M1 as an "Industry Game Changer"

[KPOM]

I mostly bring up the Surface Pro X because it's about the only Windows 10 ARM device you can reliably get your hands on right now. Which is part of the problem. And in terms of impact/speed, Apple's approach is ridiculously more effective.

Apple announced the ARM switch at WWDC. We already have Office and Photoshop betas available to use on the M1. Most of the smaller software I use has already gotten ARM builds pushed out into an official release of that software before I even get my M1 Mini. I really wouldn't be surprised if Apple completes the transition, and we're mostly just talking about the handful of straggler developers, while Microsoft is still "making a big push".

The advantage Apple has is that, because of their relatively small market share, they can afford to drop compatibility with legacy software/hardware. Microsoft has built Windows to be backward compatible. While they did finally drop support for 16-bit software, it took a while. Dropping 32-bit software support isn’t a realistic option for them as it was for Apple. By doing so, they could optimize Rosetta 2 by focusing on translating x86-64 code.

That said, Microsoft could clearly improve its efforts on ARM, even with the limitations of the need to maintain backward compatibility. That the M1 appears to be twice as fast as the Surface Pro X running a completely non-optimized version of Windows ARM on a hacked QEMU VM demonstrates as much. Perhaps the new Qualcomm 888 chip will help.

 

[wyatterp]

This discussion is very interesting! I think yes, but it will not be an instant pivot point. This is the harbinger moment akin to iPhone being introduced in 2007. I'm not a complete Apple Junkie in saying that, I own a nice high end desktop PC that uses 500 watts of power - however I'm using my MBA M1 everywhere and I've loaded 18 games from my PC games library on it that all run well enough on a machine that can do so for hours away from a plug, and with ZERO noise emissions. I think things pivot in conjunction with 5G rollout once Apple gets the 5G modem integrated on the M1 SoC - hasn't this "always connected, always on" battery champion mobile device the goal for years now? MS has wanted to do this - they just don't have the chipset to use - and yes, Nvidia sees this too. I'm buying more Nvidia and Apple Stock for now.

 

[Gerdi]

It is true, but I don't think the situation is as simple as you make it sound. Cortex X1 represents a big jump in performance simply because it is the first attempt by ARM to design CPU IP focused on performance rather than on die area and low power consumption. It is far from certain they will be able to keep this momentum going.

ARM has the last few years provided an IPC gain of at least 20% - and this was their official roadmap target. Of course ILP is limited - so i dare the predication that ARM will rather close the gap instead of Apple being able to widen it.
The Cortex X1 was just a logical consequence of what ARM is doing since Cortex A73 - providing a wider design each year.

 

[KPOM]

ARM has the last few years provided an IPC gain of at least 20% - and this was their official roadmap target. Of course ILP is limited - so i dare the predication that ARM will rather close the gap instead of Apple being able to widen it.
The Cortex X1 was just a logical consequence of what ARM is doing since Cortex A73 - providing a wider design each year.

The gap will undoubtedly narrow since Apple caught the industry off guard the last few years, but I do think Apple will maintain its lead because it has such great vertical integration.

 

[theorist9]

Agreed, from an Intel Mac pricing for new and used dropping like a rock - I'd say the M1 is a gamechanger!

I can't even pay someone to take my old Intel Macs off my hands while before the M1, it fetched a premium!

I expect I'd be willing to take those old Intel Macs off your hands without your having to pay me much at all. ?

 

[littlepud]

The problem with that is that Apple Silicon is not arm alone. It's arm +. The + is the unbelievable number of accelerators built around the chip. Two problems for those who want to copy- One Apple has this thing patent protected to the hilt and two, Apples' year over year performance gains mean that you'll never keep up.

Nvidia looks to be interested in the server market. Much higher margins and if used for VDI, you can sell GPU's (very expensive ones) too.

Very much THIS. The real game-changer with Apple Silicon is not because the processor uses the ARM ISA. The ISA itself is cleaner than x86 but not intrinsically better-performing. The big deal with Apple Silicon is that it eschews the traditional system architecture for a SoC, and basically puts an ASIC on-die for anything Apple feels is important.

 

[Argon_]

Very much THIS. The real game-changer with Apple Silicon is not because the processor uses the ARM ISA. The ISA itself is cleaner than x86 but not intrinsically better-performing. The big deal with Apple Silicon is that it eschews the traditional system architecture for a SoC, and basically puts an ASIC on-die for anything Apple feels is important.

