Windows 10 machines running on ARM will be able to emulate x64 apps soon

erek

Supreme [H]ardness
Joined
Dec 19, 2005
Messages
6,988
"While Windows on ARM has been a relatively slow mover, that hasn't stopped competitors from going forward with ARM plans. Apple is expected to launch the first ARM-based Mac later this year.

macOS already completely dropped support for 32-bit applications somewhat recently, and Apple will offer Rosetta 2 to emulate 64-bit macOS apps on ARM Macs (which the company calls "Macs with Apple Silicon").

However, whether (and how) Mac users will be able to virtualize Windows x86 applications on Apple Silicon Macs remains unknown. x64 emulation will first be introduced to ARM Windows machines via the Windows Insider Program next month."


https://arstechnica.com/gadgets/202...on-arm-will-be-able-to-emulate-x64-apps-soon/
 

deruberhanyok

[H]ard|Gawd
Joined
Aug 22, 2004
Messages
1,758
This is a necessary step for the whole industry to transition away from x86. Or at least to make ARM a viable option in windows space. I imagine Intel isn’t too thrilled about it.

I’m wondering whether this is the kind of thing that will just add some windows “legacy compatibility bloat” though, which it seemed for a little while wouldn’t be a thing on Windows ARM. I didn’t realize they’d already included 32-bit compatibility.

Hopefully they’re not planning to put full on WoW32/64 in there and allow ARM devs to use comdlgs from windows 95. /snark

If Microsoft plans to introduce it but never actually deprecate it, it just further extends the lifetime of software that should be considered for retire / replace / overhaul. I’ve worked with a lot of companies still running windows 2000 / server 2003 era software that haven’t wanted to upgrade because “it still works.”

And I get that, I really do, but I’m usually working with those companies because something stopped working, and you can only hold on to that cash for so long before you need to bite the bulletnand upgrade. there’s a point where holding off goes from “saving money” to “massive liability that could sink your business” and my main concern with x86 emulation on ARM is that we’ll just see that line continue to be blurred.
 

Lakados

2[H]4U
Joined
Feb 3, 2014
Messages
3,025
This is a necessary step for the whole industry to transition away from x86. Or at least to make ARM a viable option in windows space. I imagine Intel isn’t too thrilled about it.

I’m wondering whether this is the kind of thing that will just add some windows “legacy compatibility bloat” though, which it seemed for a little while wouldn’t be a thing on Windows ARM. I didn’t realize they’d already included 32-bit compatibility.

Hopefully they’re not planning to put full on WoW32/64 in there and allow ARM devs to use comdlgs from windows 95. /snark

If Microsoft plans to introduce it but never actually deprecate it, it just further extends the lifetime of software that should be considered for retire / replace / overhaul. I’ve worked with a lot of companies still running windows 2000 / server 2003 era software that haven’t wanted to upgrade because “it still works.”

And I get that, I really do, but I’m usually working with those companies because something stopped working, and you can only hold on to that cash for so long before you need to bite the bulletnand upgrade. there’s a point where holding off goes from “saving money” to “massive liability that could sink your business” and my main concern with x86 emulation on ARM is that we’ll just see that line continue to be blurred.
My NT4 server should be gone by Halloween... :) finally convinced accounting to sign off on their upgrade (it runs the accounting software) 6 months in and still on budget. Converting the DB over to SQL, upgrading the version of the software from v2, to like v14, and bring it all up to server 2019. This also gets rid of the need to keep IE7 running as that was the only browser that worked with it.
 

Aurelius

2[H]4U
Joined
Mar 22, 2003
Messages
3,006
This is a necessary step for the whole industry to transition away from x86. Or at least to make ARM a viable option in windows space. I imagine Intel isn’t too thrilled about it.

I’m wondering whether this is the kind of thing that will just add some windows “legacy compatibility bloat” though, which it seemed for a little while wouldn’t be a thing on Windows ARM. I didn’t realize they’d already included 32-bit compatibility.

