I do video training for Autodesk Revit, at least until there is a real, MODERN BIM tool that is Mac native, and not a second longer! ;-)
I often wonder why Autodesk doesn't port some of their major titles to Mac.
I suppose the AutoCAD community is entrenched in Windows and offering a Mac version probably wouldn't be particularly profitable, but 3DS Max? Wouldn't the people who use that rather use OS X?
Apple is iterating through new ARM CPU upgrades faster than intel: intel's tic-tok model is a two year cycle and they are falling behind that, which is why new Mac announcements are stalled and overdue: the relevant Intel CPU upgrades aren't available.....
Quote:
Originally Posted by rcfa
.....Parallels can license/develop Rosetta-type technology and Windows compatibility won't be an issue, and who knows, maybe Apple and M$ have a deal for a full desktop-class ARM based windows, which would also help M$' tablet/laptop hybrids.
The technology is there, so it's just a matter of marketing.
It would also be easy to use ARM in the consumer space and for server products, and use x86 or an x86/ARM combo for pro laptops and desktops. Apple already has multiple CPUs in the system so this is totally doable and allows Apple to go places where that competition can't easily follow.
Not remotely likely on the MS/Apple front (see more below)
Also, Intel's delays are mostly (certainly somewhat) related to fabbing up for the next round of process/die shrinks (28, 22, 14 nm)... ...and no one spends more on this tech than Intel, so unlikely Apple's fabs could keep advancing any faster.
Quote:
Originally Posted by iMat
....So, if Apple uses ARM for their Mac line. What happens? They will:
a) have to move the entire Mac line to ARM (from the Air to the Pro) or
b) have OSX run on two different architectures simultaneously. Which is a complete mess.
If Apple moves to their own CPU (sort of like the A series) then they will have to constantly upgrade and develop an entire CPU/GPU system just for their Mac line, which doesn't sell enough to support such an investment.
So all in all.
No. Just no.
It doesn't make any sense now to switch from Intel to ARM (in my opinion) because ARM is pushing slowly into "computer" territory and doesn't offer a wide enough array of CPUs (to my knowledge) to cover the Mac line from "Mini" to "Pro".
An A series targeted at notebooks/desktops does seem a big ongoing engineering effort for product lines that hold a static shrinking market share of computing devices - and Apple's left building servers (another ARM target as noted by rcfa) behind for some years, so don't see them going back there.
On the other hand, there's some indication that tablet sales growth, at least in the consumer sphere, may have a lower ceiling than PC's historically had as many to most are finding a phone (or phablet) sufficient for ALL their needs. We're not all prolific content creators, writers, number crunchers, etc.
Another factor is Intel's massive investment in Atom processors. Playing from behind, their cumulative effort could at least catch up and make the full X-86 instruction set available at ARM power consumption levels. Windows 8 laptops running Baytrail are already out and reviewers note they're plenty fast enough.
So might we see an Atom Mac.....??
Quote:
Originally Posted by delreyjones
....A "complete mess"? Is that possibly just a bit of hyperbole? The original Mac OS ran simultaneously on Motorola and PPC for a while. OSX ran on PPC and Intel. And since half of OSX is shared with IOS, we can say that today, much of the current OS is already shared by two architectures. I can see that if we asked Microsoft to do this it my devolve into your "complete mess", but I think the engineers at Apple are pretty organized.
MS already engaged in this "devolution" with Win RT, a colossal flop that's now being merged with WinPhone into likely some "tablet-esque" version of WinPhone 8 (or 9) - that won't have a "desktop stub" to run PC office.
However, MS can clearly build a Win that runs on ARM (tho' the market for Win on Mac seems far too small to justify the software engineering). And the phones are super-snappy in my limited playing around, if nothing else...
Quote:
Originally Posted by rcfa
An A8 may already have closed the performance gap, and given the low power consumption of ARM chips, you can double or triple the CPU core count compared to x86 CPUs meaning anything that can run in parallel will be significantly FASTER than on equivalent x86 platforms with similar power consumption.
Also, while CPU emulation is a factor of two-four slower than native execution, the same does NOT apply to Rosetta-style on-the-fly recompilation, which has only a moderate overhead of maybe 25% or less, IIRC, and which in some cases even ran code faster than on the native CPU.
You'd first have to close the "coding for parallel processing" bottleneck. My understanding's that software developers by and large still aren't taking advantage of the multi-threading and cores available today.
And even if the OS can split up Apps and higher level chunks to use 'em between processes, they can do little (or nothing) about the code running within 'em to take advantage.
