Apple to Announce its own Mac Processor

1_rick

[H]ard|Gawd
Joined
Feb 7, 2017
Messages
1,264
This Apple move is the next wave.... going after the millions of people that no longer have desktops. Having chosen instead to buy a 500-1500 laptop and use nothing else.
I'm less opposed to using Macs than I used to be, but bear in mind you can get a pretty good Windows laptop for like $700 or maybe a bit less on sale, and have been able to for a while now. And, as I said, I have applications I use, both at home and at work, that won't run on Macs, so one would be a secondary machine for me, which means it probably wouldn't get used much.
 

Zarathustra[H]

Fully [H]
Joined
Oct 29, 2000
Messages
30,450
OK seeing as everyone thought I was nuts 4 years back when I said Apple would dump Intel for ARM shortly... (ok I was a little early).

Consumer x86 has 5 years left.

LOL I know I sound nuts. But ya its going to happen. Apple is no doubt going to come out swinging on this... expect aggressive ARM Macbook pricing with performance that will make = priced windows laptops look 5 years out of date. When those actually sell well... I fully expect Microsoft will panic and take another stab at Windows ARM. Perhaps they go with silly Qcom chips like last time... or perhaps they get Samsung involved. And another not so long shot... perhaps they talk AMD into reviving their K12 designs and updating them into a modern Zen chiplet package.

... and when Intel gets wiff of it. I fully expect Intel will expand there current ARM licence for IOT and low power to a low power mid range performance design so they can put Intel ARM chips in widows ARM laptops from companies like HP and Dell.

At some point Intel is going to go full in on their new marketing slogans... where performance isn't everything and benchmarks are bad. They will make their chips lifestyle parts... and talk about their Intel designed XE graphic ARM SOC, and whatever new marketing buzz they can slap into the SOC design.

5 years.. thats my call. 2 years for Apple to prove ARM macbooks will sell very well, and Apple ARM macs at least won't loose any market ground. Then the MS/Intel folks will scramble to get their MEtoo stuff out... and 5 years from now x86 in consumer devices will be relegated to home brew folks like us.

The only way x86 doesn't die now is if Apple launches Arm Macbooks and no one buys them... or Apple accidentally includes a multiplication error, or massive security flaw in their new chips or something. lol
I think we are moving to a market with more diverse instruction sets, but I don't think x86 is going anywhere any time soon.

Apple still has less than 10% of the combined desktop/laptop market. They don't exactly lead the market in any appreciable way, and just because it makes sense for them to move to their in house CPU's doesn't signal that the market is about to change in any appreciable way.
 
Last edited:

Aurelius

2[H]4U
Joined
Mar 22, 2003
Messages
2,809
I've long been an advocate for using the software/hardware necessary for you to do your job and I acknowledge personal preference as being a part of that. So if you're running Wintel or Lintel or whatever because that's what you need for work and for what you do then by all means continue that path so long as it's relevant.

But the one thing to think about and perhaps challenge yourself on is about relevance. At a certain point if things head towards their current conclusion ARM via Apple will exceed what Intel/AMD are doing in the desktop PC market.
Apple themselves have stated they expect their hardware to transition completely over to ARM within 2 years. That means they are already working towards a system that is at least as good, but likely faster, than what their current top Mac Pro offers, which is an Intel 28 Core Xeon with up to 4x Vega II's, which squarely places it above every other OEM workstation.

If they can outclass that hardware and do it for less, how much faster for the cost does Apple have to be for you to at minimum consider them? If Apple ARM literally becomes 4x faster than equivalent Wintel offerings would you switch? What about 10x faster? In other words, it will eventually become adapt or die. I don't think any company is worth dying on a hill for (Apple included).
It's possible that AMD and Intel both might come up with a better x86 solution that is a quantum leap faster. But there is nothing like that on the horizon or on any roadmap. Zen 3 will crush Intel, but crushing ARM will be an entirely different bag of worms. Especially when performance per watt is through the roof and there is a much more obvious roadmap to increase performance than on x86.

All is yet to be revelaed, but in 2 years when an ARM Mac Pro drops, it's going to get real interesting in the PC space. Whether you're on Mac ARM or hanging on Intel/AMD x86 Windows/Linux.
This is something I don't think folks entirely understand. Apple built up its CPU designs to the point where a $399 iPhone is faster than virtually any Android phone. If that translates to Macs (no guarantees it will, but work with me here), there could be a point where you have to get a Mac if you want the fastest laptop, maybe even the fastest desktop, and battery life may see a similar jump. If you're a creative or enthusiast, how much would it pain you if your $2,000 x86 Windows laptop was outperformed by a $1,500 Mac? If you aren't dependent on Windows-only apps at that point, you might consider switching.

Apple is a bit like a freight train. It take a while to get up to speed, but when it does it's hard to stop.
 
Joined
Apr 11, 2017
Messages
671
This is something I don't think folks entirely understand. Apple built up its CPU designs to the point where a $399 iPhone is faster than virtually any Android phone. If that translates to Macs (no guarantees it will, but work with me here), there could be a point where you have to get a Mac if you want the fastest laptop, maybe even the fastest desktop, and battery life may see a similar jump. If you're a creative or enthusiast, how much would it pain you if your $2,000 x86 Windows laptop was outperformed by a $1,500 Mac? If you aren't dependent on Windows-only apps at that point, you might consider switching.

Apple is a bit like a freight train. It take a while to get up to speed, but when it does it's hard to stop.
If Apple's products were priced lower than the competition then I think a black hole will spontaneously form and the world would end.

Apple's entire lineup could be 100% ARM right now and they'd still have less than 10% of the market. They're a 300ft train vs a 3 mile train that is x86 as a whole. Apple can lead the way all they want, companies still have to follow them.

Also I haven't seen anything ARM come close to something like a 3990x.

And there's the whole building a PC thing and upgrading parts. Apple doesn't really do that ever.
Media is already probably 90% consumed on ARM devices and has been for a while.

I think x86 will be just fine for another 10 years at least
 
  • Like
Reactions: Halon
like this

bman212121

[H]ard|Gawd
Joined
Aug 18, 2011
Messages
1,646
That's probably going a bit far. Sure the iPhone SE does have good performance at a reasonable price point, but it's also a stripped down system. The Mac lineup is generally considered a premium device, so I definitely wouldn't expect them to undercut anyone on price. Keep in mind we're talking about the company who released a $999 monitor stand. If Apple can make a $1,500 laptop that performs comparable to a $2,000 Windows one, expect them to charge $2,199 for it. They will enjoy the higher profit margin and most people will gladly pay a few extra dollars to get a premium device. They could maybe bring back the entry level macbook (non pro) and get aggressive with pricing on that, but I don't see that happening on the mid / high end where they will be undercutting Windows PCs.
 

aokman

[H]ard|Gawd
Joined
Jan 3, 2012
Messages
1,043
If Apple's products were priced lower than the competition then I think a black hole will spontaneously form and the world would end.