Hopefully they’re not planning to put full on WoW32/64 in there and allow ARM devs to use comdlgs from windows 95. /snark

If Microsoft plans to introduce it but never actually deprecate it, it just further extends the lifetime of software that should be considered for retire / replace / overhaul. I’ve worked with a lot of companies still running windows 2000 / server 2003 era software that haven’t wanted to upgrade because “it still works.”

And I get that, I really do, but I’m usually working with those companies because something stopped working, and you can only hold on to that cash for so long before you need to bite the bulletnand upgrade. there’s a point where holding off goes from “saving money” to “massive liability that could sink your business” and my main concern with x86 emulation on ARM is that we’ll just see that line continue to be blurred.

I've long held that one of Microsoft's greatest liabilities is its longstanding "legacy support above all else" mantra. That's been changing in recent years, but it seems like that past emphasis on legacy support could really come to bite the company if there's a broader ARM transition. Support for ARM has been poor, and companies/developers might be reluctant to adopt it regardless because there's some x86 app they'd like to run for all eternity.

Apple serves as the obvious contrast. It expects the entire Mac lineup to finish — finish! — transitioning to ARM in two years. It'll be easy to create ARM binaries, and most apps (if not the vast majority) should "just work" under emulation until they have ARM-native versions. In other words, Apple will likely have made a wholesale switch to ARM by 2022 while Microsoft might have trouble getting even a small percentage of Windows users to buy ARM computers by that point.

I don't expect Apple to suddenly be all-conquering in computers, but it could be embarrassing for Microsoft (and Intel, and AMD) if even the lowliest MacBook becomes faster and longer-lasting than most high-end Windows laptops.
 

4saken

[H]F Junkie
Joined
Sep 14, 2004
Messages
11,328
I've long held that one of Microsoft's greatest liabilities is its longstanding "legacy support above all else" mantra. That's been changing in recent years, but it seems like that past emphasis on legacy support could really come to bite the company if there's a broader ARM transition. Support for ARM has been poor, and companies/developers might be reluctant to adopt it regardless because there's some x86 app they'd like to run for all eternity.

Apple serves as the obvious contrast. It expects the entire Mac lineup to finish — finish! — transitioning to ARM in two years. It'll be easy to create ARM binaries, and most apps (if not the vast majority) should "just work" under emulation until they have ARM-native versions. In other words, Apple will likely have made a wholesale switch to ARM by 2022 while Microsoft might have trouble getting even a small percentage of Windows users to buy ARM computers by that point.

I don't expect Apple to suddenly be all-conquering in computers, but it could be embarrassing for Microsoft (and Intel, and AMD) if even the lowliest MacBook becomes faster and longer-lasting than most high-end Windows laptops.


pretty sure they aren't worried. Well maybe AMD.
 

Lakados

2[H]4U
Joined
Feb 3, 2014
Messages
3,025
I've long held that one of Microsoft's greatest liabilities is its longstanding "legacy support above all else" mantra. That's been changing in recent years, but it seems like that past emphasis on legacy support could really come to bite the company if there's a broader ARM transition. Support for ARM has been poor, and companies/developers might be reluctant to adopt it regardless because there's some x86 app they'd like to run for all eternity.

Apple serves as the obvious contrast. It expects the entire Mac lineup to finish — finish! — transitioning to ARM in two years. It'll be easy to create ARM binaries, and most apps (if not the vast majority) should "just work" under emulation until they have ARM-native versions. In other words, Apple will likely have made a wholesale switch to ARM by 2022 while Microsoft might have trouble getting even a small percentage of Windows users to buy ARM computers by that point.

I don't expect Apple to suddenly be all-conquering in computers, but it could be embarrassing for Microsoft (and Intel, and AMD) if even the lowliest MacBook becomes faster and longer-lasting than most high-end Windows laptops.
Microsoft is big enough that they could support two ecosystems, most developers wont notice a difference, those that would be impacted generally have customers who know enough to be purchasing the correct hardware. It's not like ARM coming along means x86 hardware goes away, besides Apple is quite literally shouldering the costs on this, so by the time Microsoft gets around to it Apple will have mainstreamed it. What Microsoft needs to do is develop their own Citrix like environment and pair that with Azure to bring application virtualization to a more affordable point for the small and medium business and blam suddenly arm based laptops and surface pro devices has a business case they can get behind.
 

