Technology Android
- Thread starter THEV1LL4N
- Start date
Some websites selling laptops powered by the AMD Ryzen 4000 series are reporting that the battery lasts around 4-5 hours. Confused with this as I thought efficiency had been significantly improved in the Zen 2 microarchitecture as it moved 7nm CPUs. I think I will wait a little longer to get a laptop with a 5000 series CPU which I hope gets released this year.
New Macs announced today. I missed the live event but I drifted in and out of some post-event livestreams discussing what they were ordering from the event. I'll also wait until a deep dive is done by someone like Anandtech because right now it's all reactions and I feel like there will be some things that come up, good and band, in the next few days.
There's no mention of the clock speed of the ARM chips and I'm curious as to why Apple did that. Someone will report it, though, but I still wonder what the reason is for Apple. Most likely "hiding" something that may turn buyers away but we'll see.
MBA, 13" MBP, and the Mac Mini. It'll be funny if someone bought a Mac Mini to use with their $6K ProDisplay from Apple lol. I don't think there will be many in the professional crowd that will do this but still, it's such a stark difference in cost.
I imagine we see the 16" models get ARM by next Summer, but that's just a guess. It make sense Apple rolls it out slowly, much like it's done in the past with big changes, like the USB-C ports in the 2015 MB, as well as the BF keyboard.
Pricing is still kind of shitty. $1299 gets you a 13" with 8 GB of RAM and 256GB of SSD but it's not surprising since it's Apple. But a 16/512 config will put you at $1700. That feels like the same price as a similar Intel 13" MBP. Maybe I was naive in hoping that a switch to ARM would see Apple lower its prices, even just a little bit, like $200 or so.
There's no mention of the clock speed of the ARM chips and I'm curious as to why Apple did that. Someone will report it, though, but I still wonder what the reason is for Apple. Most likely "hiding" something that may turn buyers away but we'll see.
MBA, 13" MBP, and the Mac Mini. It'll be funny if someone bought a Mac Mini to use with their $6K ProDisplay from Apple lol. I don't think there will be many in the professional crowd that will do this but still, it's such a stark difference in cost.
I imagine we see the 16" models get ARM by next Summer, but that's just a guess. It make sense Apple rolls it out slowly, much like it's done in the past with big changes, like the USB-C ports in the 2015 MB, as well as the BF keyboard.
Pricing is still kind of shitty. $1299 gets you a 13" with 8 GB of RAM and 256GB of SSD but it's not surprising since it's Apple. But a 16/512 config will put you at $1700. That feels like the same price as a similar Intel 13" MBP. Maybe I was naive in hoping that a switch to ARM would see Apple lower its prices, even just a little bit, like $200 or so.
https://www.theverge.com/2020/11/11/21560810/google-photos-unlimited-cap-free-uploads-15gb-ending
lol what a joke. Just migrated over from Flickr when they pulled the same crap about a year ago, and now this.
lol what a joke. Just migrated over from Flickr when they pulled the same crap about a year ago, and now this.
So I saw something that kind of addressed my question earlier about why Apple didn't state the clock speed of its ARM chips on the Mini, Air, and Pro models of the Mac. Dave2D said they are all the same clock speed, but maintaining the peak clock speed comes down to how well each model cools itself. That still seems dumb to not list the clock speed, though; just because one chip is spread across three models doesn't mean you don't give the specs on it.
I'll have to check and see if/when Anandtech does its review.
I'll have to check and see if/when Anandtech does its review.
I know we had this discussion, probably, a year ago about ARM chips in laptops and even desktops but the benchmarks I'm seeing for the M1 so far are pretty impressive.
https://www.theverge.com/2020/11/19...laptop-performance-intel-qualcomm-competition
And some the reactions by people know who and understand more about hardware also look to be really excited and impressed with the performance and expect it to grow with each update.
I know one thing we discussed was if these chips would be powerful enough to do heavy duty stuff, like video editing. I can't remember which video I saw but I think it was Dave2D's video that showed how long it took to do a task in Final Cut on an Intel MBP, the M1 MBA, and the M1 MBP and it was off by, maybe, a minute.