Apple's entire lineup could be 100% ARM right now and they'd still have less than 10% of the market. They're a 300ft train vs a 3 mile train that is x86 as a whole. Apple can lead the way all they want, companies still have to follow them.

Also I haven't seen anything ARM come close to something like a 3990x.

And there's the whole building a PC thing and upgrading parts. Apple doesn't really do that ever.
Media is already probably 90% consumed on ARM devices and has been for a while.

I think x86 will be just fine for another 10 years at least
Just because we haven’t seen anything yet, doesn’t mean it isn’t coming. There has been no demand for such an ARM processor or a form factor for it to exist in. Apple are creating such a platform right now.

Anyone who is into ARM and wants to see x86 being put in its place should be happy with what Apple is trying to achieve.
 

Zarathustra[H]

Fully [H]
Joined
Oct 29, 2000
Messages
30,450
This extreme tech article is a little old now (2014 vintage) but it analyzes the different ISA's based on energy used to complete a certain task (power used X time it takes)

The conclusion? ISA doesn't really matter.

Personally, if any ISA had to replace x86 I'm rooting for RISC-V because it is royalty free and open source, so competition could flourish without licenses getting in the way.
 

UnknownSouljer

Supreme [H]ardness
Joined
Sep 24, 2001
Messages
6,555
If Apple's products were priced lower than the competition then I think a black hole will spontaneously form and the world would end.
Well, I've mentioned this multiple times in this thread too. But there is a misnomer that every Apple product isn't competitive and that simply isn't true.

The iMac is generally one of the most competitive computers in their product stack. The 5k iMac has historically been the same cost as just a 5k Dell display using the same panel. It's hard to argue with getting a display and a "free computer" built into it.
The Mac Pro is dollar for dollar way more competitive than HEDT offerings from other OEMs like Dell and HP.
And I would argue the Macbook Pro is very similar in pricing structure to high end business lines like Lenovo and Dell. Both of those companies might be less expensive than specifically the Macbook Pro, but it's definitely in the same ballpark as, and not some ridiculous out of place amount.
And as was already brought up, the iPhone SE 2 has the same A13 chip as the current Apple flagship iPhone 11 Pro and it costs $399. There is no (new) phone that can compete anywhere close to this price compared to its features. Granted not all of its specs are headlining and some of them are from older generations, but as a package it doesn't lack anything and is capable of doing anything any other smart phone is capable of.

Apple's entire lineup could be 100% ARM right now and they'd still have less than 10% of the market. They're a 300ft train vs a 3 mile train that is x86 as a whole. Apple can lead the way all they want, companies still have to follow them.
I'm not saying this will happen, but bare with me here, if Apple is able to produce an SoC that is 2x-3x faster than competing hardware from AMD and Intel for the same cost how long do you think those other platforms will remain relevant?

If you want a machine that costs more or the same for less performance than I suppose that it doesn't matter. But this is why processors like RISC and SPARC lost relevance, because they couldn't keep up anymore, not because their architectures weren't powerful or good.

Blackberry had the best low wattage processors, their smart phones could literally run for days. But after Apple changed the paradigm to full touch screen interfaces and devices in your pocket that could do it all and Blackberry as a result lost relevance. I'm not saying it will be the same level for x86, I'm just saying that it's happened before. And I wouldn't ever say that any piece of technology is secure, when basically everything is better, faster, more efficient, every 1.5x years. The history of technology is litered with upsets. You have to have a very short memory of history to think it hasn't happened before.

Also I haven't seen anything ARM come close to something like a 3990x.
I've stated this before in this thread: Apple's intention is replace their entire product stack with custom in-house made ARM chips. Which means that at some point they are going to have to replace a 28-Core Skylake-X system with up to 4x Vega II GPUs. So, Apple has in their product development stack something that is able to compete with that, or obviously with the purposes of moving architectures, something that is greater than that. Now part of the equation we don't know is if they will continue working with AMD to get GPU's (like Navi 2) or if they not only will try to compete in the CPU space but the high end GPU space as well. That might be a two part equation. But certainly they are going to take Intel's high end HEDT part and replace it.

I don't think it's out of the question at all that they could scale their ARM processors to easily have 16-Core mobile parts (ala Macbook Pro) and 64-core desktop workstation parts (for iMac Pro and Mac Pro). Especially with their advantages with TSMC's 5nm process node that they're using. They have excellent people creating their hardware now, all we're really discussing is their ability to scale it up. With their timeline of 2 years, I'd say that it's more than likely such parts are already in the pipeline. They've thought this far ahead to make the transition easy. It's not hard to have the faith that they've thought far enough ahead to have all their Mac parts and timelines planned out.

We haven't even seen their first Mac processors yet. As was stated by Johnny that all of their Mac's will be using custom SoC's. The dev unit is just that: a dev unit. It isn't supposed to be final hardware that will be sold to consumers. It's just hardware that devs can program on in the meantime before actual consumer hardware is available to them.

And there's the whole building a PC thing and upgrading parts. Apple doesn't really do that ever.
Media is already probably 90% consumed on ARM devices and has been for a while.
Well, there is the Mac Pro. The workstation at the top of their product stack which is, I would argue, much better dollar per dollar than competing workstations from HP or Dell.
However I don't think you're thinking of this the same way most people do. PCMR gets caught up on building your own machine, but what if Apple provides a machine for you that you "can't build" but is 3x as fast as a PC you can build for the same amount of money? What then, does building a machine still have relevance, or is that just a hobby for you?
More to the point, Apple caters far more to businesses and devs that deal with OEMs. In other words individual consumers are least concern for business desktop Mac's, they're far more interested in the market that buys 20 iMacs (or Dells or HPs), needs/wants service for 2-5 years and then replaces them. That's a totally different market than market than PCMR consumers that build machines as a hobby. And quite frankly the pockets are a lot deeper. It doesn't make business sense generally speaking to cater to the builder crowd. More or less other than on their Halo product, the Mac Pro.

If you work in an editing house for films (whether big budget Hollywood or smaller independent scale say for commercial etc) it's far easier to just buy 10 machines from an OEM rather than build them yourself. And whether you're doing PC or Mac your concern isn't about upgrading them, it's about support and ease of operation. With that in mind, it makes far more sense to get iMac Pros that don't take up a lot of space on a desk or on the ground that grind hard and are silent rather than trying to build 10 machines and manage them all yourself. And that's small scale. What about 150 machines? No one is building that, they're always having an OEM do it and like I say that's where Apple fits in.

