passing them to the native library where possible. So a well-behaved app using OS libraries for everything it can, and really only driving the business logic itself, would be running mostly-native with the business logic emulated/translated.
(This is hopes/dreams/speculation with no insider knowledge.)
If Windows could do the same, then letting windows-arm do the translation of windows-x86/64 binaries would allow it to leverage windows-arm libraries - so an app could be running in mostly-virt with some-emu. If we let parallels/qemu/etc do the emu, it can only ever be 100% emu.
Look at N64 emulation, for example.
Current (and presumably future) macOS does this and you can’t turn it off, except with Little Snitch. New APIs in macOS 11 means that Little Snitch will no longer be able to block OS processes, so it will require external network filtering hardware.
I’ll likely end up with Linux on the 16”, and use the new one for things that are not secret/private.
https://appleterm.com/2020/10/20/macos-big-sur-firewalls-and...
Looks like it will be impossible to use Apple Silicon (without external network hardware) without revealing your track log to the CIA. How cool is that?!
Many don't even want to pay for the MacBook Pro's Touch Bar and many will probably see an Air's fanless design as an advantage over Pro, even if its CPU is throttled a little more often in sustained high CPU workloads. Complete silence is just that good. And it's going to be so much cheaper.
I think the star of the show yesterday was definitely the MacBook Air.
Those are supposed at just as much of a disadvantage compared to the new Apple Silicon Air and Pro.
That's where shock comes in.
my only pb are the smears on the screen, otherwise i think i'll keep it for another 7 years.
I'll be interested to know how much active cooling and binning real do for Rustc or LLVM compilation times.
And the GPGPU: https://browser.geekbench.com/v5/compute/compare/1800719?bas...
Also the i7-1165G7 is a 12-28w part, configurable by the OEM. I'd assume the XPS 13 is running it at top spec, but that'd also need validation.
You will see much higher specs on the Mac Mini Pro, MacBook Pro 14/16 and Mac Pro planned for next year.
Video editing's all done on disk isn't it? It's not like editing programs are loading a 20 GB file into memory?
Even just 8 GB has never given me any problems with video editing.
What am I missing here?
It's just that... swapping data in and out of ram is far slower than clock speed, right? So if the bottleneck is RAM, would it make sense to get the most RAM? (ie: Intel Mac)
I'm using Tableau Desktop in my daily work and until it is available on this new platform - these Macs are not an option for me even if they are 10x more performant. I guess there are a lot of professionals that are constrained in a similar manner. So we will see if M1 is adopted in such scenarios at all.
Final Cut Pro is one of the options for the new Macbook, on the item's page.
And whilst they are under NDA you do often see them running anonymous benchmarks like Geekbench.
I'm just making sure everyone understands that it does not mean Air will have the same real life performance as Pro or Mini. If you compare top MacBook Air 2020 (Intel)[1] with lowest MacBook Pro 2020 (Intel)[2] their results are almost identical, but their real life performance was not even close - Air starts to throttle after just few minutes of work. At the moment there is no reason to believe that Apple's CPU will behave differently (after all, the fan is in Pro and Mini for a reason).
[1] https://browser.geekbench.com/v5/cpu/4652268 [2] https://browser.geekbench.com/v5/cpu/4653210
New MBA has a 1719/6967 on gb 5.3.0
Intel is now #3
The reason that graph doesn't include the A14 Firestorm -> M1 jump was simply timing. We know the thermal envelopes of the M1 and the cooling designs. We now have clock info thanks to GB5. So yes, the data is pretty solid. No one's saying that the iPhone beats the Mac (or a PC) at performance when you consider the whole system. Just that the CPU architecture can and will deliver higher performance given the M1 clock, thermals and cooling. Remember that The A14/M1 CPUs are faster at lower clock speeds.
it's unfortunately drowned out by the cpu throttling scandal on google, but, its well-known in ar dev (and if you get to talk to an apple engineer away from stage lights at wwdc) that you have to proactively choose to tune performance, or you'll get killed after a minute or two due to thermal throttling.
