Arm Takes Aim at Laptops: Cortex-A78C Processor for PCs Introduced

[erek]

"The Cortex-A78C is a new design direction which Arm thinks is a good way for laptops, to get some more serious work done. Unlike the regular Cortex-A78 which is a heterogeneous design of big and little cores (Arm's famous big.LITTLE architecture), the new Cortex-A78C revision aims to change that. The "C" version of the processor is actually a homogeneous structure made up of all the big cores. Where you would typically use four big and four little cores for a design, Arm has decided to go all big with this design. For multithreaded purposes, this is the right decision to boost performance. The level three (L3) cache has seen a boost as well, translating to 8 MB of L3$ found on the die. We are eagerly awaiting to see the first designs based on this configuration and see how it performs."

https://www.techpowerup.com/274296/arm-takes-aim-at-laptops-cortex-a78c-processor-for-pcs-introduced

 

[Zarathustra[H]]

Hard Pass.

I want my PC's to be x86, now and forever.

 

[Lakados]

I want to see some benchmarks on it, I would love to get one and try installing Win10 IoT onto it. Most of the staff apps are now either web-based optimized for chrome, have a native arm variant, or are running on a Citrix server, give me something that looks like a MacBook Pro, or a Surface Pro, that runs an ARM chip that has a good screen, lots of ram, and 12+h battery life and I am a happy person.

 

[UnknownSouljer]

ARM Mac's are set to be announced on the 11th. I think it will be interesting to see the performance per dollar we end up seeing from ARM.
Obviously if ARM doesn't have a performance per dollar benefit as well as energy benefit, then it will have no place in the market.

 

[Lakados]

ARM Mac's are set to be announced on the 11th. I think it will be interesting to see the performance per dollar we end up seeing from ARM.
Obviously if ARM doesn't have a performance per dollar benefit as well as energy benefit, then it will have no place in the market.

I know that Amazon is pretty clear that their ARM based servers are leading in overall performance per dollar in their data centers for most applications, so I am somewhat excited to see how this tech scales down, especially Apple's implementation of it. Apple's ARM game has been on point for a number of years now, I am holding off a refresh until after the announcement, depending on how it goes determines if I try for the new stuff or last years. They are mostly used for Adobe and during my site license renewal they told me that they "were ready for the new lineup", so should be a good show regardless.

 

[Mega6]

ARM will always be the high efficency lower workload choice. X86 will continue to dominate the high performance, complex workloads. Each have their niche.

 

[LukeTbk]

ARM will always be the high efficency lower workload choice. X86 will continue to dominate the high performance, complex workloads. Each have their niche.

I feel there is a question mark there as well, depending on how much money there is in the complex workloads (to make worth it hardware made just for it in mind) and I imagine that where the fear come from.

Going from a very general computing world to one where the big player able to have custom made just for them hardware completely destroy any small developers/products that cannot jump into that game. Because I can see a world where for a specific workloads it is impossible to beat in performance custom hardware for it and that arm is the way to go for that.

 

[Mega6]

Only giant companies with huge budgets can afford custom ASIC chip design (Yes, incuding using ARM as a base - see Apple, Amazon, ect.) along with developing a custom code base to take advantage of the custom design. Your generic template ARM just isn't enough to displace the general compute X64 in most cases.

 

[LukeTbk]

Only giant companies with huge budgets can afford custom ASIC chip design (Yes, incuding using ARM as a base - see Apple, Amazon, ect.) along with developing a custom code base to take advantage of the custom design. Your generic template ARM just isn't enough to displace the general compute X64 in most cases.

Yes (at least for the next 50 year's or what not before AI can make good one with little help and the compiler that take advantage), that why it can be a bit scary trying to compete with an Adobe/Apple partnership that run on custom hardware with generic X86/64 could be an impossible task, if the API are all open at least maybe, but who knowns. Because if there is a very good amount of money, we could see them more and more do it, even if the workload is complex.

 

[Bowman15]

But can it run Crysis...