Aurelius

2[H]4U
Joined
Mar 22, 2003
Messages
3,006
pretty sure they aren't worried. Well maybe AMD.

If it it isn't, it probably should be.

Remember, Microsoft missed the boat on mobile in part because it thought there was no reason to worry -- surely Windows phones would continue to thrive because Windows PCs do, right? No, Apple probably won't dominate computers, but the balance could shift if it gets to the point where Macs are routinely outperforming Windows machines.

Having followed tech for long enough, I've noticed that the companies that persist in the modern era are the ones worrying about what's next even when it looks like they're secure. The market tends to shift in such a way that it's fine... until it suddenly isn't.
 

westrock2000

[H]F Junkie
Joined
Jun 3, 2005
Messages
9,215
I'm thinking Intel is suddenly going to get it's act together and make some big moves. Right now they are repeating Pentium 4 and RAMBUS days.
 

Red Falcon

[H]F Junkie
Joined
May 7, 2007
Messages
10,729
I'm thinking Intel is suddenly going to get it's act together and make some big moves.
We've been hearing that for nearly half a decade now, and I will believe it when I see it.
Intel stagnated the market for long enough, and competition is good.
 

ManofGod

[H]F Junkie
Joined
Oct 4, 2007
Messages
12,071
If it it isn't, it probably should be.

Remember, Microsoft missed the boat on mobile in part because it thought there was no reason to worry -- surely Windows phones would continue to thrive because Windows PCs do, right? No, Apple probably won't dominate computers, but the balance could shift if it gets to the point where Macs are routinely outperforming Windows machines.

Having followed tech for long enough, I've noticed that the companies that persist in the modern era are the ones worrying about what's next even when it looks like they're secure. The market tends to shift in such a way that it's fine... until it suddenly isn't.

The likelyhood that Macs with ARM outperforms X86 Windows machines at all is nothing but a pipe dream. Sure, Apple's ARM stuff is getting faster, although not all that significantly but, X86, thanks to AMD, is getting significantly faster and more efficient, with no end in sight. :) Do I think they could be some competition between ARM and X86? Sure but, didn't Nvidia purchase the rights to ARM?
 

kirbyrj

Fully [H]
Joined
Feb 1, 2005
Messages
27,116
No, Apple probably won't dominate computers, but the balance could shift if it gets to the point where Macs are routinely outperforming Windows machines.

Out performing is irrelevant. The problem is the price Apple charges for the image piece. Look at the way Chromebooks are dominating the education space right now. This has nothing to do with performance. It's all about price.

Pretty soon you'll have a whole generation of kids growing up knowing ChromeOS better than anything else.
 

Aurelius

2[H]4U
Joined
Mar 22, 2003
Messages
3,006
The likelyhood that Macs with ARM outperforms X86 Windows machines at all is nothing but a pipe dream. Sure, Apple's ARM stuff is getting faster, although not all that significantly but, X86, thanks to AMD, is getting significantly faster and more efficient, with no end in sight. :) Do I think they could be some competition between ARM and X86? Sure but, didn't Nvidia purchase the rights to ARM?
And your evidence for this is?

I'm not saying it's guaranteed by any means, but dismissing it as a "pipe dream" is dangerous, too. Apple is ahead on adoption of manufacturing processes, and has a demonstrated history of aggressive improvement (despite your claims, the A14 is significantly faster than the A13). When Apple started on custom mobile chips, the A4 was really just competitive with contemporary rivals; now, the cheapest iPhone with a year-old SoC is generally faster than the most expensive Android phones. Companies ignore Apple's chip design capabilities at their peril, and that includes brands like Intel and AMD.