Massive reduction in latency, less space taken up by PCB=more space for batteries and cooling.

 

[torncanvas]

The Secret sauce is that these two things go together. Every device has a maximum thermal envelope. And at every given level, AS offers more performance at a lower wattage. Wattage leads to heat. Less wattage =less heat. So if a cpu is more efficient at power per watt then you can have more performance in a given thermal envelope. I saw a test running Cineabench 23. The Apple silicon system completed the test , never drawing more than 13 watts. The 2020 iMac completed it drawing 125 watts. The AS system stayed cool and it’s fan never left idle. The iMac ramps up its fans to max in less than 30 seconds. When Apple takes the large thermal envelope of the iMac and puts a more powerful M series processor in it, they will be able to bring much more performance than in today’s iMac because more performance is possible because the AS brings more performance per watt.

Yeah it'll be interesting to see what happens. It's not as inevitable as you make it sound, since every performance curve is logarithmic and it's unclear where the "knee" is with the fundamental design of Apple SoCs. But chances are good that it's a ways farther to the right than 10W, the nominal M1 wattage. I expect in the next year they'll have conquered the laptop segment and most consumer desktops in terms of performance for a given class.

IMO the big question now is the higher end; can they conquer the Ryzen 5950X or even Threadrippers? PC cooling is pretty sophisticated these days, you can build/procure a system with top-of-the-line CPU & GPU that fits in <15L, runs acceptably cool, and stays in the high-30s dB at load (larger systems can be noise-floor silent of course). That's a good quality of life, and hitting the same performance but only lower wattage isn't really going to draw all the creative pros I know away from the fast PC desktops they're leaving Apple for. It'll need to be faster, regardless of wattage. But that's only possible if the knee of Apple Silicon's curve is high enough.

Massive reduction in latency, less space taken up by PCB=more space for batteries and cooling.

Have you seen a teardown of the Mac Mini? The "less space taken up" was replaced with...nothing. For those familiar with small form factor PCs, it's like a cave in there with all that air! While I think they could have definitely made an upgradable SSD, it does make me wonder if the M1 isn't limited to 16GB RAM for some technical reason.

 

[Argon_]

Have you seen a teardown of the Mac Mini? The "less space taken up" was replaced with...nothing. For those familiar with small form factor PCs, it's like a cave in there with all that air! While I think they could have definitely made an upgradable SSD, it does make me wonder if the M1 isn't limited to 16GB RAM for some technical reason.

Well we're talking about a first gen device, and evidently one with a lot of headroom; just the beginning of a relentless march. Also we may see a smaller form factor Mini eventually.

I wonder what a M1 Mini could do with a water cooling loop, and 40C full load temperatures. It might exceed Zen3 in single core just from the reduced resistive losses.

 

[Gerdi]

Very much THIS. The real game-changer with Apple Silicon is not because the processor uses the ARM ISA. The ISA itself is cleaner than x86 but not intrinsically better-performing

Designs implementing ARM ISA have intrinsically higher performance potential (under both iso power or iso area) for 4 main reasons:
1) fixed length instruction coding
2) load/store architecture
3) weakly ordered memory model
4) more GP registers

There is no way of fixing above issues with x64 ISA. In fact all more modern ISAs like RISC-V have above properties - because there is the experience and learning from the past 40 years of computer architecture put into these modern ISA designs.

 

[Argon_]

delete

 

[torncanvas]

Well we're talking about a first gen device, and evidently one with a lot of headroom; just the beginning of a relentless march.

I wonder what a M1 Mini could do with a water cooling loop, and 40C full load temperatures. It might exceed Zen3 in single core just from the reduced resistive losses.

Heck yeah haha! That'd be fun to see people get nerdy with it. Speaking of, one of my reasons for choosing a MBP vs MBA is holding out hope that it can be overclocked before too long. There's a lot of unused thermal headroom with the Mac Mini, yes, but also with the MBP to a lesser extent.

 

[Argon_]

Heck yeah haha! That'd be fun to see people get nerdy with it. Speaking of, one of my reasons for choosing a MBP vs MBA is holding out hope that it can be overclocked before too long. There's a lot of unused thermal headroom with the Mac Mini, yes, but also with the MBP to a lesser extent.

Yeah the 2 port MBP cooler can handle about 25W with a bit of margin for dust and thermal paste degradation. I know this from my machine. That's plenty of headroom for the 12-14 watt continuous M1. I wonder what the benchmark results would be if you compared the stock fan curve to a custom one, say a 45-70C ramp up. The reduction in resistive losses might get you a few percent.