I'm sure it's not the only metric needed to judge these new chips but that's still some impressive performance for a task that a lot of people do. And since these are just the entry-level models of the MBP and Mac (the Mac Mini), I can see the 16" MBP packing some more power like this. And I think Apple has to be thinking about putting it in the iMac and Mac Pro in the next few years. Sooner rather than later. And I am excited to see what that brings. I don't know if 3.5 years is considered "old" by my MBP is doing fine right now and I expect another three years out of it, but the alleged 20 hour battery life and better performance of the MBA/MBP on M1 is going to be tough to resist when the 16" MBP starts to put out similar numbers.
https://www.theverge.com/2020/11/19...laptop-performance-intel-qualcomm-competition
And some the reactions by people know who and understand more about hardware also look to be really excited and impressed with the performance and expect it to grow with each update.
I know one thing we discussed was if these chips would be powerful enough to do heavy duty stuff, like video editing. I can't remember which video I saw but I think it was Dave2D's video that showed how long it took to do a task in Final Cut on an Intel MBP, the M1 MBA, and the M1 MBP and it was off by, maybe, a minute.
I'm sure it's not the only metric needed to judge these new chips but that's still some impressive performance for a task that a lot of people do. And since these are just the entry-level models of the MBP and Mac (the Mac Mini), I can see the 16" MBP packing some more power like this. And I think Apple has to be thinking about putting it in the iMac and Mac Pro in the next few years. Sooner rather than later. And I am excited to see what that brings. I don't know if 3.5 years is considered "old" by my MBP is doing fine right now and I expect another three years out of it, but the alleged 20 hour battery life and better performance of the MBA/MBP on M1 is going to be tough to resist when the 16" MBP starts to put out similar numbers.
I always thought it was a great move for Apple if they can actually move forward with the transition - their ARM processors are second to none and they made crazy progress each year, up until A13. The experience on Macbooks is going to be excellent as soon as the most used third party programs are ported to ARM instead of being emulated. Apparently at the moment it's mostly just Apple programs that run natively, and experience on emulated programs is always going to be sub-par. That said, with the small software library circling around a dozen or two of mainstream programs, it doesn't take much to transition 90% or so of their users effectively. With little in terms of legacy or exotic software, Apple's ecosystem was just a perfect candidate for a transition to ARM.
While these chips are great for casual usage, these don't have any of the advanced instructions (like AVX, AVX2, and AVX512). Apple was smooth by not even attempting to emulate them (Rosetta doesn't work on programs that take advantage of them) so you can't see comparisons. You can only see web and light multimedia tools, which is what ARM chips excel at. While that's perfectly fine for most Mac users, especially more casual users who may never even encounter more advanced instructions, this wouldn't fly on Windows anytime soon. It's these advanced instructions and the sheer amount of operations that Intel/AMD chips can accelerate that ARM chips can't that set these processor technologies apart. Compute resources on Intel/AMD chips are spread across all common instruction sets so they can handle anything you throw at them flawlessly. ARM chips use all of their resources to go 100% into the 20% of instructions that are used 90% of the time (maybe 99% of the time on Macs) and now attempt to compete or outperform traditional x86 chips there.
While these chips are great for casual usage, these don't have any of the advanced instructions (like AVX, AVX2, and AVX512). Apple was smooth by not even attempting to emulate them (Rosetta doesn't work on programs that take advantage of them) so you can't see comparisons. You can only see web and light multimedia tools, which is what ARM chips excel at. While that's perfectly fine for most Mac users, especially more casual users who may never even encounter more advanced instructions, this wouldn't fly on Windows anytime soon. It's these advanced instructions and the sheer amount of operations that Intel/AMD chips can accelerate that ARM chips can't that set these processor technologies apart. Compute resources on Intel/AMD chips are spread across all common instruction sets so they can handle anything you throw at them flawlessly. ARM chips use all of their resources to go 100% into the 20% of instructions that are used 90% of the time (maybe 99% of the time on Macs) and now attempt to compete or outperform traditional x86 chips there.
Last edited:
Will it ever change, ARM not being able to run advanced instructions? Or is it a limitation that won't be able to be overcome due to the architecture? I don't reasonably expect ARM to take over everything, at least not with more heavy, professional users, but since most of us are casual users, it looks like the vast majority of people would benefit from an ARM chip, Mac or not.
Is it possible that devs for Windows apps (some of them, that aren't heavy-duty apps for pro users) change the architecture they develop for? Or at least add ARM support? So that in the future, people with ARM Windows machines can still run these apps on a more low-powered ARM device?
Is it possible that devs for Windows apps (some of them, that aren't heavy-duty apps for pro users) change the architecture they develop for? Or at least add ARM support? So that in the future, people with ARM Windows machines can still run these apps on a more low-powered ARM device?
Also the Note lineup might be done. The S 21 Ultra and Z Fold 3 get the S Pen.