Everyone else in terms of consumers is, whether you like it or care about statistic or not, primarily on laptops. And guess what? No one really builds laptops. That's why their lineup basically looks like laptops and the iMac with Mac Pro at the top. Apple is far more interested in eating the biggest pieces of the pie and they're not after the relatively low percentage of PCMR builders that would be basically inconsequential to their bottom line. Laptops are where it's at for most folks and up to this point they haven't been really interested in catering to gamers. We'll see if that changes now that essentially every iOS game will be macOS compatible on ARM.

I think x86 will be just fine for another 10 years at least
You can coast for a while, but when you make that statement, it could be in decline the entire time. Intel could honestly become the next IBM or Blackberry slowly bleeding out as it thrashes around for relevance.
If Apple is capable of making a part that is faster than their 14nm+++++ Skylake-X 28-Core CPU then there is clearly something there. And Apple obviously feels confident they can, as their plan is to transition their entire product line in 2 years.
 
Last edited:

ChadD

Supreme [H]ardness
Joined
Feb 8, 2016
Messages
4,700
I think we are moving to a market with more diverse instruction sets, but I don't think x86 is going anywhere any time soon.

Apple still has less than 10% of the combined desktop/laptop market. They don't exactly lead the market in any appreciable way, and just because it makes sense for them to move to their in house CPU's doesn't signal that the market is about to change in any appreciable way.
I agree with you in that 10% is no threat. I can just see a future where Microsoft and Intel both do a little bit of panic mode if the Apple product is mid market attractive. If Apple launches them and they fall on their face... of course that won't happen. But if they launch them and they grab even a small bump in market share. I imagine we might see MS shopping for ARM chip partners to go toe to toe with Apple. That could end up being Qcom (which MS has already tried to dance with once) or AMD who we all know has had fully fleshed out High performance ARM parts that went as far as test fabrication. IMO if Intel gets a feel that MS is going to go for Windows ARM round 3 with a real partner like AMD or Samsung.... I don't think its crazy to think Intel may just speed push some 3D chip Stacked ARM design.

I don't know it might be crazy for Intel to throw in with ARM in regards to main stream parts. But with the PR beating they have been taking lately... I wouldn't rule it out. Wounded animals can be pretty unpredictable. I can imagine their bean counter board may see main stream laptop ARM chips as a way to steal some thunder and... with them lately saying they want to market "lifestyle" features of their chips over pure brute performance. Going as far as to say benchmarks are not a good Intel sales tool. I think the idea of Intel finally admitting x86 isn't idea for low power consumer devices... and building "lifestyle" consumer ARM parts, and focusing x86 on server and HEDC.

Of course none of that comes to pass unless Apples first Macbook out of the gate turns a lot of heads. Apple needs a jobesk launch... we'll see if they still have it in them. I am not expecting no matter what that Apple ends up with massive market share... but as they have shown with all the i stuffs. What is more interesting is how the market reacts... in almost everything they have done for years now the market tends to copy the best and many of the worst Apple design features. lol
 

ChadD

Supreme [H]ardness
Joined
Feb 8, 2016
Messages
4,700
This extreme tech article is a little old now (2014 vintage) but it analyzes the different ISA's based on energy used to complete a certain task (power used X time it takes)

The conclusion? ISA doesn't really matter.

Personally, if any ISA had to replace x86 I'm rooting for RISC-V because it is royalty free and open source, so competition could flourish without licenses getting in the way.
RISC-V would have had a better chance if ARM didn't exist.

The truth is ARMs licence is pretty open and if a company has the capital to design their own cores (such as Apple) the fees ARM is charging are a very minor part of the overall cost of the project. And what ARM brings to the table that RISC-V doesn't... is a company large enough to do the heavy lifting on design for the smaller players. There are really no chip using companies that really want to see RISC-V take off. (there is some edge cases like Western Digital using RISC-V but with the razor thin margins on storage devices that is logical) The little companies can licence for very low cost and use fully modern and solid core designs from ARM... mixing them with any other bits they want in their SOC designs. (Companies like meditek licencee and build complete ARM designed CPU+GPUs) The bigger players like Apple and Samsung can licence the ISA and design their own compliant core.

ARM is the best of both worlds.
1) They aren't Intel you want a licence no problems.
2) They aren't RISC-V where they say here is the ISA its free... but hire the engineers and design the core yourself.

ARM is wiling to licence the ISA, and are big enough to design modern cores for the companies not quite ready to design their own cores. Apple spent 1 billion to design the A4... and that was a decade back, it would cost at least double that today. So to design a actual competitive RISC-V chip would no doubt cost at least that much... and the software stacks would still be a massive issue.

Intel is in trouble long term in their war with ARM... because they are permissive with their licences, and a large enough player at this point to keep updated core designs ready to sell to anyone wanting them. The Linux nerd in me wants RISC-V to be more... but sadly it will never be anymore then a good option for companies like WD to use for storage controllers or Nvidia to use in internal control on GPUs. In those cases ARMs off the shelf designs are over kill, and frankly the ISA is irrelevant for such uses where they don't need to communicate outside that one piece of hardware.
 

Snowdog

[H]F Junkie
Joined
Apr 22, 2006
Messages
10,773
However I don't think you're thinking of this the same way most people do. PCMR gets caught up on building your own machine, but what if Apple provides a machine for you that you "can't build" but is 3x as fast as a PC you can build for the same amount of money? What then, does building a machine still have relevance, or is that just a hobby for you?
More to the point, Apple caters far more to businesses and devs that deal with OEMs. In other words individual consumers are least concern for business desktop Mac's, they're far more interested in the market that buys 20 iMacs (or Dells or HPs), needs/wants service for 2-5 years and then replaces them. That's a totally different market than market than PCMR consumers that build machines as a hobby. And quite frankly the pockets are a lot deeper. It doesn't make business sense generally speaking to cater to the builder crowd. More or less other than on their Halo product, the Mac Pro.
There two areas that have kept people on PCs that will likely continue.

1: Budget computers. Most of the market is probably in the under $1000 PC, but these days, that is mostly budget laptops. Apple has nothing here. The least expensive Apple laptop is $1000. Is Apple going to have a budget laptop? Skepticism seems warranted.

2: PCMR: This market seems to be some combination of Builder Hobbyist, and AAA Gamers. Apple is never going to appeal to Builder Hobbyists for obvious reasons. AAA gaming might be possible, but Apple would really need to work for years to make Macs a AAA gaming platform, something they have really shown nothing but disdain for in the past. It could happen, but it would take years, of work to Macs into a AAA gaming platform, and I think skepticism is still warranted.
 
Joined
Apr 11, 2017
Messages
671
Well, I've mentioned this multiple times in this thread too. But there is a misnomer that every Apple product isn't competitive and that simply isn't true.