Also, remember why Apple is making the switch to ARM in the first place — because it felt it was being held back and believed it could fare better than Intel (and clearly didn't think AMD would be much of an improvement). I'm not counting on Apple clearly outperforming in every single category, but the odds are that it was making an informed choice based on roadmaps and what it expected to achieve.

Yes, NVIDIA is in in the midst of trying to buy ARM, but it's not guaranteed to clear regulators given the influence it would give NVIDIA over the mobile industry. There's a good possibility that NVIDIA will have to make promises of equal access and fair rates if the deal goes through.
 

Aurelius

2[H]4U
Joined
Mar 22, 2003
Messages
3,006
Out performing is irrelevant. The problem is the price Apple charges for the image piece. Look at the way Chromebooks are dominating the education space right now. This has nothing to do with performance. It's all about price.

Pretty soon you'll have a whole generation of kids growing up knowing ChromeOS better than anything else.

I'd say it's still relevant. Absolute pricing definitely matters, but perceived value is important as well. Apple's main challenge in the x86 era has been to debunk stereotypes that its systems are overpriced compared to equivalent Windows hardware (sometimes true, but not as often as anti-Apple types care to admit). With ARM, there's a chance Apple can outperform or outlast equivalent Windows PCs and become the easy option for anyone who isn't tied to Windows apps.

With that said: it would be great to see significantly cheaper Mac laptops, even if they're unlikely to rival Chromebooks on price anytime soon.
 

kirbyrj

Fully [H]
Joined
Feb 1, 2005
Messages
27,116
I'd say it's still relevant. Absolute pricing definitely matters, but perceived value is important as well. Apple's main challenge in the x86 era has been to debunk stereotypes that its systems are overpriced compared to equivalent Windows hardware (sometimes true, but not as often as anti-Apple types care to admit). With ARM, there's a chance Apple can outperform or outlast equivalent Windows PCs and become the easy option for anyone who isn't tied to Windows apps.

With that said: it would be great to see significantly cheaper Mac laptops, even if they're unlikely to rival Chromebooks on price anytime soon.

It seems like they are trying to position the iPad for that market with the newer smart keyboard with the trackpad.

I also want to see what their plan is for the Mac Pro line. I would think they would need some sort of discrete GPU for some workloads.
 

Ready4Dis

2[H]4U
Joined
Nov 4, 2015
Messages
2,425
pretty sure they aren't worried. Well maybe AMD.
Maybe AMD? I'm not sure I follow that logic, AMD has higher end consumer parts than Intel, so why would they be more worried? And they are more power efficient to boot, so again, why would AMD be the ones concerned at this point?

Anyways, ARM has been coming to windows for how many years now? It's failed how many times now? I don't think it's a huge concern at this point and will take a long time to be more than a novelty item.
 

4saken

[H]F Junkie
Joined
Sep 14, 2004
Messages
11,328
Maybe AMD? I'm not sure I follow that logic, AMD has higher end consumer parts than Intel, so why would they be more worried? And they are more power efficient to boot, so again, why would AMD be the ones concerned at this point?

Anyways, ARM has been coming to windows for how many years now? It's failed how many times now? I don't think it's a huge concern at this point and will take a long time to be more than a novelty item.
$$ is in the enterprise. not with consumers.
 

Krenum

Fully [H]
Joined
Apr 29, 2005
Messages
16,446
A friend of mine told me back in the mid 2000's that ARM chips were the future...he was right. I should've asked him for the lottery numbers.
 

Ready4Dis

2[H]4U
Joined
Nov 4, 2015
Messages
2,425
$$ is in the enterprise. not with consumers.
I get that, but if you remove the Intel name, will it really help them in market penetration in enterprise? I think a lot of companies buy Intel because it's Intel, not because their products are better. So buying Intel, unless they can continue to use the name (which I doubt), doesn't really give them any help. Taking an existing product and slapping SK Hynix on the side of it is probably enough to turn off a lot of big companies, not because it's inferior in any way, just because they buy what they are familiar with.
 