One YTer I watch has a Galaxy Fold. Or had one, until some software bugs had him try for an exchange and he ended up getting a Note 20 instead. But the Fold looked really cool but the reliability of the hinge was always in question as well as finding a case and screen protector for it.
Maybe after a few more revisions, Samsung will make it more reliable or less prime to damage. And maybe bring that price down, too
One YTer I watch has a Galaxy Fold. Or had one, until some software bugs had him try for an exchange and he ended up getting a Note 20 instead. But the Fold looked really cool but the reliability of the hinge was always in question as well as finding a case and screen protector for it.
Maybe after a few more revisions, Samsung will make it more reliable or less prime to damage. And maybe bring that price down, too
ARM is becoming "wider", meaning it's becoming heavier and less like original ARM was. Support for more stuff is being added. But if you go too far you'll end up with ARM chips looking more like today's x86 chips, except without the benefit of running legacy software. Which is imho why everyone leaves ARM as the light chips specialized at a narrow variety of tasks. They are efficient and these days perform well at those tasks. Make them more bloated and they will lose those benefits.
Who knows about the future, but people use Windows largely because of compatibility - every tool you can imagine will work on every Windows-running computer. It's a platform that had by far the most software made for it over many decades, it's still the case, and you can still run all of it no matter who makes your computer. Its biggest advantage would be gone if it went ARM, which is why it's not going to happen for a very long time, and the only reason it may happen in a very distant future is if ARM chips way significantly overtake x86/AMD/Intel processors in overall performance.
At the moment Microsoft is trying to offer ARM support just in case that happens, and there is an official ARM build of Windows that can run a small suite of ARM programs, but nobody takes it seriously, as it's not sufficient even for the most casual Windows users.
Who knows about the future, but people use Windows largely because of compatibility - every tool you can imagine will work on every Windows-running computer. It's a platform that had by far the most software made for it over many decades, it's still the case, and you can still run all of it no matter who makes your computer. Its biggest advantage would be gone if it went ARM, which is why it's not going to happen for a very long time, and the only reason it may happen in a very distant future is if ARM chips way significantly overtake x86/AMD/Intel processors in overall performance.
At the moment Microsoft is trying to offer ARM support just in case that happens, and there is an official ARM build of Windows that can run a small suite of ARM programs, but nobody takes it seriously, as it's not sufficient even for the most casual Windows users.
Last edited:
I wonder if Apple has plans and guidelines for devs to make sure ARM Macs are fully supported within the next few years, then. I mean, they have to if they've made such a huge shift from Intel to Apple Silicon. No way they sacrifice whatever userbase they have right now with Macs just to "try their hand at AS and see what happens. And while it would be a difference in CPU architecture instead of OS, like it was in the past, maybe Apple is OK with keeping Windows and macOS as parallels and never bothering to make them play nice with each other. Apple kind of tried bridging the gap 10+ years ago with Boot Camp but since you mentioned that Windows has such a stronghold because of compatibility, maybe Apple is banking on people adopting AS and Apple will do their part to make sure whatever programs are pivotal for Mac users with AS can still do everything their Windows counterparts can do.
Basically I'm saying Apple might have something up their sleeves that continues to keep Windows at an arm's distance while Apple tries to break ground on having apps for the AS architecture that everything that Windows versions of those apps can do. I guess that won't bode well for gaming as it has always sucked on Macs but maybe content creators can get a fully functional creative suite on an ARM machine from Apple.
Apple basically made a statement saying "you have to move over your software to ARM now, since we're moving over there and that's where all users will be soon". Apple has the benefit of having a small software ecosystem circled around a couple dozen of key programs. Barely supported niche tools or programs/games that haven't been updated in years and likely won't ever be are not something that's all that common on Macs.
Like I mentioned earlier, It's easy to manage 30-50 devs that make almost everything that your userbase uses and hope the remaining bunch ports their tools too. If not, too bad, some programs will stop working, but few people use them, and there are so few of those programs that some alternatives would crop up not long after.
That was never a problem that bothered Apple. So it's a perfect ecosystem to move to ARM.
If this happened on Windows, I would not be able to ever use most programs and games I used ever again. Businesses would collectively have to invest trillions to have their software rewritten, updated, entirely new ecosystems supported, processes redesigned and the hardware itself upgraded. Suddenly most companies would have to undergo major, expensive transformations, and many vendors would be fucked, especially if their tools were written like 20 years ago and nobody really understands how they work anymore. This was already a huge deal when Windows XP support ended, despite the fact that probably 99% of Windows XP software is compatible with Windows 10 with no rework required. Suddenly you heard of major national organizations relying 100% on some obscure software suite that turns out only works on Windows XP and they were willing to pay insane money to Microsoft to maintain support for a couple more years - then there were multi-billion dollar projects lasting years just to develop a new suite and transform the entire organization around it. As a matter of fact, I work on such project now.