 

[Mega6]

Yes (at least for the next 50 year's or what not before AI can make good one with little help and the compiler that take advantage), that why it can be a bit scary trying to compete with an Adobe/Apple partnership that run on custom hardware with generic X86/64 could be an impossible task, if the API are all open at least maybe, but who knowns. Because if there is a very good amount of money, we could see them more and more do it, even if the workload is complex.

What if a completely new architecture pops up designed by new AI that is more efficent than ARM or X86. Maybe, who knows.

OR - maybe the world ends tomorrow.

 

[Deleted member 89018]

Hard Pass.

I want my PC's to be x86, now and forever.

i am curious as to why it would matter what instruction set your PC uses?

 

[Zarathustra[H]]

i am curious as to why it would matter what instruction set your PC uses?

I have a library of software and games dating back to the 80's, most of which I doubt would ever be ported.

That, and despite Apple's great work on making ARM chips more capable, I don't have much confidence in them in high performance applications.

If I am going to be convinced to change ISA, it better be for a royalty free, open source ISA.

 

[Lakados]

I have a library of software and games dating back to the 80's, most of which I doubt would ever be ported.

That, and despite Apple's great work on making ARM chips more capable, I don't have much confidence in them in high performance applications.

If I am going to be convinced to change ISA, it better be for a royalty free, open source ISA.

I am sure at this point much of that older software could easily be run in an x86 emulation layer with no perceivable drop in performance.

 

[ChadD]

ARM has a handful of major advantages over x86.... and with Nvidia now in that picture are few of them may disappear anyway.

Biggest draw back in regards to x86... is Intel AMD.... and a couple chinese players via Vias license are the only possible players ever.

ARM as long as Nvidia doesn't do anything incredibly stupid will always be alive with competition. Apple is free to take the ISA and lock the gates. However with MS seeming to be onboard ARM Linux and Windows hardware is wide open. Samsung Amazon or any other future player that wants to design a ARM chip can.

Yes today x86 is for sure the desktop performance king... and that isn't likely to change in regards to pure CPU performance. However we are also getting to a time where co processors are going to start to really matter. Apple is going to be talking about all the things bolted on to their ARM chips that are not ARM cores. So far x86 isn't that agile.

Apples announcements will be interesting. These newer generic ARM core designs are interesting if they can compete with what Apple is doing. Windows ARM laptops next year might actually not suck. Having 8 big ARM cores instead of a big.Little setup is interesting. We will finally see what real performance first ARM chips look like in the consumer space. They won't be in workstations and gaming machines anytime soon.... but almost everything under that computer wise can probably be better or at least more economically accomplished with ARM.

 

[ChadD]

I am sure at this point much of that older software could easily be run in an x86 emulation layer with no perceivable drop in performance.

Chances are he already runs 100% of that 1980s software in an emulator anyway. DOSbox for the win. lol

 

[Lakados]

Chances are he already runs 100% of that 1980s software in an emulator anyway. DOSbox for the win. lol

Oh yeah.... I forgot there was a DOSBox port for the Pi4...

 

[UnknownSouljer]

I have a library of software and games dating back to the 80's, most of which I doubt would ever be ported.

Cool? Do you still hang onto SGI and Motorolla based machines too? I think there's more benefit to virtual machines and emulation for that stuff in the future.

That, and despite Apple's great work on making ARM chips more capable, I don't have much confidence in them in high performance applications.

We'll see won't we? We know Apple will have to move Mac Pro to ARM (there are already rumors circling around what that will be) and it will be possible to see how the new ARM Mac Pro compares to the old one. That will be a two year wait though - 2022 at the soonest.

If I am going to be convinced to change ISA, it better be for a royalty free, open source ISA.

I think that's a very short sighted incredibly limited way to look at things. So if ARM's performance is literally 20x greater performing in all tasks and it costs 20% as much as any competing architecture and it has 100% of all current work software on it, you'd still not move to it? So in other words you'd pay 80% more for 5% of the performance? Or in dollar terms if an ARM chip costed $500 dollars and performed 20x better than a $10,000 x86 chip, you'd rather spend $10,000?
Obviously I'm speaking in hyperbolic terms, but at what point is the cost versus performance "good enough" for you to switch to? How great of a software and performance gap does there have to be? If you're immovable even if literally 99% of the general computing work force isn't on x86 anymore I guess that's on you. At that point you'll be Motorola powered SGI or Mac powered PowerPC.