Imagine going on a hike and climbing an exponential slope like 2^x. You go up to 2^4 and then go down again and repeat this three times so you have hiked 12km (43) in total. Then there is a athlete who is going up to 2^8. He says he has hiked 8km and you laugh at him because of how sweaty he is despite having walked a shorter distance than you. In reality 32^4 (48) is nowhere near 2^8 (256). The athlete put in a lot more effort than you.
That's why A14 only runs at 1.8ghz base, 3ghz boost. That's how it has low power consumption. And similarly Intel pushing 5ghz is why it has high power consumption.
TSMC's 5nm will have a raw transistor performance/watt advantage, but it's not huge
Check out for example the per core power charts that Anandtech does: https://www.anandtech.com/show/16214/amd-zen-3-ryzen-deep-di...
Compare for example the 1 core power numbers between the chips. The 5600X 1 core result is 11w @ 4.6ghz, whereas the other two chips boost higher and hit 4.8-4.9ghz 1 core turbos, but it costs 17-18w to do it. Huge increase in power for that last 1-2% performance. So you really can't or shouldn't compare more power-concious configurations with the top end desktop where power is infinite and well worth spending for even single digit percentage gains.
And then of course you should also note that the single-core power draw in all of those is vastly lower than their TDP numbers (65w for the 5600x, and 125w for the 5800x/5900x).
Yeah comparing TDP is meaningless even within the same processor. The 4 core workload in this table uses 94W and the 16 core workload uses 98W. There is also an anomaly at 5 cores where the CPU uses less power than if it only used 4 cores.
If you tried to derive conclusions about the power efficiency of the CPU you would end up making statements like "This CPU is 3-4 times more power efficient than itself"
For what it's worth, I have a fully specced out 16 inch MacBook Pro with the AMD Radeon Pro 5600m and even with that I'm regularly hitting 100% usage of the card, and not to mention the fan noise.
Looking forward to a version from Apple that is made for actual professionals, but I imagine these introductory M1 based devices are going to be great for the vast majority of people.
It will be very interesting to see what the performance will be of the more "pro" chip that overcomes those limitations that they'd put in the 16" and iMacs
https://appleinsider.com/articles/20/11/11/how-apple-silicon...
I wonder if that includes AirPlay screens, or just wired.
This could be solved by the GPU engineers by removing the legacy compatibility. "Nobody" boots a modern PC from BIOS anymore
Doesn't seem very "pro" to me. The MBP16" intel has 4 x USB-c ports, can drive two monitors, and can have >= 32GB ram.
Again, what definition of "pro" are you using, and how is that relevant to professionals with other occupations?
But regardless of professional requirements, "Pro" in Apple's product line just means "the more expensive slightly better version." Nobody's arguing that AirPods Pro are the wireless earbuds of choice for people who make money by listening to wireless earbuds.
I'm a Sofwware Engineer / pro photographer / videographer.
I sling code from time to time, edit thousands of RAW files from my cameras and edit together 1080p footage day in and day out.
I did that for years on a 2013 MBA with 8GB of RAM. Now I have a 2015 MBP with 16GB of RAM. It's perfectly adequate.
I ordered a 16GB Pro the other day to be my personal dev machine. I’m sure it’ll be more than fine. I’m upgrading from a 2013 8GB Pro which was only just starting to slow me down.
My resource hogs are slack, mainly the browser, and Zoom calls are apparently the most computationally intensive thing in the world, especially if you screen share while you have an external plugged in.
Memory wise the reason I had to go from 8Gb to 16GB on my personal laptop was literally just for TravisCI.
Honestly, adding external monitors cripples MacBooks pretty quick, even unscaled 2 2k monitors will slow a 2015 15 down significantly (don't try and leave YouTube on either), and it gets worse from there once you start upgrading to 4k monitors. a 2017 15 is good for a 4k and a 2k, and gets a bit slow if you try and go dual 4k.