The iMac is generally one of the most competitive computers in their product stack. The 5k iMac has historically been the same cost as just a 5k Dell display using the same panel. It's hard to argue with getting a display and a "free computer" built into it.
The Mac Pro is dollar for dollar way more competitive than HEDT offerings from other OEMs like Dell and HP.
And I would argue the Macbook Pro is very similar in pricing structure to high end business lines like Lenovo and Dell. Both of those companies might be less expensive than specifically the Macbook Pro, but it's definitely in the same ballpark as, and not some ridiculous out of place amount.
And as was already brought up, the iPhone SE 2 has the same A13 chip as the current Apple flagship iPhone 11 Pro and it costs $399. There is no (new) phone that can compete anywhere close to this price compared to its features. Granted not all of its specs are headlining and some of them are from older generations, but as a package it doesn't lack anything and is capable of doing anything any other smart phone is capable of.


I'm not saying this will happen, but bare with me here, if Apple is able to produce an SoC that is 2x-3x faster than competing hardware from AMD and Intel for the same cost how long do you think those other platforms will remain relevant?

If you want a machine that costs more or the same for less performance than I suppose that it doesn't matter. But this is why processors like RISC and SPARC lost relevance, because they couldn't keep up anymore, not because their architectures weren't powerful or good.

Blackberry had the best low wattage processors, their smart phones could literally run for days. But after Apple changed the paradigm to full touch screen interfaces and devices in your pocket that could do it all and Blackberry as a result lost relevance. I'm not saying it will be the same level for x86, I'm just saying that it's happened before. And I wouldn't ever say that any piece of technology is secure, when basically everything is better, faster, more efficient, every 1.5x years. The history of technology is litered with upsets. You have to have a very short memory of history to think it hasn't happened before.


I've stated this before in this thread: Apple's intention is replace their entire product stack with custom in-house made ARM chips. Which means that at some point they are going to have to replace a 28-Core Skylake-X system with up to 4x Vega II GPUs. So, Apple has in their product development stack something that is able to compete with that, or obviously with the purposes of moving architectures, something that is greater than that. Now part of the equation we don't know is if they will continue working with AMD to get GPU's (like Navi 2) or if they not only will try to compete in the CPU space but the high end GPU space as well. That might be a two part equation. But certainly they are going to take Intel's high end HEDT part and replace it.

I don't think it's out of the question at all that they could scale their ARM processors to easily have 16-Core mobile parts (ala Macbook Pro) and 64-core desktop workstation parts (for iMac Pro and Mac Pro). Especially with their advantages with TSMC's 5nm process node that they're using. They have excellent people creating their hardware now, all we're really discussing is their ability to scale it up. With their timeline of 2 years, I'd say that it's more than likely such parts are already in the pipeline. They've thought this far ahead to make the transition easy. It's not hard to have the faith that they've thought far enough ahead to have all their Mac parts and timelines planned out.

We haven't even seen their first Mac processors yet. As was stated by Johnny that all of their Mac's will be using custom SoC's. The dev unit is just that: a dev unit. It isn't supposed to be final hardware that will be sold to consumers. It's just hardware that devs can program on in the meantime before actual consumer hardware is available to them.


Well, there is the Mac Pro. The workstation at the top of their product stack which is, I would argue, much better dollar per dollar than competing workstations from HP or Dell.
However I don't think you're thinking of this the same way most people do. PCMR gets caught up on building your own machine, but what if Apple provides a machine for you that you "can't build" but is 3x as fast as a PC you can build for the same amount of money? What then, does building a machine still have relevance, or is that just a hobby for you?
More to the point, Apple caters far more to businesses and devs that deal with OEMs. In other words individual consumers are least concern for business desktop Mac's, they're far more interested in the market that buys 20 iMacs (or Dells or HPs), needs/wants service for 2-5 years and then replaces them. That's a totally different market than market than PCMR consumers that build machines as a hobby. And quite frankly the pockets are a lot deeper. It doesn't make business sense generally speaking to cater to the builder crowd. More or less other than on their Halo product, the Mac Pro.

If you work in an editing house for films (whether big budget Hollywood or smaller independent scale say for commercial etc) it's far easier to just buy 10 machines from an OEM rather than build them yourself. And whether you're doing PC or Mac your concern isn't about upgrading them, it's about support and ease of operation. With that in mind, it makes far more sense to get iMac Pros that don't take up a lot of space on a desk or on the ground that grind hard and are silent rather than trying to build 10 machines and manage them all yourself. And that's small scale. What about 150 machines? No one is building that, they're always having an OEM do it and like I say that's where Apple fits in.

Everyone else in terms of consumers is, whether you like it or care about statistic or not, primarily on laptops. And guess what? No one really builds laptops. That's why their lineup basically looks like laptops and the iMac with Mac Pro at the top. Apple is far more interested in eating the biggest pieces of the pie and they're not after the relatively low percentage of PCMR builders that would be basically inconsequential to their bottom line. Laptops are where it's at for most folks and up to this point they haven't been really interested in catering to gamers. We'll see if that changes now that essentially every iOS game will be macOS compatible on ARM.


You can coast for a while, but when you make that statement, it could be in decline the entire time. Intel could honestly become the next IBM or Blackberry slowly bleeding out as it thrashes around for relevance.
If Apple is capable of making a part that is faster than their 14nm+++++ Skylake-X 28-Core CPU then there is clearly something there. And Apple obviously feels confident they can, as their plan is to transition their entire product line in 2 years.
I can't see Apple making any kind of gaming desktop, let alone one that I would want.
My last build was the most amount of money I've ever spent on a computer, around $1200 (see sig). That gets me a 3900x and a 5700.
$1200 doesn't get me much at all with Apple. The closest thing it can get me is a 21 inch imac with a 8th gen quad core i3 with 8GB of RAM and a 1TB spinner and a Radeon 555X 2GB. And that costs $100 more.
For Apple to be relevant to me, they are going to have to make some serious changes and be more consumer friendly and cheaper.
The strategy of telling the consumer what they need won't fly with everyone. People want options.
Also, you can have the fastest CPU in the world, but if I can't use it for anything, it's pretty useless to me except for a showpiece. Same goes for Apple, they could have the fastest computers in the world, but it for sure isn't going to be cheap enough for it to matter to me. For the stuff that is in my price range, it will be more worth it to buy something else. That is Apple's business model and its not for me.
ARM still has a long way to go to be relevant in the consumer desktop space and enterprise server space. I won't be worried until I start seeing ARM desktops for personal and businesses from Dell, HP, etc and people actually buying them. At that point, I will probably already have one and won't be as "worried"
 

OFaceSIG

2[H]4U
Joined
Aug 31, 2009
Messages
2,426
Nobody, including Amazon using ARM for client workloads. Amazon is slowly building up their internal OS distros to support it because they want to save costs on hardware they generates them zero revenue. Customer facing equipment? Will be Intel, and more and more AMD overtime.