ManofGod

[H]F Junkie
Joined
Oct 4, 2007
Messages
12,071
And your evidence for this is?

I'm not saying it's guaranteed by any means, but dismissing it as a "pipe dream" is dangerous, too. Apple is ahead on adoption of manufacturing processes, and has a demonstrated history of aggressive improvement (despite your claims, the A14 is significantly faster than the A13). When Apple started on custom mobile chips, the A4 was really just competitive with contemporary rivals; now, the cheapest iPhone with a year-old SoC is generally faster than the most expensive Android phones. Companies ignore Apple's chip design capabilities at their peril, and that includes brands like Intel and AMD.

Also, remember why Apple is making the switch to ARM in the first place — because it felt it was being held back and believed it could fare better than Intel (and clearly didn't think AMD would be much of an improvement). I'm not counting on Apple clearly outperforming in every single category, but the odds are that it was making an informed choice based on roadmaps and what it expected to achieve.

Yes, NVIDIA is in in the midst of trying to buy ARM, but it's not guaranteed to clear regulators given the influence it would give NVIDIA over the mobile industry. There's a good possibility that NVIDIA will have to make promises of equal access and fair rates if the deal goes through.

I do not think dismissing it as a pipe dream is dangerous, at all. AMD is improving their performance every single generation, and in no small measure. Their X86 processors are faster than anything ARM related at this time I do not see ARM of any kind catching up, unless Intel gets ahead and stagnates that market again. Heck, the only reason X86 is not as far ahead as it should be now is because of Intel's intentional stagnation of that market.

Truth is, I think Apple is doing what they are doing so they be in complete control from top to bottom. Then they will make even more money, although their machines will not be faster than X86 machines.
 

ManofGod

[H]F Junkie
Joined
Oct 4, 2007
Messages
12,071
A friend of mine told me back in the mid 2000's that ARM chips were the future...he was right. I should've asked him for the lottery numbers.

Then you would have most likely lost with those numbers. Yeah, ARM is in phones but that does not make them the future but simply a competitor.
 

ManofGod

[H]F Junkie
Joined
Oct 4, 2007
Messages
12,071
$$ is in the enterprise. not with consumers.

AMD also has higher end enterprise products as well, although it will take time to get into that space in any meaningful way. This time, Intel does not have their anti competitive practices to lean on. Also, $$$ is also in the consumer market, as well.
 

ManofGod

[H]F Junkie
Joined
Oct 4, 2007
Messages
12,071
I'd say it's still relevant. Absolute pricing definitely matters, but perceived value is important as well. Apple's main challenge in the x86 era has been to debunk stereotypes that its systems are overpriced compared to equivalent Windows hardware (sometimes true, but not as often as anti-Apple types care to admit). With ARM, there's a chance Apple can outperform or outlast equivalent Windows PCs and become the easy option for anyone who isn't tied to Windows apps.

With that said: it would be great to see significantly cheaper Mac laptops, even if they're unlikely to rival Chromebooks on price anytime soon.

The likelyhood of Mac laptops ever being significantly cheaper is even less likely than ARM outperforming X86.
 

cdabc123

2[H]4U
Joined
Jun 21, 2016
Messages
2,762

Attachments

  • 1280px-Processor_families_in_TOP500_supercomputers.svg.png
    1280px-Processor_families_in_TOP500_supercomputers.svg.png
    543 KB · Views: 0

UnknownSouljer

Supreme [H]ardness
Joined
Sep 24, 2001
Messages
6,806
I do not have to look that up to realize that probably means they can shove more low powered cores into the same space. Therefore, ARM is not really outperforming X86 itself, at least in my opinion. Of course, you can always show me that I am wrong, if you can?
The list itself is the evidence. Even by your own statement, you're at minimum acknowledging greater performance per watt out of ARM, which isn't insignificant - in fact quite the opposite. That's an incredibly important metric.

ARM is the only architecture I know that has 15% or greater performance improvements year over year for at least a decade. We know this because a new iPhone comes out every year. And every year there is a 15-30% performance gain.