I don't think PC would ever recover from a loss of such incredibly huge and diverse legacy software support - it's what makes that platform so great, and discarding it most likely would be the death of it.
Like I mentioned earlier, It's easy to manage 30-50 devs that make almost everything that your userbase uses and hope the remaining bunch ports their tools too. If not, too bad, some programs will stop working, but few people use them, and there are so few of those programs that some alternatives would crop up not long after.
That was never a problem that bothered Apple. So it's a perfect ecosystem to move to ARM.
If this happened on Windows, I would not be able to ever use most programs and games I used ever again. Businesses would collectively have to invest trillions to have their software rewritten, updated, entirely new ecosystems supported, processes redesigned and the hardware itself upgraded. Suddenly most companies would have to undergo major, expensive transformations, and many vendors would be fucked, especially if their tools were written like 20 years ago and nobody really understands how they work anymore. This was already a huge deal when Windows XP support ended, despite the fact that probably 99% of Windows XP software is compatible with Windows 10 with no rework required. Suddenly you heard of major national organizations relying 100% on some obscure software suite that turns out only works on Windows XP and they were willing to pay insane money to Microsoft to maintain support for a couple more years - then there were multi-billion dollar projects lasting years just to develop a new suite and transform the entire organization around it. As a matter of fact, I work on such project now.
I don't think PC would ever recover from a loss of such incredibly huge and diverse legacy software support - it's what makes that platform so great, and discarding it most likely would be the death of it.
Last edited:
That makes sense about Apple. I think they have good relations with certain devs so while it may seem to us, on the outside, that Apple is twisting arms to make things work for Apple devices, it might be more of a mutual thing and Apple and select few companies and devs just move as a herd, together. So if Apple makes a big shift, such as a switch to ARM, they still have a strong support cast for the most popular apps, like you said.
For Windows, it seems like its big strong point is its ability for apps from 20+ years ago to still run on Windows 10, like you said. And I agree, if Windows was to make such a radical shift like Apple did, then I think more people would find themselves at square-one in regards to which OS they would choose and might choose macOS/iOS out of bitterness towards MS for such a big shift.
I haven't looked too deep in to the implementation of iOS apps in Big Sur for ARM machines but I bet that might help cover some bases that macOS developers are unable to cover. But I feel like the first batch of iOS apps that run on ARM devices will be centered around social media and maybe a few games before the productivity apps start to make their way over, optimized for a Mac display.
Interesting. The M1 really seems to hold its own in single core stuff. It gets ass-blasted in multi-core, as the graph shows.
Is it physically impossible for this to improve to the level of Ryzen or even Intel, or is it something that will just take time and a lot of R&D to get it to that point? I'm not sure Apple would let it stay as-is in multi-core performance if it can ultimately be improved, albeit with time. And I'm not sure if they'd make such a radical shift to ARM if multi-core is going to suck compared to the others no matter what they do. I suppose it's still possible that Apple keeps the ARM designs for the entry-level models of their Macs but the Mac Pro and maybe iMac Pro stay with Intel/AMD, possibly, since it wouldn't be up to the mark with professional Mac users. Like the ones who buy the $6K ProDisplay and a $14K+ Mac Pro for work
Multicore is easy to improve by just making bigger chips with more cores if power allows, it's not the problem. Take a look at Handbrake though. The Intel chip with the same multicore performance is over three times faster. A Ryzen 7 is over 8 times faster. The ARM chip is basically useless in that workload. It's because ARM chips don't support the instructions needed and they're being emulated. A complex instruction set is broken down into simple instructions that ARM can handle, it's a very inefficient process and its performance can't be meaningfully improved. The fact ARM chips don't support complex instructions is what allows ARM to be so lightweight in the first place. So the only thing that can be done to improve this situation is for devs behind Handbrake to write their program from scratch for ARM using exclusively instructions that it natively understands so there's no emulation. This would make it less slow than it is now, but since the ARM instruction sets they would use to try to achieve the same result aren't optimal for these workloads it would still not be anywhere as fast as the x86 version.