[Said the layman who's neither engineer or coder... ...so correct away if I'm wrong.]
Maybe Apple will test the market by releasing 2 versions of mbp or imacs: the current Intel based and the new ARM based.
That didn't work out particularly well for Microsoft with the Surface and Windows RT... I don't know if consumers are savvy enough to understand that an ARM/iOS version isn't just a less expensive version of the Intel machine.
That didn't work out particularly well for Microsoft with the Surface and Windows RT... I don't know if consumers are savvy enough to understand that an ARM/iOS version isn't just a less expensive version of the Intel machine.
Microsoft launched Window RT / Surface RT a year before Intel became a viable option for low price tablets.
There are still IT departments out there, that refuse to support Mac. Anyone in the corporate world knows this. At this time, either remotely or via Parallels etc.
So what you are saying is that the processor doesn't make a difference.
To give this up, would be not only irritating, but for me, a terrible choice. How many remember not being able to use PowerPC native apps when the OS switches took place? This would be equally as bad.
Yeah but we really don't know what the capabilities of the system would be.
It would cause me to not buy a new mac, for the first time in over 25 years.
Honestly you sound just like the CP/M and Apple 2 guys complaining about the new IBM hardware and then the new Mac. Times change and you either get left in the dust or you work to keep up.
There is something significant here. as you point out it wouldn't be about the use of an ARM SoC if they went that route. It would be about enabling a design that wouldn't otherwise be possible. That would just be one part of it rather than a fad element.
Going by i5 and i7 is a terrible way to disambiguate things. Blame intel for that. They don't guarantee anything, because what they indicate changes based on the class of the cpus in question. For example dual core i5s, which are restricted to notebook chips, have a total of 4 logical cores due to hyperthreading. In chips binned as desktop class, mainstream, or whatever they decide to call it, there are quad i5s. They still have 4 logical cores, but these have hyperthreading disabled. The 4 logical cores are all physical cores. In terms of actual performance differences between i5 and i7, that also varies in both benchmarks and real applications.
Or you could simply say intel is dishonest in their marketing. I find their current naming conventions to be extremely confusing and without a product map in front of you it is hard to tell what you are actually getting. Even AMD does better with the naming of generations of processors.
Windows is dying. Windows 8 and the iPad have been terminal cancer for windows.
OSX and iOS are unix based - inherently multiplatform. All apple has to do is release a new version of Xcode and OSX & apps will run just fine on ARM.
Apple's done this three times now: 68k to ppc, ppc to x86 and x86 to ARM.
Added bonus: all iOS apps will also run under OSX. They already do this under Xcode in emulation but the Ax ARM processors will run iOS apps natively.
Actually those iOS programs get compiled to i86 and then run in a simulator. A slight but important difference. This is why I like the idea of an ARM based Mac OS that supports running iPad apps in a window. The infrastructure is already there and jut needs to be tidied up and made transparent to the users.
forget parallels. Remember how bad windows used to run on the 680x0 and PPC? It sucked - poky and slow.
Apple doesn't have to care about windows anymore. Their margins will skyrocket when they stop buying processors from Intel.
Well it is obviously important to some. However I think we are well past the point where it is mandatory to do so. More so if people have a choice between i86 Macs and ARM Macs they could make the choice they need to make.
Also by not supporting the wintel cadre, they will help kill wintel once and for all.
Microsoft is in total disarray. Their dev tools for mobile suck.
Imagine touchscreen MacBook Airs that run everything (OSX &iOS) except windows for the price of iPads.
No, actually, I'm complaining that applications we rely on, may not be supported for some time, by the software vendors. Similar to what happened with some of the Adobe apps, and what happened to Quicken, when the OS could no longer support PowerPC apps. What were we told in that year or two before they re-supported (or rather supported) the newest OS10? The vendors told us to switch to windows. Doubt that would happen these days. However, we are stuck (for good or bad) using what works, until the rest of the world jumps forward to support any new OS based on ARM. Now, on the other hand, if no "PC" emulation or connection is supported under an Arm processor, us "corporate" types will simply not be able to use a new Mac, until a workaround comes about. Could take years, and thats lost productivity, if nothing works. Thats why we would have to stick with old technology until things fall in place.
The reason they switched from PowerPC to Intel was that the CPUs were becoming a limiting factor in making devices smaller and thinner (i.e. the Powerbook G5 that never happened). And certainly a switch from Intel to ARM could be justified on the same grounds. The thing I don't believe about the report is that Macs would have 4 or 8 of these CPUs. Surely that would require some nasty motherboard logic, with 8 CPUs all access the same memory? Doesn't sound believable.