Here is a interesting graphic representing the top 500 list. And there are quite a few architectures used in supercomputers that do not pose much of a challenge to x86 cpus at there own game.
The major reason why other ISA's aren't a challenge to x86 has more to do with the ubiquity of the x86 platform. This very thread is a precise example of that. It's not uncommon even in this forum (a bastion of the PCMR) that the great advantage of running a PC and Windows in particular is that "I can play every game from the past 40 years" and/or "emulate every console and play those too" or roughly equivalent paraphrases. The important bit of course being that on Windows you can "run anything" and nothing that we do from a consumer computing perspective is coded to run on these ISAs (Power9, ARM, etc).

At a certain point however it becomes a question of the lowest hanging fruit. At what point does a particular ISA's performance and cost have to exceed and outperform the previous one in order to over take it? Or in other words, how cheap and how fast does a competing ISA have to become versus the incumbent for people to not leave the previous one in droves (does ARM as an example have to performance twice has fast and cost half as much as x86? Unlikely that such a performance gap will be necessary, but it illustrates the point)? We've only seen this happen a few times, even in the length of computing history (which dates back to the turn of the 20th century). It has been far more common for certain software (as in OS's or coding languages) to rise and fall long before an ISA does. x86 has existed essentially since the late 70s. But either way, just because it is rare to have a toppling of ISAs, I wouldn't count it out.

Certainly for Microsoft to invest resources they at least see that this could be the future. At minimum it signals they are hedging their bets (especially since we're literally talking about emulating x86 on ARM, Windows to Windows). Even if ARM doesn't go anywhere the potential is clearly there and Microsoft is wise to see that. Apple is obviously playing their hand hard. And as has been noted by more than one individual in this thread and just in general: it's foolish to discount what Apple is doing just because they aren't popular amonst PCMR.

Power cpus are closer then arm to desktop and server dominance.
I don't agree really that PowerPC is going to have a desktop come back. Certainly if they did than they're having big problems, because literally the top two PC/Software manufacturers haven't invested a drop into it. Microsoft which obviously has the lions share. And Apple which basically has the rest. Feel free to let me know when you have a thread that's exactly like this one, except it states that Microsoft is developing Windows for PowerPC and is working on emulating their copy of Windows on it to run x86 and you also have Apple moving back to it.
Certainly without getting more obtuse, ARMs licensing structure lends itself to way more people investing capital into it than Power. And that is again evidenced by literally every smart phone and tablet running off of that architecture. It's 100% obvious that ARM has a more direct move into the consumer computing space than Power does by a long shot. At this point the ARM powered A14 chip in the iPad Air is way more powerful than an 'equivalent' $600 laptop. Its big issue is that Apple isn't running a desktop OS on it - not because of performance, efficiency, or cost.

EDIT: If you're referencing how I stated that ARM was at the top of the Super Computer 500 you're not really understanding my point - or certainly trying to change it. I was responding to a statement that "ARM isn't fast enough to replace x86". Citing that it's on the top of that particular list directly refutes that claim. The issue with Power in comparison with ARM on desktop, despite all the spots on the top 500 list that it occupies, is all of the things I referred to earlier: there is nothing programmed for it and it costs too much.
For building specialized super computers neither of those two things matter, as the chart clearly shows. But clearly there is more to it than how many spaces are filled up in that list by a particular ISA, which above I have acknowledged. ARM clearly doesn't have those same hurdles or at the very least is actively removing those hurdles and is already in the consumer space, unlike Power at any level.
 
Last edited:

Krenum

Fully [H]
Joined
Apr 29, 2005
Messages
16,446
Then you would have most likely lost with those numbers. Yeah, ARM is in phones but that does not make them the future but simply a competitor.

Nvidia would disagree, they didn't spend 40 billion for nothing. ARM is in a lot more than just phones.
 

cdabc123

2[H]4U
Joined
Jun 21, 2016
Messages
2,762
The list itself is the evidence. Even by your own statement, you're at minimum acknowledging greater performance per watt out of ARM, which isn't insignificant - in fact quite the opposite. That's an incredibly important metric.