Unless RHEL comes out with a serious distro that runs better on ARM than it does on x86, ARM is dead on arrival in the client facing server space.
 

kirbyrj

Fully [H]
Joined
Feb 1, 2005
Messages
26,278
Apple is a bit like a freight train. It take a while to get up to speed, but when it does it's hard to stop.
Nothing in the entire history of Macs leads me to believe this is true in the slightest. Apple succeeded in smart phones and music players because there were few other serious options at the time of release. In laptops, desktops, and servers, there are so many other options available. If nothing else, they can afford this move because they are so far behind in marketshare if it doesn't work out for them they didn't lose much.
 
Last edited:

DWolvin

2[H]4U
Joined
Nov 13, 2002
Messages
2,292
I pretty much agree; If they can pull it off, good for them. I don't see them matching the 28 core xeon right away, but it's aimed at such a specific sector that it will be easier to match than most would expect. At the end of the day, competition is good, even if it's only for 12% of the computers out there. It will keep AMD up at night, and Intel (I feel) is already furiously looking to improve. Win-Win for us consumers.
 

Zarathustra[H]

Fully [H]
Joined
Oct 29, 2000
Messages
30,450
RISC-V would have had a better chance if ARM didn't exist.

The truth is ARMs licence is pretty open and if a company has the capital to design their own cores (such as Apple) the fees ARM is charging are a very minor part of the overall cost of the project. And what ARM brings to the table that RISC-V doesn't... is a company large enough to do the heavy lifting on design for the smaller players. There are really no chip using companies that really want to see RISC-V take off. (there is some edge cases like Western Digital using RISC-V but with the razor thin margins on storage devices that is logical) The little companies can licence for very low cost and use fully modern and solid core designs from ARM... mixing them with any other bits they want in their SOC designs. (Companies like meditek licencee and build complete ARM designed CPU+GPUs) The bigger players like Apple and Samsung can licence the ISA and design their own compliant core.

ARM is the best of both worlds.
1) They aren't Intel you want a licence no problems.
2) They aren't RISC-V where they say here is the ISA its free... but hire the engineers and design the core yourself.

ARM is wiling to licence the ISA, and are big enough to design modern cores for the companies not quite ready to design their own cores. Apple spent 1 billion to design the A4... and that was a decade back, it would cost at least double that today. So to design a actual competitive RISC-V chip would no doubt cost at least that much... and the software stacks would still be a massive issue.

Intel is in trouble long term in their war with ARM... because they are permissive with their licences, and a large enough player at this point to keep updated core designs ready to sell to anyone wanting them. The Linux nerd in me wants RISC-V to be more... but sadly it will never be anymore then a good option for companies like WD to use for storage controllers or Nvidia to use in internal control on GPUs. In those cases ARMs off the shelf designs are over kill, and frankly the ISA is irrelevant for such uses where they don't need to communicate outside that one piece of hardware.
I see what you are saying, but I still think it is a bad idea to allow one company, ARM Holdings to have that much power and control.

They may be permissive with their licensing today, but will that continue indefinitely when they run everything?

Don't get me wrong, I understand why companies large and small use it. ARM has made it easy for them. I just feel like we are setting ourselves up for the next problem company in 15 years.
 
  • Like
Reactions: ChadD
like this

SOAREVERSOR

Limp Gawd
Joined
Apr 10, 2017
Messages
326
I see what you are saying, but I still think it is a bad idea to allow one company, ARM Holdings to have that much power and control.

They may be permissive with their licensing today, but will that continue indefinitely when they run everything?

Don't get me wrong, I understand why companies large and small use it. ARM has made it easy for them. I just feel like we are setting ourselves up for the next problem company in 15 years.
we are setting ourselves up for the next massive problem company but that's capitalism.
 

melk

[H]ard|Gawd
Joined
Jan 21, 2014
Messages
1,030
Nobody, including Amazon using ARM for client workloads. Amazon is slowly building up their internal OS distros to support it because they want to save costs on hardware they generates them zero revenue. Customer facing equipment? Will be Intel, and more and more AMD overtime.

Unless RHEL comes out with a serious distro that runs better on ARM than it does on x86, ARM is dead on arrival in the client facing server space.
Client workloads? Like AWS EC2 compute?

https://www.anandtech.com/show/15578/cloud-clash-amazon-graviton2-arm-against-intel-and-amd


Also just announced yesterday, as measured on the Top500 list there is a new #1 supercomputer in the world, powered in part by yes, Arm cores, the first time Arm has been on the list. Roughly 2.5x times faster than the current IBM Summit system.

https://www.servethehome.com/supercomputer-fugaku-by-fujitsu-and-riken-revealed-at-no-1/
 

Aurelius

2[H]4U
Joined
Mar 22, 2003
Messages
2,809
If Apple's products were priced lower than the competition then I think a black hole will spontaneously form and the world would end.

Apple's entire lineup could be 100% ARM right now and they'd still have less than 10% of the market. They're a 300ft train vs a 3 mile train that is x86 as a whole. Apple can lead the way all they want, companies still have to follow them.

Also I haven't seen anything ARM come close to something like a 3990x.

And there's the whole building a PC thing and upgrading parts. Apple doesn't really do that ever.
Media is already probably 90% consumed on ARM devices and has been for a while.

I think x86 will be just fine for another 10 years at least
Oh, don't get me wrong, any kind of market change (if it happens) would take years to play out. And I wouldn't be surprised if Apple simply focused on stuffing as much hardware into a Mac as possible at a given price rather than lowering prices. I'm just thinking that there could be a point where the performance advantage may be clear enough that a Mac one step down the rung might outperform a given Windows PC.

I do agree that Apple's biggest challenge is building an ARM chip that can outperform very high-end, many-core x86 processors. Clearly it thinks it can as the transition will last just a couple of years, but I wonder how an ARM-based Mac Pro will fare in practice.

Custom-built PCs will still likely run Windows, but remember that laptops generally outsell desktops, and that most desktops are off-the-shelf systems. Apple has a pretty large potential audience.
 

melk

[H]ard|Gawd
Joined
Jan 21, 2014
Messages
1,030
I pretty much agree; If they can pull it off, good for them. I don't see them matching the 28 core xeon right away, but it's aimed at such a specific sector that it will be easier to match than most would expect. At the end of the day, competition is good, even if it's only for 12% of the computers out there. It will keep AMD up at night, and Intel (I feel) is already furiously looking to improve. Win-Win for us consumers.
You don't see them matching a 28 core Xeon right away, because....? You should see what current high end Arm CPU's look like... check the links I've posted above. Here's another:

https://www.servethehome.com/ampere-altra-max-targets-a-128-core-arm-cpu-shipping-in-2021/

Here's the current shipping high end. To give an idea of what high end Arm looks like right now. 128-core chips shipping next year.