I planned on looking into eGPU solutions until IT offered me a new Macbook, and I convinced them I needed a 16" Pro.
tldr: External monitors or badly optimized applications (Zoom, YouTube, or browser based CI) will make most MacBooks feel sluggish pretty quick.
The two limit seems to be for 6016x3384 which I assume is 5k.
https://www.reddit.com/r/macbookpro/comments/gs6bal/2019_mbp...
My XPS15 regularly throttles and sounds like it’s about to take off.
Thin and light is great for short bursts of activity, but, when you need sustainable heavy usage, you'll need a bigger computer, even if it's just to have a bigger heatsink.
By the way, “throttling” refers to CPU _slowing down_ despite cooling working at full capacity, so loud fans in itself isn’t one.
e: Another way to explain thermal throttling would be “thermal fading”, like brake fading on a car. Whether brake fading is considered a design fault or a feature that allow bursts of stronger brakes is semantic.
https://www.theage.com.au/technology/apple-fans-burned-by-ho...
We have no idea what the M1 would look like. Though it's likely. I imagine they wouldn't have taken the fan out of the Air if that was a big concern.
And it's doing this while using more than an order of magnitude less power (10W vs. a TDP of 125W for that Intel part).
That's stunning.
The fanless Intel Core-M CPUs could post excellent benchmark scores (for its time). But if you give it a lengthy compile task, it'll slow down dramatically.
Or are you saying that the test needs to run for way longer to be fair?
Yes, the main computing constraint of mobile devices is heat management (This doesn't really reflect the CPU but the complete device. Putting the CPU in a more ideal setup like a traditional desktop or water cooling will improve the CPU's performance in longer tasks)
Short duration single core workloads workout a fan is trivial even for CPUs that aren't trying to do so.
That's the wrong conclusion to make. For instance, the Lenovo ThinkBook 14s (with a Ryzen 4800u) with a 15W TDP posts the same Geekbench multicore scores [1] as the M1 Macbook. But the ThinkBook isn't in any way faster than the top-end iMac for real world compute intensive tasks.
The M1 certainly looks efficient, but there's little you can conclude from a single benchmark running for a very short period of time.
[1]: https://browser.geekbench.com/v5/cpu/4642736
And it’s 45% slower in single core.
Most importantly, the M1 is estimated to cost Apple $65, the 4800u is a $300+ part.
This comparison looks at different segments of the fab<>manufacturer<>OEM relationship. Add the user in there and you might say that you can buy an AMD CPU for $100 but an Apple CPU will cost you $1000. Not very meaningful as a comparison.
Two and a half times higher cost to build a slower, more power hungry CPU is not actually very similar.
You can argue that this particular Ryzen has a higher gross margin, say 50%, and lower ASP than $300, but that only gets your cost down to what, $140? And with RAM costing extra.
https://www.anandtech.com/show/16226/apple-silicon-m1-a14-de...
I imagine they will eventually have enough chips to start binning for different performance levels like AMD and Intel do.
The benchmark is just ... that accurate.
They are not binning for how high the cores will clock, which is just how business is done with Intel and AMD.
Does the M1 performance have to be ramped down during sustained use due to exceeding thermal envelope of the fanless MBA? Of the fan’d MBP?
We’ll know soon enough!
That said, it's promising and I'm really curious to see where this development will lead to in a few years' time.
iOS/iPhones/iPads already smoke every other device in running javascript. Apple Silicon may end up being the best thing to ever happen for Electron apps.
Nevertheless, because Apple has chosen to not increase their manufacturing costs by including more big cores, the multi-threaded performance is not at all impressive, being lower than that of many much cheaper laptops using AMD Ryzen 7 4800U CPUs.
So for any professional applications, like software development, these new Apple computers will certainly not blow away their competition performance-wise, and that before taking into account their severe limitations in memory capacity and peripheral ports.