ARM is the only architecture I know that has 15% or greater performance improvements year over year for at least a decade. We know this because a new iPhone comes out every year. And every year there is a 15-30% performance gain.


The major reason why other ISA's aren't a challenge to x86 has more to do with the ubiquity of the x86 platform. This very thread is a precise example of that. It's not uncommon even in this forum (a bastion of the PCMR) that the great advantage of running a PC and Windows in particular is that "I can play every game from the past 40 years" and/or "emulate every console and play those too" or roughly equivalent paraphrases. The important bit of course being that on Windows you can "run anything" and nothing that we do from a consumer computing perspective is coded to run on these ISAs (Power9, ARM, etc).

At a certain point however it becomes a question of the lowest hanging fruit. At what point does a particular ISA's performance and cost have to exceed and outperform the previous one in order to over take it? Or in other words, how cheap and how fast does a competing ISA have to become versus the incumbent for people to not leave the previous one in droves (does ARM as an example have to performance twice has fast and cost half as much as x86? Unlikely that such a performance gap will be necessary, but it illustrates the point)? We've only seen this happen a few times, even in the length of computing history (which dates back to the turn of the 20th century). It has been far more common for certain software (as in OS's or coding languages) to rise and fall long before an ISA does. x86 has existed essentially since the late 70s. But either way, just because it is rare to have a toppling of ISAs, I wouldn't count it out.

Certainly for Microsoft to invest resources they at least see that this could be the future. At minimum it signals they are hedging their bets (especially since we're literally talking about emulating x86 on ARM, Windows to Windows). Even if ARM doesn't go anywhere the potential is clearly there and Microsoft is wise to see that. Apple is obviously playing their hand hard. And as has been noted by more than one individual in this thread and just in general: it's foolish to discount what Apple is doing just because they aren't popular amonst PCMR.


I don't agree really that PowerPC is going to have a desktop come back. Certainly if they did than they're having big problems, because literally the top two PC/Software manufacturers haven't invested a drop into it. Microsoft which obviously has the lions share. And Apple which basically has the rest. Feel free to let me know when you have a thread that's exactly like this one, except it states that Microsoft is developing Windows for PowerPC and is working on emulating their copy of Windows on it to run x86 and you also have Apple moving back to it.
Certainly without getting more obtuse, ARMs licensing structure lends itself to way more people investing capital into it than Power. And that is again evidenced by literally every smart phone and tablet running off of that architecture. It's 100% obvious that ARM has a more direct move into the consumer computing space than Power does by a long shot. At this point the ARM powered A14 chip in the iPad Air is way more powerful than an 'equivalent' $600 laptop. Its big issue is that Apple isn't running a desktop OS on it - not because of performance, efficiency, or cost.

EDIT: If you're referencing how I stated that ARM was at the top of the Super Computer 500 you're not really understanding my point - or certainly trying to change it. I was responding to a statement that "ARM isn't fast enough to replace x86". Citing that it's on the top of that particular list directly refutes that claim. The issue with Power in comparison with ARM on desktop, despite all the spots on the top 500 list that it occupies, is all of the things I referred to earlier: there is nothing programmed for it and it costs too much.
For building specialized super computers neither of those two things matter, as the chart clearly shows. But clearly there is more to it than how many spaces are filled up in that list by a particular ISA, which above I have acknowledged. ARM clearly doesn't have those same hurdles or at the very least is actively removing those hurdles and is already in the consumer space, unlike Power at any level.
I would agree with many things, the reason we still use x86 chips is entirely software support. However it is notable to mention that windows is not the entirety of what I would describe to be the "consumer computing" market. Linux does pose a viable alternative. That is helpful to ANY other architectures as windows does not support anything other then x86 well. I would also agree that ARM has a possibility to enter the market, It is receiving software support from the mobile side of things which could absolutely result in a "windows device" running on a ARM cpu. The only computers worth noting for there performance with ARM cpus runs on linux and thats fine. As far as emulation goes (either through windows or a hypervisor) you do lose a portion of the performance to the emulation. IBM Power cpus are already capable of x86 emulation within decent margin of running a native OS. The Power cpus are also impressive pieces of silicone even compared to top of the line arm chips and are far more suited to virtualization then most ARM chips I have seen.