In terms of having a Rosetta type solution on the ARM Macs to let them run old Intel apps, I don't think they should do that. Rather they should do the opposite, and have a Rosetta solution on Intel Macs to let them run ARM apps go forward. The ARM Macs should take on the policy of the iDevices, and only be allowed to run apps from the Mac App Store. And the only apps that should show up for downloading are ones that developer have uploaded as pure ARM or ARM/Intel fat binaries.
One out-there thing I wonder is whether Apple could possibly do static in-place Intel to ARM translation to the apps on the Mac App Store. Would that even be possible, or are there some things that simply can not be done until runtime?
Anyway I would be first in line to buy one of these new Macs. But mainly for software reasons not hardware. Switching to a whole new architecture is a golden opportunity to get rid of a lot of cruft and declare a lot of old frameworks deprecated and will not be supplied in ARM versions.
I often wonder why Autodesk doesn't port some of their major titles to Mac.
I suppose the AutoCAD community is entrenched in Windows and offering a Mac version probably wouldn't be particularly profitable, but 3DS Max? Wouldn't the people who use that rather use OS X?
Most of the film and TV stuff from Autodesk has Mac versions already, because they where all Mac first stuff I think, and Autodesk just bought them, and continues to develop for that entrenched market. AutoCAD actually did get a Mac port a few years ago, and my understanding is it is actually a really good translation of (most) AutoCAD functionality to a familiar but still very Mac UI. 3DS and Revit likely never will because the code base is such crap it would require a ground up rewrite, and Autodesk doesn't do that usually. Acad had a much cleaner codebase due to the major rewrite it got around the turn of the millennium. In general Autodesk is not even a Fast Follower, so much as a JIT Buyer. They let others actually think big thoughts and lead, and when something is beating them in the market they buy it and milk it till the next better idea comes along.
As for 3DS users preferring Mac, I really doubt it. Anyone who likes the craptastic UI and workflow that is 3DS probably doesn't even notice that Mac is elegant and easy.
Why wouldn't they make the move to ARM? Clearly they have plans to do so otherwise they wouldn't have machines already running in the labs. It's just a matter of time, I'd stake money on it happening.
I have no doubt they have experiments in the labs, but there is absolutely no way such a device would sell except to a power-conscious "green" customer. No Desktop will ever sell like this except maybe macmini "servers" relegated to NAS/Servers. An iMac would certainly fail, and a Mac Pro would be dead before it hit the store.
This is not the same argument as the PowerPC problem. With PowerPC, IBM was the sole CPU supplier to Apple, and IBM was much more interested in getting Nintendo/Sony/Microsoft to make consoles using their IP than they were building CPU's that could be dropped into a desktop or laptop. Part of the reason the PowerPC platform held on as long as it did was because those same parts were also being used for the Commodore Amiga. IBM makes servers with the POWER parts now. Sticking a server cpu in a laptop is a joke, and Apple knew it.
ARM's strength, and only strength is in power to performance ratio. Intel is terrible at this (Note how the average intel desktop is a 95 TDP watt target, not something reasonable like 30 watts, and the only parts that actually achieve low-power, are heavily neutered i3 and "Celeron" Atom type of parts which are an insult to use.) 8 quad core ARM parts may bring the performance of a ARM based desktop to that of a Intel based desktop, but I dare you to find any software that has been compiled to magically use as many threads as possible. No UNIX-land still programs like it's 1979 (and many windows apps too, just look at Google Chrome,) forking processes, not threading. It's pretty useless as a desktop when one ARM cpu core is the same power as a 15 year old intel CPU. if you had 4x8 cores (eg 32) threads, 31 of those threads would be idle, just like how 7 intel cores are idle on a i7 hyperthreaded CPU, and those other CPU cores are only turned on when a process is explicitly thread-aware (eg x264)
No, more cores doesn't work unless the software out there is rewritten for it, and only Apple's OS and Apple's software will ever do that. Adobe has enough problems just getting their flash player to synchronize across cpu cores. If a large company can utterly fail on their flagship products, that doesn't bode well for everyone else.
Most software that isn't outright number-crunching software (eg SETI@home) just fails to take advantage of multiple cores. It's just "too hard" for developers to not screw up. I've played countless games, and all of them still do everything single-threaded, and when they do take advantage of threads, it ultimately is too little. Dedicated consoles have a minor advantage here because games designed for them know exactly what they will have, while the average desktop or laptop can't be relied on having more than one cpu core.