But the truth is comparing to future offerings is bullshit, and we have to stick to what's available today. Impressive power/performance and all that, I have to say. We will see how sustained load looks like and how it runs non-optimized software. But to put in perspective 1 CCX of zen 3 performs better on 7nm (but draws up to 65W). With approximately the same die size (although w/o GPU and other things, the M1 has).
I still bet on the i9, but it'd be interesting to run a test.
When I said there is no magic, I was warning that we shouldn't expect huge speedups or a crushing advantage, at least not for long. The edge M1 has is due to a simpler ISA (which is less demanding to run efficiently, freeing more resources for optimization and execution) and a faster memory interface (which makes an L3 miss less of a punishment). This fast memory interface also limits it to, for now, 16GB of memory. If the dataset has 17GB, it'll suffer. Another difference is that all of the i9 cores are designed to be fast, whereas only 4 cores of the M1 are. This added flexibility can be put to good use by moving CPU-bound processes to the big cores and IO-bound and low-priority ones to the little ones.
In the end, they are very different chips (in design and TDP). It'd be interesting to compare them with actual measurements, as well as newer Intel ones.
Anandtech's deep dive provides several examples of advances in Apple's core design that didn't involve magic or breaking the laws of physics. For example...
Instruction Decode:
>What really defines Apple’s Firestorm CPU core from other designs in the industry is just the sheer width of the microarchitecture. Featuring an 8-wide decode block, Apple’s Firestorm is by far the current widest commercialized design in the industry. Other contemporary designs such as AMD’s Zen(1 through 3) and Intel’s µarch’s, x86 CPUs today still only feature a 4-wide decoder designs
Instruction Re-order Buffer Size:
>A +-630 deep ROB is an immensely huge out-of-order window for Apple’s new core, as it vastly outclasses any other design in the industry. Intel’s Sunny Cove and Willow Cove cores are the second-most “deep” OOO designs out there with a 352 ROB structure, while AMD’s newest Zen3 core makes due with 256 entries, and recent Arm designs such as the Cortex-X1 feature a 224 structure.
Number of Execution Units:
>On the Integer side, we find at least 7 execution ports for actual arithmetic operations. These include 4 simple ALUs capable of ADD instructions, 2 complex units which feature also MUL (multiply) capabilities, and what appears to be a dedicated integer division unit.
On the floating point and vector execution side of things, the new Firestorm cores are actually more impressive as they a 33% increase in capabilities, enabled by Apple’s addition of a fourth execution pipeline.
https://www.anandtech.com/show/16226/apple-silicon-m1-a14-de...
This is one place where the 64-bit ARM ISA design shines: since all instructions are exactly 4 bytes wide and always aligned to 4 bytes, it's easy to make a very wide decoder, since there's no need to compute the instruction length and align the instruction stream before decoding.
That's exactly what I said. It's faster, but not an order of magnitude faster and different workloads will perform differently depending on a multitude of factors (even if benchmarks don't). Do not expect it to outperform a not-too-old top-of-the-line mobile CPU by a large margin.
Zen 3 slightly outperforms the iPhone chip, but it runs it's clocks slower to stay inside a 5 watt power draw.
https://www.anandtech.com/show/16226/apple-silicon-m1-a14-de...
So, yes. Expect it to outperform Tiger Lake and Zen 3, at least on a per core basis.
Apple chips with more cores will come in time as well.
It's the per core performance, especially at a given power draw, that matters going forward.
cf. Anandtech article from 2003:
https://www.anandtech.com/show/1083/2
I've become something of CPU collector in recent years, and I have a nice line of p6 cpus from thePentium Pro -> Pentium 2 -> Pentium 3 -> Pentium M -> Core 2 that conveniently sidesteps those awful Netburst p4 CPUs.
It feels like this (p6+) microarch has finally run out of road and needs a rethink. What 'saved' intel was a change in philosophy rather than chasing MHz they chased power savings. And with Apple's new chips that history is repeating itself (and appears to be with a similar outcome).