As for the supercomputing list ARM can create a powerful supercomputer, so can a gpu cluster or a cluster of the intel phis. However some of these chips would be very poor for attempting to run all programs windows is capable of. They are fundamentally differnt processing styles and emulation of such is expensive to the chip. As such I find it hard to compare the power of ARM vs x86. Especially when there are industry soulutions that are already efficiently capable of that emulation. Software support is the key for any architecture and any program can be efficiently ran on ARM if the developers choose. The issue now is making that choice insted of building onto the mountain of software on x86.
 
Last edited:

cdabc123

2[H]4U
Joined
Jun 21, 2016
Messages
2,762
AMD was fine until Nvidia bought ARM. Now everyone has to step back and look at future plans. It could push open source RISC into the spotlight.
AMD just needs to buy Xilinx and create some silicone allowing for more then a developer coding for a specific architecture.
 

juanrga

Pro-Intel / Anti-AMD Just FYI
Joined
Feb 22, 2017
Messages
2,740
The likelyhood that Macs with ARM outperforms X86 Windows machines at all is nothing but a pipe dream.
No when one is familiar with the microarchitectures and the benchmarks.

Their X86 processors are faster than anything ARM related at this time I do not see ARM of any kind catching up
The last time I checked it, ARM server processors offered the same performance than Zen-based EPYCs but at one fraction of the cost, and I just checked now that ThunderX3 is faster than Zen2-based EPYC

MarvellTX3_16.jpg
MarvellTX3_18.jpg

Yeah, ARM is in phones but that does not make them the future but simply a competitor.
ARM is everywhere from microcontrollers to supercomputers. ARM architecture is the #1 architecture in the global internet infrastructure

https://fuse.wikichip.org/news/1751...ith-new-roadmaps-architectures-and-standards/

I do not have to look that up to realize that probably means they can shove more low powered cores into the same space. Therefore, ARM is not really outperforming X86 itself, at least in my opinion. Of course, you can always show me that I am wrong, if you can?
ARM achieves here more performance with less cores;

Fujitsu A64FX (48C) == 3.3792 TFLOPS

AMD 7H12 (64C) == 2.6624 TFLOPS
 
Last edited:

Aurelius

2[H]4U
Joined
Mar 22, 2003
Messages
3,006
I do not think dismissing it as a pipe dream is dangerous, at all. AMD is improving their performance every single generation, and in no small measure. Their X86 processors are faster than anything ARM related at this time I do not see ARM of any kind catching up, unless Intel gets ahead and stagnates that market again. Heck, the only reason X86 is not as far ahead as it should be now is because of Intel's intentional stagnation of that market.

Truth is, I think Apple is doing what they are doing so they be in complete control from top to bottom. Then they will make even more money, although their machines will not be faster than X86 machines.

Scientists and pollsters will tell you to never dismiss possibilities that are significant and supported by evidence, however probable they may be, and that's the case here.

Based on your posting in this thread, it looks like you're trying to justify your predetermined conclusion (x86 will always be at least as fast) rather than open your mind to see where the market might be going. I'm not saying that Apple will immediately outperform x86, or even that it'll outperform x86 in the long run, just that there's significant potential and that ignoring it is risky. You certainly shouldn't dismiss it based on the failures of Microsoft and Qualcomm to make Windows viable on ARM in previous years.
 

tangoseal

[H]F Junkie
Joined
Dec 18, 2010
Messages
9,089
I apologize as I dont understand what makes arm so hyped? So why is arm all the talk lately? Didnt nVidia buy thier IP?
 
Top