This is why every time I see a story about "switching to ARM" in the context of desktops I groan and don't see it happening. At best/worst Apple is keeping their options open for a time when the ARM parts actually have comparable single thread performance. That is not this year, and not likely until at least 2016 when Intel runs out of die shrinks.
That didn't work out particularly well for Microsoft with the Surface and Windows RT... I don't know if consumers are savvy enough to understand that an ARM/iOS version isn't just a less expensive version of the Intel machine.
I'll blame Microsoft for just screwing up the marketing. the RT was a device looking to solve a problem that didn't exist. It was trying to run "full windows" on hardware that would normally be tablet hardware. The Surface Pro on the other hand is a laptop in a tablet form factor. The latter is utterly incapable of doing anything CPU intensive because Intel parts are bad (AMD are worse) in the side of device. The Surface Pro 3 just proves this point that Microsoft is trying to make a tablet using laptop parts, and doesn't give a care about the parts lasting very long. They removed the Wacom parts and replaced with inferior nTrig just to make it lighter. Let's not even get started on WindowsCE/Mobile/Phone
See in the Apple world, the difference between OSX and iOS is some UI API layer and a recompile. On Windows, you can't do this, because recompiling a windows app designed to use a two-button mouse doesn't work on a tablet that doesn't have a right-click button. Apple software however has always been able to work without a right-click button, as they traditionally only ever had a 1-button mouse.
Apple could switch to ARM for their desktops and laptops but that would be forgetting the mistake Microsoft already made too quickly. Software has to specifically be designed to take advantage of the many lower-performance cores, and most software aside from bechmarking software itself doesn't lend well to that.
I often wonder why Autodesk doesn't port some of their major titles to Mac.
Good question! It might be due to poor sales of AutoCAD on the Mac, or it could be code bases that don't easily port.
The poor sales of things like AutoCAD are likely due to the software costing way too much for what you get vs what people need. Let's face it CAD was modern technology in the 70's & 80's, these days it isn't anymore difficult than a modern word processor. So if an excellent word processor can be had for $300 why not a CAD program. It should be noted that Dassault offers its basic CAD system for free.
I suppose the AutoCAD community is entrenched in Windows and offering a Mac version probably wouldn't be particularly profitable, but 3DS Max? Wouldn't the people who use that rather use OS X?
I'm not sure what the problem is. I'm left with the impression that AutoDesk is on greedy company.
The reason they switched from PowerPC to Intel was that the CPUs were becoming a limiting factor in making devices smaller and thinner (i.e. the Powerbook G5 that never happened). And certainly a switch from Intel to ARM could be justified on the same grounds.
Basically the PPC consortium fell apart.
However, as far as ARM goes this would not be a justification for going to ARM. The best reason is that Apple would have no choice if it wants to continue to innovate in the desktop/laptop computer space. This is the same reason they had to start designing their own cell phone chips, in a nut shell silicon is the printed circuit board of the last century. In order to innovate in PC space you either need to design your own chips or get Intel to truly go the custom route. Intel seem highly resistant to custom chips so in the end it is either AMD or ARM.
By the way it seems like AMD has seen the writing on the wall here and is very willing to do custom SoC. In the end the business of computers is changing for the same reason the business of cell phones is changing. It is now possible to do SoC that are truly deserving of the SoC label.
The thing I don't believe about the report is that Macs would have 4 or 8 of these CPUs. Surely that would require some nasty motherboard logic, with 8 CPUs all access the same memory? Doesn't sound believable.
Not believable at all actually.
In terms of having a Rosetta type solution on the ARM Macs to let them run old Intel apps, I don't think they should do that. Rather they should do the opposite, and have a Rosetta solution on Intel Macs to let them run ARM apps go forward.
How about neither?
The ARM Macs should take on the policy of the iDevices, and only be allowed to run apps from the Mac App Store.
If they did this I wouldn't be able to support them by buying the hardware. Right now the only reason I have to run a Mac is to run apps not available on the App Store. Apps that will likely never be on the App Store. It would kill the platform for more advanced users of the machines.
And the only apps that should show up for downloading are ones that developer have uploaded as pure ARM or ARM/Intel fat binaries.
Given the above, the App Store is key to getting this to work. I could see Apple giving developers an ultimatum, either have the ARM binaries by XYZ date or your app gets dropped from the store. I can see most developers jumping on the bandwagon due to Mac OS being the only real growth opportunity out there for PC type apps.