1592926553686.png


Now, that's not Apple of course, and Ampere Altra is not the only shipping high-end Arm chips, but Apple now has a decade+ of custom chip design at this point, which it happens to absolutely lead in performance in the market (power constrained phones and tablets). They've proven they know what they are doing when they are constrained by power and thermals... it's not hard to imagine the performance they can get from their cpu design when they are given an significant bump in power budget.

X86 will be around a long time still, certainly. but it's not hard to see the momentum of innovation is radically different between the two ISA's.
 

Aurelius

2[H]4U
Joined
Mar 22, 2003
Messages
2,809
Nothing in the entire history of Macs leads me to believe this is true in the slightest. Apple succeeded in smart phones and music players because there were few other serious options at the time of release. In laptops, desktops, and servers, there are so many other options available. If nothing else, they can afford this move because they are so far behind in marketshare if it doesn't work out for them they didn't lose much.
It's not guaranteed to translate to Macs by any means, but you're downplaying things too much.

Remember, when the iPhone debuted in 2007 there were numerous serious options from RIM, Palm, Microsoft's partners, Nokia... you get the idea. Many of those vendors dismissed the iPhone because they had dominant share and, yes, because the original iPhone was lacking a few features (most notably 3G, video recording and enterprise support). But they didn't understand that Apple was determined to build on that foundation, and that it would be much easier for Apple to iterate on a breakthrough design than for rivals to counter inertia and revamp their own phone designs. That freight train built momentum and, a few years later, every single one of those dominant companies was in danger -- some of them before Android took off.

The PC market is certainly different and I do agree that Apple doesn't have much to lose, but one of the golden rules of tech is that you ignore Apple at your peril. I'm not expecting Apple to become the majority, but I wouldn't rule out significant gains in market share, either.
 

kirbyrj

Fully [H]
Joined
Feb 1, 2005
Messages
26,278
It's not guaranteed to translate to Macs by any means, but you're downplaying things too much.

Remember, when the iPhone debuted in 2007 there were numerous serious options from RIM, Palm, Microsoft's partners, Nokia... you get the idea. Many of those vendors dismissed the iPhone because they had dominant share and, yes, because the original iPhone was lacking a few features (most notably 3G, video recording and enterprise support). But they didn't understand that Apple was determined to build on that foundation, and that it would be much easier for Apple to iterate on a breakthrough design than for rivals to counter inertia and revamp their own phone designs. That freight train built momentum and, a few years later, every single one of those dominant companies was in danger -- some of them before Android took off.

The PC market is certainly different and I do agree that Apple doesn't have much to lose, but one of the golden rules of tech is that you ignore Apple at your peril. I'm not expecting Apple to become the majority, but I wouldn't rule out significant gains in market share, either.
That's a fairer statement than your previous one ;). The nascent smartphone market circa 2007 was definitely different than the PC market. For good or bad, Apple has already been a major (and minor) player in that market for most of its history with mixed results.

We'll see what happens though.
 

OFaceSIG

2[H]4U
Joined
Aug 31, 2009
Messages
2,426
Client workloads? Like AWS EC2 compute?

https://www.anandtech.com/show/15578/cloud-clash-amazon-graviton2-arm-against-intel-and-amd


Also just announced yesterday, as measured on the Top500 list there is a new #1 supercomputer in the world, powered in part by yes, Arm cores, the first time Arm has been on the list. Roughly 2.5x times faster than the current IBM Summit system.

https://www.servethehome.com/supercomputer-fugaku-by-fujitsu-and-riken-revealed-at-no-1/
That's new, so we'll see if anyone puts anything of consequence on it. I doubt it anytime soon. But you are correct, it is "available".

Supercomputers aren't for client, aka customer, workloads.
 

Shoganai

Gawd
Joined
Dec 5, 2018
Messages
730
That's new, so we'll see if anyone puts anything of consequence on it. I doubt it anytime soon. But you are correct, it is "available".

Supercomputers aren't for client, aka customer, workloads.
I don’t think that’s the point. This is the first time something with ARM has ever made the top 500 list, let alone the number one spot.
 
Joined
Apr 11, 2017
Messages
671
It's not guaranteed to translate to Macs by any means, but you're downplaying things too much.

Remember, when the iPhone debuted in 2007 there were numerous serious options from RIM, Palm, Microsoft's partners, Nokia... you get the idea. Many of those vendors dismissed the iPhone because they had dominant share and, yes, because the original iPhone was lacking a few features (most notably 3G, video recording and enterprise support). But they didn't understand that Apple was determined to build on that foundation, and that it would be much easier for Apple to iterate on a breakthrough design than for rivals to counter inertia and revamp their own phone designs. That freight train built momentum and, a few years later, every single one of those dominant companies was in danger -- some of them before Android took off.

The PC market is certainly different and I do agree that Apple doesn't have much to lose, but one of the golden rules of tech is that you ignore Apple at your peril. I'm not expecting Apple to become the majority, but I wouldn't rule out significant gains in market share, either.
You must of forgotten that Apple has done this before when switching from PowerPC to x86. They still have inconsequential market share after switching. Switching from x86 to ARM will probably have the same effect. The only difference is that PowerPC was never really a consumer oriented arch (I've never really heard of PowerPC computers besides old Macs) anyways. Most people ran x86 already. Today, basically all Servers and Desktops are still x86, and now Apple is switching to ARM. I don't expect x86 to move that much until "equivalent ARM PCs" are somewhat readily available.
The closest thing we got right now are Raspberry Pi's which is basically low-end smartphone parts, and its been over 8 years since the 1st one was released. Their latest one has a quad core A72 CPU running at 1.5Ghz and an option for 8GB of RAM for $75. I still want one to mess with, but that's about it, they are "toys" to mess with and do basic office tasks that smartphones can already do.
On one end of the spectrum we have Rasberry Pi at $75, the other end, we will have ARM Macs that will likely be $1000+ minimum. Nothing really in the range of $100-$999 that is easily accessible.
 

Aurelius

2[H]4U
Joined
Mar 22, 2003
Messages
2,809
You must of forgotten that Apple has done this before when switching from PowerPC to x86. They still have inconsequential market share after switching. Switching from x86 to ARM will probably have the same effect. The only difference is that PowerPC was never really a consumer oriented arch (I've never really heard of PowerPC computers besides old Macs) anyways. Most people ran x86 already. Today, basically all Servers and Desktops are still x86, and now Apple is switching to ARM. I don't expect x86 to move that much until "equivalent ARM PCs" are somewhat readily available.
The closest thing we got right now are Raspberry Pi's which is basically low-end smartphone parts, and its been over 8 years since the 1st one was released. Their latest one has a quad core A72 CPU running at 1.5Ghz and an option for 8GB of RAM for $75. I still want one to mess with, but that's about it, they are "toys" to mess with and do basic office tasks that smartphones can already do.
On one end of the spectrum we have Rasberry Pi at $75, the other end, we will have ARM Macs that will likely be $1000+ minimum. Nothing really in the range of $100-$999 that is easily accessible.
I'm definitely aware of the transition -- I bought one of the last pre-Intel iMacs!