It's an exciting time for hardware again because Intel and AMD are going to have to react to this and I think there's still legs to x86, it's survived everything thats been thrown at it so far...
I bought i486DX for $20 month ago, as a memo of my first CPU.
Also, some one, some day, had to disrupt it. Maybe this is it, maybe not.
[1] https://www.anandtech.com/show/12689/cpu-design-guru-jim-kel...
Edit: hadn't clicked through to your article. my comment is redundant
https://www.macworld.com/article/3572624/tsmc-details-its-fu...
This seems like a really cool piece of technology, and I'm kind of bummed that everyone is so cynical and pessimistic about everything these days (albeit understandably so).
Hopefully they'll reverse that decision. (or their GPUs in the higher end machines will have to be really good)
Apple lead the fray again, it's incredible.
Here is the image: http://www.cgchannel.com/wp-content/uploads/2020/06/200623_A...
I’m not sure if it supports it immediately, but this isn’t a difficult change, so it will certainly come soon.
For GCC targeting arm64 macOS, the dev branch is at https://github.com/iains/gcc-darwin-arm64/ currently.
For Rosetta, everything runs except virtualisation apps.
https://en.m.wikipedia.org/wiki/Windows_RT
That said, I am definitely waiting for at least one generation to pass before I jump on the train.
I’ve been through this before. It will be great, but Apple is a master at smoke & mirrors. Things will not go as smoothly as the sizzle reels make it seem.
Docker still doesn't run on Apple Silicon Macs, so the migration path is already disrupted here.
Many use cases, especially when it comes to microservices can almost always better served by something else
https://www.apple.com/newsroom/2020/06/apple-announces-mac-t...
The problem here is that in WWDC, Apple showed Docker Desktop running on an Apple Silicon system (maybe suggesting that it is at least running) and here we are in November it is still not running or not known if Apple Silicon is supported.
I don't see any patches in the Docker repositories on such support, thus maybe Apple has a private fork for Apple Silicon. In general it is not available to us and we don't know when it will be.
I think their new architecture will be awesome. Personally, I look forward to being able to debug iOS Bluetooth apps in the simulator, and I think some of the new form factors will be pretty cool.
But I can’t help but notice the current dearth of AAA apps that are already universal.
A transition like this is a big deal; especially if you have an app with thousands of function points, as it requires a complete, top-to-bottom re-test of every one.
In the unlikely case that you won’t find issues, it will still take a long time. Also, most companies won’t bet the farm on prerelease hardware; instead, using it to solve issues. They will still need to test against the release hardware before signing off for general distribution.
Also, I have a couple of eGPUs that I use. I don’t think the new architecture plays well with them.
I’m hoping that the next gen will obviate the need for them. They are a pain.
https://developer.apple.com/documentation/apple_silicon/abou...
Everyone is on Xcode these days and using much higher level frameworks. Adobe and Microsoft already announced early 2021 availability of native binaries.
Besides, Rosetta 2 is even more impressive than the original was. I’m betting it will be a breeze. 6 months in and almost everything will be native.
> fun fact: retaining and releasing an NSObject takes ~30 nanoseconds on current gen Intel, and ~6.5 nanoseconds on an M1
> …and ~14 nanoseconds on an M1 emulating an Intel
https://mobile.twitter.com/Catfish_Man/status/13262384342355...
I won't be replacing my workstation this weekend but then, I won't be updating it to Big Sur just yet either. I am getting a new personal machine sometime in the next year though I don't imagine I'll be buying an Intel Mac.
The more interesting thing is the power efficiency, which doesn't have that much impact on single thread performance because higher power CPUs don't actually use their entire power budget for a single thread. But that's an impressive multi-threaded score for that TDP. It gets stomped by actual desktop CPUs for the obvious reason, but it has better multi-threaded performance than anything with the same TDP. Though that's also partially because the low-TDP Zen 3 CPUs aren't out yet.
What I'd really like to see is some benchmarks that aren't geekbench.