As far as being open source; I don't think that's something any of us can control. The bottom line is the platform we all end up on is likely the one that is the most democratized. If you work in a corporate office now you're on x86 and Windows basically >90% of the time. If in that same situation it switches to ARM (in some hypothetical situation where >90% of corporate is on ARM and is using whatever OS), there won't be really much of a choice. Either use an esoteric system that no one else is using and no one knows how to use or be forced to move in order to stay relevant. At the end of the day, you as a user can choose to do whatever you want. Including spend $10,000 on x86 CPU's that don't compete and use ISA's that no one else is on, but again I don't think there is nearly as many choices in the matter as you want to believe there are - that is if again you want to be relevant.

 

[Deleted member 243478]

Hard Pass.

I want my PC's to be x86, now and forever.

Good luck with that...

 

[Zarathustra[H]]

Good luck with that...

If you think x86 is going anywhere in the next 20 years you have completely lost your mind.

Sure, Apple will use ARM and ARM will make introads inroads for ultrabooks, but anything further than that seems farfetched.

 

[Zarathustra[H]]

I think that's a very short sighted incredibly limited way to look at things. So if ARM's performance is literally 20x greater performing in all tasks and it costs 20% as much as any competing architecture and it has 100% of all current work software on it, you'd still not move to it? So in other words you'd pay 80% more for 5% of the performance? Or in dollar terms if an ARM chip costed $500 dollars and performed 20x better than a $10,000 x86 chip, you'd rather spend $10,000?

These are fantasy scenarios.

Won't happen. Not now, not ever.

 

[LukeTbk]

If you think x86 is going anywhere in the next 20 years you have completely lost your mind.

Sure, Apple will use ARM and ARM will make introads inroads for ultrabooks, but anything further than that seems farfetched.

Just NVidia, Microsoft Apple and Amazon using arm sound like already a lot (that possibly the 2 biggest cloud company I think), those are really big giants.

x86 will have a giant momentum/legacy to overturn and will make it stay relevant a long time, but it could become a minorityshare in some space where it is currently popular by 2040 like personal computer and gaming console (and obviously laptop/mobile personal computing where it is almost a given to happen for the section where it is yet to be the case like laptops).

 

[UnknownSouljer]

These are fantasy scenarios.

Won't happen. Not now, not ever.

You're awfully sure of yourself and being incredibly dismissive considering that SGI and Motorola, as a few examples, had chips at the top at one point and now are irrelevant. If you wanted to buy something containing any of their ISA's and instruction sets at this point you can't. You could say it's even greater than a 200% speed difference for 20% the cost, as there hasn't been new development on any of them in over a decade.

The top Supercomputer is ARM. ARM is being used increasingly in Server Workloads. And Apple is showing that it's more than competitive for general compute. I'd say you lack vision if anything. EDIT: As LukeTbk above noted, it's being invested in by nVidia, Qualcomm, Samsung, Apple, Amazon, and Fujitsu amongst others.

Look. Let's take ARM out of the equasion for a second. Let's say ARM never makes it like you say. To believe that in the next 40 years of your life there won't be an ISA that is faster/better/cheaper than x86 in that time is incredibly short sited. All of modern computing is arguably less than 50 years - x86 is just a fraction of that. The processing power in the Apollo 11 is less than a calculator at this point.
Whether you think it's ARM or "Garboflavens" isn't really relevant. And I think you missed the point of my statement by focusing on ARM. I personally think it will be ARM that takes over but that belief isn't relevant to the points I made.

 

[ChadD]

Oh yeah.... I forgot there was a DOSBox port for the Pi4...

Yep even under Newer versions of windows.... all that old software is already running in emulation. Translating a few instruction calls for software designed to run on 30-40 year old x86 chips is no big deal. Dosbox can be compiled for Android and things run just as well.... have to artificially slow down the CPU to properly emulate DOS software even on ARM. lol

 

[GiGaBiTe]

i am curious as to why it would matter what instruction set your PC uses?