 

[torncanvas]

I'm mostly interested in overclocking the GPU, if I could get an extra 5-10% that'd be nice (for content creation). Undervolting my gaming laptop utterly transformed it into an entirely different-feeling machine, so it's like a whole new world to me now (cue singing)

 

[dingclancy23]

The M1 has been lauded, by tech journalists and users alike, as an industry game changer for personal computing. I'm curious: Do you agree with that statement? If so, how do you think the personal computing industry has been forever changed by the advent of the M1 and Apple Silicon Macs at large?

Personally, I do think it was a fantastic move for Apple and I would argue that it's a game-changer for the Mac. But I don't see the personal computer industry changing or adopting similar strategies to this as a result of Apple doing it. Do I think we'll see more SoCs in non-Mac personal computers? Abso-friggin-lutely. But we're never going to see a computer maker own the entire hardware and software stack the way Apple now does with Apple Silicon Macs like the ones we now have with M1. Microsoft may have an SQ1 or SQ2 for the Surface Pro X, but that thing is a Qualcomm SoC. Samsung makes SoCs for its phones and tablets, but Samsung isn't Samsung's only customer for those SoCs. And while they do have their own version of Android (albeit one of the worst ones out there), it's not their OS underneath it all! I think Microsoft and Samsung have the best chance of trying to follow Apple on something like this. Maybe NVIDIA, now that they own ARM Holdings. But I think any one of the three of them doing it would take so much time to catch up to Apple. So, no, I don't think it's an "Industry Game Changer"; though I do think it's a massive game changer for the Mac itself. What say you all on this? Do you think the personal computing industry will forever be changed by this? And if so, how and when?

The scary thing is, there will be a version in 2022 that will run M1 which is positioned like an iPad or iPhone SE.

Mac Mini is already at $699. They can pull of a $599 Mac that will be a workhorse now that they own the chips.

There are also a lot of tech in the current iPhones and iPads that are not on the Mac because of Intel. Think Pro-motion display, thinner and lighter, more sensors, Face-ID, better cameras, 5G and more.

The will re-engineer the heck out of this thing.

Other laptop manufacturers will have these own set of features too, but their lack of integration will result in a poorly-designed device. Anything with a large noisy fan will not allow the form factor that Apple will be making.

The only manufacturer capable of holding their own is Samsung since they are the only other company making great phones. HP, Asus, Dell, will not be ready.

 

[throAU]

The problem with that is that Apple Silicon is not arm alone. It's arm +. The + is the unbelievable number of accelerators built around the chip.

Aware of that. Again, these aren't magical apple pixie dust. They're just dedicated ASICs doing stuff that anyone else could also add if they're on TSMC 5nm manufacturing process (and thus have a state of the art manufacturing node) and have the silicon die space.

The big problem will be on the software side, but Nvidia already has a pretty good foothold there with regards to ML/AI.

What I'm saying is that Nvidia or someone else could implement on die resources to do the same sort of things. It isn't magic that only apple know how to do. It just that Apple are the only company to try it and actually put a product out to market. Yet.

What apple is doing in M1 is cutting edge for sure - but it isn't stuff that nobody else understands how to do. It simply that Apple had the financial resources to spend on it without it being a "safe bet". Now Apple HAVE done it and proven what is possible (and more importantly, financially viable, and producible in volume), others will follow.

 

[Brien]

Zen3 can match the M1 in single core, albeit with a much higher power demand.

Intel better get it’s stuff together.

 

[throAU]

Very much THIS. The real game-changer with Apple Silicon is not because the processor uses the ARM ISA. The ISA itself is cleaner than x86 but not intrinsically better-performing. The big deal with Apple Silicon is that it eschews the traditional system architecture for a SoC, and basically puts an ASIC on-die for anything Apple feels is important.

Yup, but I'd say its not "because it is apple silicon".

More that "it isn't x86 and hindered by 40 years of hacks and bandaids on top of a sub-optimal design, that tries to maintain backwards compatibility at the binary level".

As you say, its due to the decisions apple made (that anyone else can also make - if they so choose) - not just because "Apple Magic!".


edit:
Also, m1 is just the start for apple as well. they aren't even using HBM on the thing which will be significantly stifling the thing for bandwidth.