The circumstances are different, I'd say. Apple was moving to Intel chips to catch up and protect its share, hopefully gaining some users along the way. It's switching to ARM to move ahead of Intel, if not x86 as a whole.

I don't doubt that Apple can make ARM processors that are competitive with mainstream x86 chips, if not faster. The challenge is in making a good use case and otherwise pitching them. If Apple can definitively say "our $999 MacBook Air is faster and longer-lasting than any similarly-priced Windows ultrabook," that's when it might reel in users.
 

juanrga

Pro-Intel / Anti-AMD Just FYI
Joined
Feb 22, 2017
Messages
2,642
This extreme tech article is a little old now (2014 vintage) but it analyzes the different ISA's based on energy used to complete a certain task (power used X time it takes)

The conclusion? ISA doesn't really matter.
ISA matters. Why do you believe Intel is inventing new ISAs like AVX and AMD using them?

Also I haven't seen anything ARM come close to something like a 3990x.
The Q80-33 is shipping and fastest chips will start sampling by end of year.
 
Last edited:

juanrga

Pro-Intel / Anti-AMD Just FYI
Joined
Feb 22, 2017
Messages
2,642
Also just announced yesterday, as measured on the Top500 list there is a new #1 supercomputer in the world, powered in part by yes, Arm cores, the first time Arm has been on the list. Roughly 2.5x times faster than the current IBM Summit system.

https://www.servethehome.com/supercomputer-fugaku-by-fujitsu-and-riken-revealed-at-no-1/
What do you mean by "powered in part"?

Fugaku is not the first time an ARM computer enter in the list.

https://www.top500.org/system/179565/
 

ChadD

Supreme [H]ardness
Joined
Feb 8, 2016
Messages
4,700
I see what you are saying, but I still think it is a bad idea to allow one company, ARM Holdings to have that much power and control.

They may be permissive with their licensing today, but will that continue indefinitely when they run everything?

Don't get me wrong, I understand why companies large and small use it. ARM has made it easy for them. I just feel like we are setting ourselves up for the next problem company in 15 years.
I can see that... of course the problem for ARM at that point would be. In order to be everywhere ARM the industry has built frameworks and tools to make it as easy as possible to port code. As much as I am saying Risc-v has a long road.... IF a company like Samsung or one of the chinese phone companies or any other tech company with a couple billion to burn wanted to build a modern Risc-V chip way they go. Apple claiming a simple Xcode recompile is perhaps a bit optimistic... but its also not to far off, and that cross platform, cross isa compile work gets better in every ecosystem every month. Android has in the past supported x86 and MIPS... nothing stopping anyone from designing a updated MIPS core either.

You have a point though... if you let one ISA rule it all you could just end up with the next Intel. It would be a complete 180 from what they are currently doing however. The core of what has made ARM successful, has been their ability to turn out new modern ARM core designs every couple years and licence them to anyone. That is what has kept MIPS or RISC-V from really being more popular with hardware manufacturers, MIPS didn't keep up on the design front, and the only real RISC-V designs are university student level projects. In the end I trust ARM more then I trust Intel. Intel has been able to keep CPUs nice and lucrative... as they have been able to keep all but a handful from having the ability to make compatible chips. AMD doesn't really have the funds to low ball them all that hard after years of Intel squeezing.

Hopefully ARM sticks to the business model that got them where they are now... at least if they don't it only gets easier and easier to hop ISAs if need be. Especially when you consider in 15 years there is a very real reason that a major majority of stuff will be coming out of China, where they may decide to push some home grown MIPS like ISA anyway.
 

Zarathustra[H]

Fully [H]
Joined
Oct 29, 2000
Messages
30,450
ISA matters a lot of. Why do you believe Intel is inventing new ISAs like AVX and AMD using them?
Maybe I should have been clearer. I thought in context it was clear that I was specifically referring to power efficiency.

In the testing per that test, when calculating the energy necessary to complete a task (power use during run x time of run) there were some differences between the different CPU's, but they did not exhibvit a trend tied to which instruction set the chip was using.

There is this popular belief out there that ARM is more power efficient by its very nature, and that just seems to be false.
 

Zarathustra[H]

Fully [H]
Joined
Oct 29, 2000
Messages
30,450
Android has in the past supported x86 and MIPS... nothing stopping anyone from designing a updated MIPS core either.
MIPS and RISC-V are very closely related. Might not be too difficult to cross compile between them.

MIPS forked off from the original first gen RISC design, and has been developed some since then. RISC-V is the fifth gen of that design.

You have a point though... if you let one ISA rule it all you could just end up with the next Intel. It would be a complete 180 from what they are currently doing however. The core of what has made ARM successful, has been their ability to turn out new modern ARM core designs every couple years and licence them to anyone. That is what has kept MIPS or RISC-V from really being more popular with hardware manufacturers, MIPS didn't keep up on the design front, and the only real RISC-V designs are university student level projects. In the end I trust ARM more then I trust Intel. Intel has been able to keep CPUs nice and lucrative... as they have been able to keep all but a handful from having the ability to make compatible chips. AMD doesn't really have the funds to low ball them all that hard after years of Intel squeezing.
I guess I just don't trust any for-profit organization. ARM's interest right now is to expand as much as possible into new architectures. Once they've cemented that position I could easily see them try to pull a 3DFX when they bought STB (albeit more successfully). Buy out Ampere or some other smaller ARM chip manufacturer, gradually strangle everyone else out of licensing, and now control the ARM CPU market.

Hopefully ARM sticks to the business model that got them where they are now... at least if they don't it only gets easier and easier to hop ISAs if need be. Especially when you consider in 15 years there is a very real reason that a major majority of stuff will be coming out of China, where they may decide to push some home grown MIPS like ISA anyway.
Yeah, that's a real alarming future.
 

ChadD

Supreme [H]ardness
Joined
Feb 8, 2016
Messages
4,700
What do you mean by "powered in part"?

Fugaku is not the first time an ARM computer enter in the list.

https://www.top500.org/system/179565/
Its the first time they score a First place.

So far #1 and fastest computer in the world titles have been held by Alpha / PowerPC / a Custom Chinese Alpha variant / x86 (for a very short time).... and now ARM.

So no more can people say ARM is not a capable of high performance computing. It would seem Fujitsu just proved ARM can in fact be very high performance. Cray has also started selling machines based on the same Fujitsu chip... now that Fugaku has become the #1 machine in terms of power, AND the #1 in terms of performance per watt. Expect Cray to pick up some new and probably very interesting wins for ARM machines.