It's more than instruction sets, disposability is a massive problem and upgradability is non-existent in the ARM world. There is not, and has never been a standardized socket or platform for ARM CPUs that large numbers of companies or devices support. When you buy an ARM device, its features are frozen and non-upgradeable, meaning that when it gets to be slow, it gets tossed for something faster.

In x86 land for the past 40 years, the majority of x86 CPUs have been in sockets, which allow you to swap in faster parts and keep your computer relevant longer. ARM is also 100% incompatible with all existing x86 slotted cards that rely on option ROMs, unless a new ROM is written for the device, which will never happen. Drivers also have to be written specifically for ARM, which tosses out pretty much everything else. Then there's the whole problem of Microsoft locking down the ARM platform to make it difficult or impossible to install other OSes like Linux on them because of their backroom dealings with hardware vendors. Sort of like how they tried to take over UEFI, and still hold the keys to the lock when it comes to secure boot.

Going from x86 to ARM is backwards, promotes disposability and gives users less choice over their systems. Who really cares whether ARM is more efficient, it's just more BS "in the name of progress" that benefits nobody but people holding the cards.

I am sure at this point much of that older software could easily be run in an x86 emulation layer with no perceivable drop in performance.

Emulation is never perfect, there are a lot of older games that don't work properly or at all on emulators, for any architecture. The same can be said for programs as well, I have several applications and server programs that run badly on virtual machines because of timing issues. You can't blame the programmer back then for it either, because nobody had a notion that x86 was going to eventually be virtualized.

And one of the most annoying are programs that have some sort of licensing or hardware dongles, those can be impossible to emulate and would require someone with assembly knowledge to patch them.

 

[GNUse_the_force]

Since Nvidia bought ARM im much more interested in RISC-V for an opensource platform. I expect within the next 5 years RISC-V is going to make huge strides. It won't replace (and doesn't need to) x86 but it will be a compelling alternative for those who want PC/tinker/IOT units to do 90% of their regular tasks at minimal power usage on a secure platform. The increasingly thirsty & expensive regular PC can be for everything else.

That said i don't think x86 as an open platform is totally out of the question as afaik AMD is opensourcing their coreboot system. They don't appear to have made any moves to do that to their current PSP 'blackbox' system yet but time will tell. At the very least if you are concerned about freedom/security/privacy currently i would put my money towards building an AMD system to support their more open business practices. There is a real possibility that AMD release a small form factor competitor.


edit*: Pricing is currently in the realm of custom/business, but RISC-V platforms are surprisingly way past the drawing board phase. Maybe in another 5 years the pricing/performance will be 'good enough' ..

https://www.phoronix.com/scan.php?page=article&item=sifive-riscv-unmatched&num=1

 

[ChadD]

It's more than instruction sets, disposability is a massive problem and upgradability is non-existent in the ARM world. There is not, and has never been a standardized socket or platform for ARM CPUs that large numbers of companies or devices support. When you buy an ARM device, its features are frozen and non-upgradeable, meaning that when it gets to be slow, it gets tossed for something faster.

In x86 land for the past 40 years, the majority of x86 CPUs have been in sockets, which allow you to swap in faster parts and keep your computer relevant longer. ARM is also 100% incompatible with all existing x86 slotted cards that rely on option ROMs, unless a new ROM is written for the device, which will never happen. Drivers also have to be written specifically for ARM, which tosses out pretty much everything else. Then there's the whole problem of Microsoft locking down the ARM platform to make it difficult or impossible to install other OSes like Linux on them because of their backroom dealings with hardware vendors. Sort of like how they tried to take over UEFI, and still hold the keys to the lock when it comes to secure boot.

Going from x86 to ARM is backwards, promotes disposability and gives users less choice over their systems. Who really cares whether ARM is more efficient, it's just more BS "in the name of progress" that benefits nobody but people holding the cards.



Emulation is never perfect, there are a lot of older games that don't work properly or at all on emulators, for any architecture. The same can be said for programs as well, I have several applications and server programs that run badly on virtual machines because of timing issues. You can't blame the programmer back then for it either, because nobody had a notion that x86 was going to eventually be virtualized.