Last edited:
Hey I've got a great analogy! Imagine ARM processors are multimedia devices that have just the MP3 decoder, and X86 processors are devices that have all multimedia decoders imaginable natively built-in. You'd be all awesome on both as long as you're playing MP3 files, maybe the ARM processor will even play these a bit faster because all optimizations go into maximizing the performance of that one format. But then you suddenly get a bunch of FLAC files that you'd like to play, and the x86 processor plays them amazingly like it's nothing. But the ARM processor is fucked, it has to spend a lot of time converting each of those FLAC files to MP3 first, and then play them as MP3 files trying to mimic the lossless FLAC format as best as it can without being really capable of doing so.
That's precisely how those processor technologies differ. In that world it'd be in Apple's interest to compare their processors to others exclusively in MP3 playback performance, and since let's say 80% of users buy their products just to listen to MP3s, it'd work just well enough. Run into any other workloads and you end up being 10 times slower, not as good at, or you can't handle them at all. Sometimes it's easy, akin to converting an AAC into MP3, just requiring more time. And sometimes it's ridiculous gymnastics like converting a video to MP3 + a synced slideshow to create an illusion of having video support that in that imaginary world it completely lacks.
Following that analogy you could add extra transistors to your ARM chips to natively support FLAC or video playback, but there are hundreds/thousands of other formats that x86 chips support natively for you to do the same thing for. Then you realize by the time you're done you'll end up with a much more bloated processor than Intel's or AMD's because these had decades to optimize each of those decoders and they are engineered to perfection at this point, having squeezed every smallest efficiency out of each of them over the years. Which is why everyone's just letting ARM be ARM, drawing the line on bare necessities and assuming it's going to be a way narrower architecture meant to do a few things well, remaining small, light and as effective as possible at the few things it's able to natively do.
I think this illustrates the issue perfectly, except replace MP3/FLAC with more complex stuff, such as "decoders" aimed to natively process specific complex algorithms (encryption, vectors etc.). There is a huge array of complex commands, each of which x86 can execute as a single operation (1 "clock cycle"). An ARM chip may be able to get you the same result, but instead of each being a single operation it has to break it down into myriads of many simple steps, each of them requiring their own clock cycle, taking many clock cycles to complete the whole thing instead.
That's precisely how those processor technologies differ. In that world it'd be in Apple's interest to compare their processors to others exclusively in MP3 playback performance, and since let's say 80% of users buy their products just to listen to MP3s, it'd work just well enough. Run into any other workloads and you end up being 10 times slower, not as good at, or you can't handle them at all. Sometimes it's easy, akin to converting an AAC into MP3, just requiring more time. And sometimes it's ridiculous gymnastics like converting a video to MP3 + a synced slideshow to create an illusion of having video support that in that imaginary world it completely lacks.
Following that analogy you could add extra transistors to your ARM chips to natively support FLAC or video playback, but there are hundreds/thousands of other formats that x86 chips support natively for you to do the same thing for. Then you realize by the time you're done you'll end up with a much more bloated processor than Intel's or AMD's because these had decades to optimize each of those decoders and they are engineered to perfection at this point, having squeezed every smallest efficiency out of each of them over the years. Which is why everyone's just letting ARM be ARM, drawing the line on bare necessities and assuming it's going to be a way narrower architecture meant to do a few things well, remaining small, light and as effective as possible at the few things it's able to natively do.
I think this illustrates the issue perfectly, except replace MP3/FLAC with more complex stuff, such as "decoders" aimed to natively process specific complex algorithms (encryption, vectors etc.). There is a huge array of complex commands, each of which x86 can execute as a single operation (1 "clock cycle"). An ARM chip may be able to get you the same result, but instead of each being a single operation it has to break it down into myriads of many simple steps, each of them requiring their own clock cycle, taking many clock cycles to complete the whole thing instead.
Last edited:
That makes sense. I've seen a ton of praise on the M1 on reddit. Of course, it was the Apple/Mac sub so I knew it'd be biased. This was another thing posted earlier today: https://appleterm.com/2020/11/28/m1-intel-tiger-lake/
This looks like the graphs you linked to above:
Are people just focused on being amazed by how the M1 performs, despite it lagging far behind the 3950? Or do they still understand that a power user will still prefer to get their job done on the most power machine instead of buying an M1 and just sitting there impressed that it runs at a fraction of the speed of what the user could be running on a more power Ryzen chip?
Donate
Any donations will be used to help pay for the site costs, and anything donated above will be donated to C-Dub's son on behalf of this community.
Members online
No members online now.