No doubt Super computer stuff doesn't filter down to normal servers and the like all that much... but more wins in this space will mean more software tools for high performance ARM computation. Which absolutely will make it far easier for some other company to licence the same ARM ISA and build out something interesting for server markets... and at some point high end workstations as well in all likely hood.

Fugaku taking #1 makes me feel much better about my 5 years of x86 left prediction. Looks like x86 is being attacked from a lot of fronts right now.
 

Zarathustra[H]

Fully [H]
Joined
Oct 29, 2000
Messages
30,450
Maybe I should have been clearer. I thought in context it was clear that I was specifically referring to power efficiency.

In the testing per that test, when calculating the energy necessary to complete a task (power use during run x time of run) there were some differences between the different CPU's, but they did not exhibvit a trend tied to which instruction set the chip was using.

There is this popular belief out there that ARM is more power efficient by its very nature, and that just seems to be false.
MIPS and RISC-V are very closely related. Might not be too difficult to cross compile between them.

MIPS forked off from the original first gen RISC design, and has been developed some since then. RISC-V is the fifth gen of that design.



I guess I just don't trust any for-profit organization. ARM's interest right now is to expand as much as possible into new architectures. Once they've cemented that position I could easily see them try to pull a 3DFX when they bought STB (albeit more successfully). Buy out Ampere or some other smaller ARM chip manufacturer, gradually strangle everyone else out of licensing, and now control the ARM CPU market.



Yeah, that's a real alarming future.

Lol. I totally did not know that ARM holdings had invested an unknown amount into Ampere when I posted the above..

https://www.electronicsweekly.com/news/business/arm-invests-ampere-2019-04/
 

juanrga

Pro-Intel / Anti-AMD Just FYI
Joined
Feb 22, 2017
Messages
2,642
Maybe I should have been clearer. I thought in context it was clear that I was specifically referring to power efficiency.

In the testing per that test, when calculating the energy necessary to complete a task (power use during run x time of run) there were some differences between the different CPU's, but they did not exhibvit a trend tied to which instruction set the chip was using.

There is this popular belief out there that ARM is more power efficient by its very nature, and that just seems to be false.
AVX is designed just for that. Similar computation using x86 instructions would consume much more power than using AVX instructions.

I believe that you are referring to a certain wrong paper comparing x86 to ARM from a University team loosely linked to Intel. ;)

The advantages of the ARM ISA aren't a belief. One can see them in the design of the ISA. Jim Keller even gave a talk about how ARM64 is better than x64.
 

juanrga

Pro-Intel / Anti-AMD Just FYI
Joined
Feb 22, 2017
Messages
2,642
Its the first time they score a First place.

So far #1 and fastest computer in the world titles have been held by Alpha / PowerPC / a Custom Chinese Alpha variant / x86 (for a very short time).... and now ARM.

So no more can people say ARM is not a capable of high performance computing. It would seem Fujitsu just proved ARM can in fact be very high performance.
The guy whom I replied wrote "the first time Arm has been on the list". He didn't say first place, he said "on the list". That ARM is capable of high performance computing was proved before by Cavium/Marwell, whose TX2 was able to beat the fastest Skylake and EPYC CPUs in HPC code.
 

Zarathustra[H]

Fully [H]
Joined
Oct 29, 2000
Messages
30,450
AVX is designed just for that. Similar computation using x86 instructions would consume much more power than using AVX instructions.

I believe that you are referring to a certain wrong paper comparing x86 to ARM from a University team loosely linked to Intel. ;)

The advantages of the ARM ISA aren't a belief. One can see them in the design of the ISA. Jim Keller even gave a talk about how ARM64 is better than x64.
Nah, I am referring to this article which I forgot to link in my original post above.

It's a few years old now, but they compared:
- Intel Sandy Bridge (C2700)
- AMD Bobcat (Zacate E-240)
- Intel Atom (N450)
- ARM Coretex A15 MPcore
- ARM Coretex A9 (OMAP4430)
- ARM Coretex A8 (OMAP3530)
- Longsoon STLS2F01

The interesting chart is Fig10 as it is the best measure of the actual energy used to perform a given task. Some architectures fare better than others, but ISA is not the defining characteristic. IN some tasks Intel's X86 chips do the best, in others ARM do the best, but in no case can you predict performance based on which ISA the chip is using.

Their conclusion is that ISA certainly matters on some very small 1-2mm^2 or sub milliwatt designs where RISC designs really pay off, but outside of that the architecture of the chip is much more important than the ISA.
 

ChadD

Supreme [H]ardness
Joined
Feb 8, 2016
Messages
4,700
The guy whom I replied wrote "the first time Arm has been on the list". He didn't say first place, he said "on the list". That ARM is capable of high performance computing was proved before by Cavium/Marwell, whose TX2 was able to beat the fastest Skylake and EPYC CPUs in HPC code.
I agree ARM has been shown to perform a few times... but those earlier chips always fizzled out when compared to the competition. They showed promise and by the time they got to production they where no longer all that attractive. This is the first time where all that promise shown in early leaks and tests... has actually paid off with a official Fastest around tag. Granted there are exascale x86 machines coming online next year... still for now the fastest official computer in the world is ARM powered.

https://www.top500.org/lists/top500/2020/06/

The new top system, Fugaku, turned in a High Performance Linpack (HPL) result of 415.5 petaflops, besting the now second-place Summit system by a factor of 2.8x. Fugaku, is powered by Fujitsu’s 48-core A64FX SoC, becoming the first number one system on the list to be powered by ARM processors. In single or further reduced precision, which are often used in machine learning and AI applications, Fugaku’s peak performance is over 1,000 petaflops (1 exaflops). The new system is installed at RIKEN Center for Computational Science (R-CCS) in Kobe, Japan.
 
Last edited:

Zarathustra[H]

Fully [H]
Joined
Oct 29, 2000
Messages
30,450
I agree ARM has been shown to perform a few times... but those earlier chips always fizzled out when compared to the competition. They showed promise and by the time they got to production they where no longer all that attractive. This is the first time where all that promise shown in early leaks and tests... has actually paid off with a official Fastest around tag. Granted there are exascale x86 machines coming online next year... still for now the fastest official computer in the world is ARM powered.

https://www.top500.org/lists/top500/2020/06/

The new top system, Fugaku, turned in a High Performance Linpack (HPL) result of 415.5 petaflops, besting the now second-place Summit system by a factor of 2.8x. Fugaku, is powered by Fujitsu’s 48-core A64FX SoC, becoming the first number one system on the list to be powered by ARM processors. In single or further reduced precision, which are often used in machine learning and AI applications, Fugaku’s peak performance is over 1,000 petaflops (1 exaflops). The new system is installed at RIKEN Center for Computational Science (R-CCS) in Kobe, Japan.
Fair, but for Supercomputer types of highly parallelized loads the sky is really the limit. You can just keep adding CPU's of any architecture until you get to the performance level you desire.
 
Top