And one of the most annoying are programs that have some sort of licensing or hardware dongles, those can be impossible to emulate and would require someone with assembly knowledge to patch them.

Well if Intel had its way.... x86 would basically be upgradable anyway. The idea is nice... one socket pop in a new chip. But lets all get real it hasn't worked that way in YEARS. It was a legit argument in 386 486 days... when you could slot CPUs from at least 3 different manufacturers into the same slot. (and same board never mind) However Intel ENDED that years ago. Intel chips do not slot into AMD boards and vise versa.... and Intel doesn't use the same socket generation to generation. AMD has been applauded for sticking with the AM4 socket for as long as they have... however AM4 is probably going to be the outlier where a socket lasts for more then 2 generations. (and lets also be real... even AM4 is not really 3 generations. First generation AM4 boards DO NOT support Zen 3, so AM4 is a 2 generation socket. 450/70 boards will not all support Zen3, and support is delayed)

x86 has no advantage as far as electronic scrap pile creation goes.

As for emulation of x86 stuff... really what we are talking about here is Dos/Windows emulation. There is nothing x86/ARM about a USB dongle. We are talking about software not hardware. As far as software goes... MS does a good enough job making sure things like security dongles stop working. Good luck getting any security dongle setup designed for windows in the 90s to work on a modern Windows x86 PC. Everything you run that is pre windows 7 is 100% running in a emulator even on x86.... and old software that uses real close to the metal assembler bits isn't going to run properly on a modern x86 chip any more then it would on ARM MIPS or riscv.

Having said that x86 software is some of the easiest software to emulate as its functions are well documented and easy to mimic. Where emulation normally falls short is older and in general super proprietary architectures like old consoles ect that used specific customized chips. In a lot of cases those companies either no longer exist or if they do they never really kept good records on how that stuff even worked. All the old tricks used are long forgotten and the people that wrote the software long gone.

I expect Apples switch will be actually very smooth. They have such a tight control on the inner working software of their OS that I can't imagine they will have any problem properly swapping.

 

[juanrga]

ARM will always be the high efficency lower workload choice. X86 will continue to dominate the high performance, complex workloads. Each have their niche.

The fastest computer in the world is ARM and it is running the most complex workloads available.

If you think x86 is going anywhere in the next 20 years you have completely lost your mind.

People said the same about PA-RISC or PowerPC.

It's more than instruction sets, disposability is a massive problem and upgradability is non-existent in the ARM world. There is not, and has never been a standardized socket or platform for ARM CPUs that large numbers of companies or devices support. When you buy an ARM device, its features are frozen and non-upgradeable, meaning that when it gets to be slow, it gets tossed for something faster.

Sockets for ARM platform have existed for a while. And RAM, disks, GPUs... are upgradeable.

 

[GiGaBiTe]

Well if Intel had its way.... x86 would basically be upgradable anyway. The idea is nice... one socket pop in a new chip. But lets all get real it hasn't worked that way in YEARS. It was a legit argument in 386 486 days... when you could slot CPUs from at least 3 different manufacturers into the same slot. (and same board never mind) However Intel ENDED that years ago. Intel chips do not slot into AMD boards and vise versa.... and Intel doesn't use the same socket generation to generation. AMD has been applauded for sticking with the AM4 socket for as long as they have... however AM4 is probably going to be the outlier where a socket lasts for more then 2 generations. (and lets also be real... even AM4 is not really 3 generations. First generation AM4 boards DO NOT support Zen 3, so AM4 is a 2 generation socket. 450/70 boards will not all support Zen3, and support is delayed)

Market segmentation doesn't make a socket any less valuable. Intel's artificial market segmentation since Sky Lake has been proven with weird Chinese motherboards that accept 6th to 9th gen Intel CPUs, and by people modifying their motherboards sockets and BIOSes to accept newer Intel chips with no issues.

AMD's CPU support issue is almost entirely caused by board vendors eating up all of the ROM space with dumb flashy graphics, leaving no room for microcode. I've seen a few motherboards have special UEFI updates with simplified GUIs to add support more chips. Fancy UEFI setups should have never been a thing, we got along fine for 30+ years with text menus.