Replace the DRAM with on-package HBM (More expensive, but hey - premium product) for higher end SKUs and you'll see a big performance gain from that alone is my bet. I can almost see M1 like packages with onboard HBM2 stuck into many sockets on something like a Mac Pro.

i.e., higher end Mac Pro has 8 sockets which have SOC+HBM packages installed (No traditional RAM slots) so RAM/CPU and GPU processing scales at the same time. Upgradable, but Apple proprietary. Would be right up their alley

 

[leman]

First benchmarks of ARM Cortex X1 have appeared (https://www.notebookcheck.net/The-Q...comes-out-swinging-on-Geekbench.507553.0.html).

The X1 running at 2.8 ghz is basically equivalent to an A12 running at 2.5 ghz. That’s pretty good, but not nearly enough to challenge x86 CPUs.

 

[NotTooLate]

First benchmarks of ARM Cortex X1 have appeared (https://www.notebookcheck.net/The-Q...comes-out-swinging-on-Geekbench.507553.0.html).

The X1 running at 2.8 ghz is basically equivalent to an A12 running at 2.5 ghz. That’s pretty good, but not nearly enough to challenge x86 CPUs.

The biggest gainers from this "Apple flex" are 3 of the biggest tech companies in the world that are currently non players in the PC world - Quallcom /Nvidia/ Google , if Microsoft won't transition fast enough , you will see Google picking up the slack on the low end (for starters) and then chew right through everything MS has to offer with lower prices , good enough performance and amazing battery life , they will choose between Nvidia and Quallcom (Samsung even?) for the HW portion.

I don't see MS holding back the WoA licensing for much longer , otherwise it will be a dead platform , and they can get behind Google again , history will repeat it self , I predict WoA on AS sooner rather then later , as well as putting more efforts to ramp up the OS to be usable and not just a tech demo as it stands today.

 

[laptech]

I am curious as to why people think Intel need to up their game. Intel only needed to up their game if they wanted to still be the CPU of choice for Apple.

The only people that are going to buy Apple macs are existing Apple mac owners, employees who use mac's at work and therefore want a mac for themselves and students who use mac's at colleges and universities and therefore want a mac themselves.

The only 'game changer' in my opinon is if Apple are able to produce an ARM machine that has a dedicated GPU for gaming and high end video editing whilst able to maintain low power usage. Until that day arrives Intel has nothing to worry about in my opinion.

 

[quarkysg]

The only 'game changer' in my opinon is if Apple are able to produce an ARM machine that has a dedicated GPU for gaming and high end video editing whilst able to maintain low power usage.

IMHO the M1 Macs already shows that this can be done. As it is today, the M1 Macs (entry level) has been shown to be able to run existing games titles that is impossible to even run acceptably (if at all) on entry level Intel Macs. It can already run games that requires dedicated GPUs on Intel Macs. It also does video editing tasks rivalling the 16" MBP.

It should be easy for Apple to double the M1's CPU and GPU (maybe tripling) cores, use faster memory, and it'll be a very potent Mac.

I don't Apple needs to beat Intel, AMD or Nvidia's best in term RAW performance.

 

[leman]

The only 'game changer' in my opinon is if Apple are able to produce an ARM machine that has a dedicated GPU for gaming and high end video editing whilst able to maintain low power usage. Until that day arrives Intel has nothing to worry about in my opinion.

They don’t need a dedicated GPU. Apple GPUs are currently the fastest shipping GPU IP on the market. Next year they will have machines with 16 or more GPU cores which will offer similar raw performance to mid-range dedicated GPUs while featuring zero-latency data synchronization and ultra-low power consumption.

 

[BootLoxes]

I am curious as to why people think Intel need to up their game. Intel only needed to up their game if they wanted to still be the CPU of choice for Apple.

Not really. They have pretty much lost the entire pc gaming base to AMD. If I remember, many servers are also swapping over to AMD. Not sure if thats true but could have sworn I read it somewhere.

They are gradually losing more and more markets. Intel needs to change asap

 

[jido]

First benchmarks of ARM Cortex X1 have appeared (https://www.notebookcheck.net/The-Q...comes-out-swinging-on-Geekbench.507553.0.html).

The X1 running at 2.8 ghz is basically equivalent to an A12 running at 2.5 ghz. That’s pretty good, but not nearly enough to challenge x86 CPUs.

Did you notice that it is a mobile phone being benchmarked?

The A14 is also getting lower scores than the M1, despite having the same cores.

I think that's a good showing from Qualcom and I expect more performance to come.