If 5000 series Zen chips launch on AM4, it'd make four generations of supported processors. The final Bulldozer CPUs (Bristol Ridge), and 1st-3rd gen Zen. While chipset support being an issue that prevents older working with newer, it was an issue on past sockets as well, so it's not really a new problem. Socket 7 and Super 7 had numerous chipsets all with varying compatibility, as did Slot 1, PGA370, 478 and 775. AMD should have not made the promises they did with all of their eggs in one basket and all working at the same time, but they at least partially delivered with keeping the same socket.

x86 has no advantage as far as electronic scrap pile creation goes.

I can't tell you the number of x86 systems I've saved from the scrap pile by just upgrading the CPU and RAM. Can't say the same for cheap crappy ARM devices that eventually become too slow to run modern software. I still know quite a few people with 10+ year old x86 systems that get along just fine after some upgrades. ARM? Not so much.

Having said that x86 software is some of the easiest software to emulate as its functions are well documented and easy to mimic. Where emulation normally falls short is older and in general super proprietary architectures like old consoles ect that used specific customized chips. In a lot of cases those companies either no longer exist or if they do they never really kept good records on how that stuff even worked. All the old tricks used are long forgotten and the people that wrote the software long gone.

You're forgetting that x86 had plenty of specialized chips which are difficult to emulate. MDA, CGA, EGA and VGA. VESA modes, sound cards, input devices and more. And even though the x86 architecture was extensively documented, timing is a big problem. Many old programs used function loops timed to old CPU speeds that would race if ran on faster CPUs, something which is an annoyance in emulation. Dosbox can to an extent slow itself down to help mitigate these issues, but it doesn't work in all cases.

 

[ChadD]

You're forgetting that x86 had plenty of specialized chips which are difficult to emulate. MDA, CGA, EGA and VGA. VESA modes, sound cards, input devices and more. And even though the x86 architecture was extensively documented, timing is a big problem. Many old programs used function loops timed to old CPU speeds that would race if ran on faster CPUs, something which is an annoyance in emulation. Dosbox can to an extent slow itself down to help mitigate these issues, but it doesn't work in all cases.

My point was that software doesn't run on newer x86 chips properly either. It all runs either via software like DOSbox or the emulator MS builds into windows for early windows software. It all needs to be emulated these days... so it really doesn't matter what modern ISA its running on. Dosbox compiled for Android and ARM runs the same as Dosbox compiled for Windows 10. Its not the hardware that is the issue... its the knowledge of the software emulator folks. Stuff that was well documented and not tight closed source proprietary runs fine via emulation. The stuff that doesn't, doesn't run properly on Win/x86 either.

 

[Joseph F]

Hard Pass.

I want my PC's to be x86, now and forever.

I wouldn't say forever, but for the forseeable future, yes.

 

[schmide]

Although a good step still a total nothing. I'm getting really tired of ARM getting a pass for future performance without producing a competitive platform in anything but edge cases.

Additionally, the constant comparison of normalized numbers to infer future performance. You can't do that so doing so just makes things more confusing. A current process 2ghz chip with a 2clock L1 will never run at 4ghz.

I will buy it if they produce it, but lets stop pretending that a product exists in this segment. Maybe Apple will surprise us soon, but even then, it will just be a baby step.

It is kind of neat to see intel moving towards big.little while ARM makes a move towards only big.

 

[Sycraft]

My point was that software doesn't run on newer x86 chips properly either. It all runs either via software like DOSbox or the emulator MS builds into windows for early windows software. It all needs to be emulated these days... so it really doesn't matter what modern ISA its running on. Dosbox compiled for Android and ARM runs the same as Dosbox compiled for Windows 10. Its not the hardware that is the issue... its the knowledge of the software emulator folks. Stuff that was well documented and not tight closed source proprietary runs fine via emulation. The stuff that doesn't, doesn't run properly on Win/x86 either.

Not exactly. The reason that modern virtualization is so far is because it is virtualization, not emulation. Most of the code runs natively on the chip. A program running in ring 3 in a VM is just passed through and run in ring 3 on the host. Only system calls and such have to be dealt with special (and there is hardware acceleration for that). That ends up adding very little overhead and allows for basically native performance in VMs. You don't get that if you have to emulate though. Then you have to translate all instructions from the guest ISA to the host ISA. There are lots of ways to accelerate that process, but it is never as fast as just a passthrough.

 

[ChadD]

Not exactly. The reason that modern virtualization is so far is because it is virtualization, not emulation. Most of the code runs natively on the chip. A program running in ring 3 in a VM is just passed through and run in ring 3 on the host. Only system calls and such have to be dealt with special (and there is hardware acceleration for that). That ends up adding very little overhead and allows for basically native performance in VMs. You don't get that if you have to emulate though. Then you have to translate all instructions from the guest ISA to the host ISA. There are lots of ways to accelerate that process, but it is never as fast as just a passthrough.

That is a good point.... depends on the age of the software we are talking about. I don't think a lot of people are virtualizing extremely old operating systems to run some old dos or windows 98 program. If its windows 7 age software most people just run that stuff in windows 10... and if it doesn't run properly sure you can run a VM. I'm not sure that not a massive part of the market. People do it but I'm just not sure its all that common. Software newer then that... I'm not sure why anyone would need to virtualize it if the software developers simply recompile. The industry has been moving all their frameworks and various tools toward cross compatibility for more then a few years now.

How things go for Apple over then next 6 months or so will be interesting. Guess we'll see if the software industry at least on the Apple side has gotten to a point where cross compiling is relatively simple. My prediction... it either goes super smooth and Apple end users notice no issues, OR it will be a complete cluster fuck. I don't foresee a situation where 80% works well and 20% doesn't.... its either going to be half your Apple stuff runs like shit or 95% of it runs like native. After that the next generation of Windows ARM machines is a completely different ball of wax of course. MS has nothing like the ecosystem lock Apple does.... however IF (and only if) it goes super smooth for Apple, I could imagine in a year or so when new ARM machines hit with A78 8 Big core ARM chips hit Windows devices MS may stand a chance. If Apple is a cluster.... ya then Windows will be x86 till windows dies. lol

 

[Sycraft]

Well it'll also depend on what ARM can offer. If their chip isn't significantly faster, or more efficient or cheaper then it is likely to end up being not that popular. While many ARM fans take it as an article of faith that ARM can make a chip so much better than x86, this has yet to be proven. The idea that "It is RISC so it will be better," has long since been disproven by the past 30 years of RISC (Power, MIPS, etc). The ISA just doesn't allow for magically faster chips. So I remain a skeptic until ARM produces something and we get some real benchmarks to see how it performs and we see what the street price will be.

Remember that Apple isn't switching to ARM because it is so much better, they are switching to ARM because they want to control everything. They've long had Not Invented Here syndrome really bad, and they want to work on unifying their MacOS and iOS product line. So while this doesn't mean that their ARM chips will be bad, don't think they are doing it because they are going to get some major gains. They are doing it for control and to have their own thing, any improvements are secondary.

With Windows (and even Linux) the trick isn't for ARM to just be competitive, it has to be better and by enough of a margin to make it worth switching. Otherwise the view of most is going to be "Why bother?" Once something gets entrenched as a standard, dislodging that standard takes more than just competing, you have to be not just a little better but better enough that people, companies, and so on say "Ok, this is worth changing." Otherwise they just stick with what they already have because it is "Good enough."

 

[Red Falcon]

If you think x86 is going anywhere in the next 20 years you have completely lost your mind.

Sure, Apple will use ARM and ARM will make introads inroads for ultrabooks, but anything further than that seems farfetched.

If it weren't for AMD, I was estimating x86-64 would be dead in less than a decade.
AMD is making competition for ARM on nearly all fronts except mobile (x86-64 is garbage here), and is the only company keeping that legacy ISA alive at this point.

People said the same thing about m68k, PA-RISC, MIPS, POWER/PowerPC, SPARC etc. back in the 1980s and 1990s - yet here we are.
SPARC64 and Power ISA is still alive and well in servers, but they are non-existent outside of those markets and ecosystems.

 

[Deleted member 243478]

Although a good step still a total nothing. I'm getting really tired of ARM getting a pass for future performance without producing a competitive platform in anything but edge cases.

Additionally, the constant comparison of normalized numbers to infer future performance. You can't do that so doing so just makes things more confusing. A current process 2ghz chip with a 2clock L1 will never run at 4ghz.

I will buy it if they produce it, but lets stop pretending that a product exists in this segment. Maybe Apple will surprise us soon, but even then, it will just be a baby step.

It is kind of neat to see intel moving towards big.little while ARM makes a move towards only big.

LOL so the iPhone and iPad lines are edge cases? In 10 years Apple has taken their SOC from a garbage mobile chip to a chip that can rival Intel i7’s with a tiny power power envelope / faster internal graphics, onboard encryption, secure enclave and much more. Its hilarious that Apple had to put their own SOC in Macs under the T2 branding just to give their Intel powered devices the same features with security because x86 couldn’t deliver on it period...

This week will be great to see the new Mac line ups and what their SOC can do when it is scaled up and given a higher power envelope!

 

[LukeTbk]

Although a good step still a total nothing. I'm getting really tired of ARM getting a pass for future performance without producing a competitive platform in anything but edge cases.

Are amazon graviton 2 server service really an edge case (or non competitive) ? Considering the marketshare of 60% and how much of the world run on that market.

 

[Deleted member 243478]

Well it'll also depend on what ARM can offer. If their chip isn't significantly faster, or more efficient or cheaper then it is likely to end up being not that popular. While many ARM fans take it as an article of faith that ARM can make a chip so much better than x86, this has yet to be proven. The idea that "It is RISC so it will be better," has long since been disproven by the past 30 years of RISC (Power, MIPS, etc). The ISA just doesn't allow for magically faster chips. So I remain a skeptic until ARM produces something and we get some real benchmarks to see how it performs and we see what the street price will be.

Remember that Apple isn't switching to ARM because it is so much better, they are switching to ARM because they want to control everything. They've long had Not Invented Here syndrome really bad, and they want to work on unifying their MacOS and iOS product line. So while this doesn't mean that their ARM chips will be bad, don't think they are doing it because they are going to get some major gains. They are doing it for control and to have their own thing, any improvements are secondary.

With Windows (and even Linux) the trick isn't for ARM to just be competitive, it has to be better and by enough of a margin to make it worth switching. Otherwise the view of most is going to be "Why bother?" Once something gets entrenched as a standard, dislodging that standard takes more than just competing, you have to be not just a little better but better enough that people, companies, and so on say "Ok, this is worth changing." Otherwise they just stick with what they already have because it is "Good enough."

No they are switching to ARM because they can match Intels performance with 1/4 the power consumption while also delivering more secure devices without trying to marry x86 to their own SOC. All the while increasing profits and controlling their own destiny instead of being hindered by Intel missing their targets...

As for companies moving to AMD... AMD have still proven time and time again that their platforms are not mature enough or validated enough. They will happily push products to market that aren’t ready and finish them later at the users expense. Even Ryzen 3000 series failed constantly to deliver its advertised specs.

 

[schmide]

LOL so the iPhone and iPad lines are edge cases? In 10 years Apple has taken their SOC from a garbage mobile chip to a chip that can rival Intel i7’s with a tiny pass power envelope / have Faster internal graphics, onboard encryption, secure enclave and much more. Its hilarious that Apple had to put their own SOC in Macs under the T2 branding just to give their Intel powered devices the same features with security because x86 couldn’t deliver on it period...

This week will be great to see the new Mac line ups and what their SOC can do when it is scaled up and given a higher power envelope!

Their systems are good, but a lowish edge.

Even more the systems are propitiatory. No custom drives or accelerators. There is a fair amount of compromise one has to go through to accept it as a comparable system to what we are all used to.

On the other side we have high edge like Graviton (which sadly exists in a bubble) The Thunder X3 is canceled so we have one less source of high metrics.

Niche markets do not automatically lead to market takeover.