One out-there thing I wonder is whether Apple could possibly do static in-place Intel to ARM translation to the apps on the Mac App Store. Would that even be possible, or are there some things that simply can not be done until runtime?
It is possible. What would be more likely is that they leverage LLVM and require apps be built to one of LLVM's intermediate forms and then have LLVM generate the native code on the target machine. It is an option different than fat binaries but a workable one considering how far LLVM has come in the last few years. Such an OS would never be tied to hardware in the future.
Anyway I would be first in line to buy one of these new Macs.
Yep! Given that I have the same access as I currently do with Mac OS, that is the ability to install third party software and run scripts and such then I'd be all in. I just don't run Windows software anymore. Ideally the platform would also be open enough to run Linux also.
But mainly for software reasons not hardware.
Well it is the whole deal really. For me a whole bunch of low power cores would work out far better than the current two core solutions found in most low end laptops. With the ARM cores I can see Apple putting eight cores into a SoC by the end of 2015.
Switching to a whole new architecture is a golden opportunity to get rid of a lot of cruft and declare a lot of old frameworks deprecated and will not be supplied in ARM versions.
Possibly but Mac OS has been overhauled fairly completely now. My goal or hope would be that we would get a low cost platform, with all of those cores, that is significantly better than the Intel offerings in the same price range. Better means performance that holds up under load, power usage that is absolutely minimal under load and also when sleeping, and finally cool operation.
By the way, I realize that the ARM cores Apple would lead with this year won't be the fastest in the world. That isn't as important has getting access to lots of them.
I'll blame Microsoft for just screwing up the marketing. the RT was a device looking to solve a problem that didn't exist. It was trying to run "full windows" on hardware that would normally be tablet hardware. The Surface Pro on the other hand is a laptop in a tablet form factor. The latter is utterly incapable of doing anything CPU intensive because Intel parts are bad (AMD are worse) in the side of device.
It is more about the OS sucking to me. Windows has a very negative image with the population at large. That opinion was just confirmed when the executives at MS couldn't even get through the introduction without the device crashing.
The Surface Pro 3 just proves this point that Microsoft is trying to make a tablet using laptop parts, and doesn't give a care about the parts lasting very long. They removed the Wacom parts and replaced with inferior nTrig just to make it lighter. Let's not even get started on WindowsCE/Mobile/Phone
See in the Apple world, the difference between OSX and iOS is some UI API layer and a recompile. On Windows, you can't do this, because recompiling a windows app designed to use a two-button mouse doesn't work on a tablet that doesn't have a right-click button. Apple software however has always been able to work without a right-click button, as they traditionally only ever had a 1-button mouse.
The difference between iOS and OSX are large enough that more than a recompile is needed. However people mis how much is actually shared between the two platforms.
Apple could switch to ARM for their desktops and laptops but that would be forgetting the mistake Microsoft already made too quickly. Software has to specifically be designed to take advantage of the many lower-performance cores, and most software aside from bechmarking software itself doesn't lend well to that.
That is unadulterated bull crap. Most software running on the Mac these days is threaded and does leverage cores. Beyond that if you run more than one app at a time those cores become very handy indeed. People seem to not remember anything around here but a significant number of apps got significant performance boosts when Apple released GCD (Grand Central Dispatch) a few years ago. GCD is just one way Apps can leverage cores. Sure you can go out and find a few apps that make use of no threading at all, but they are damn few these days.
I have no doubt they have experiments in the labs, but there is absolutely no way such a device would sell except to a power-conscious "green" customer. No Desktop will ever sell like this except maybe macmini "servers" relegated to NAS/Servers. An iMac would certainly fail, and a Mac Pro would be dead before it hit the store.
The funny thing here is that the Mini or a replacement for the Mini would be an ideal location for ARM based chips. You could leave it turned on and with it in sleep mode most of the time your electric bill does not get hammered. It isn't being green it is all about economics.
As for the Mac Pro nobody is talking about an ARM based Mac Pro at the moment. A laptop however would be great for the same reasons a Mini would be.
This is not the same argument as the PowerPC problem. With PowerPC, IBM was the sole CPU supplier to Apple, and IBM was much more interested in getting Nintendo/Sony/Microsoft to make consoles using their IP than they were building CPU's that could be dropped into a desktop or laptop. Part of the reason the PowerPC platform held on as long as it did was because those same parts were also being used for the Commodore Amiga. IBM makes servers with the POWER parts now. Sticking a server cpu in a laptop is a joke, and Apple knew it.
Well we can speculate all we want here but in the end IBM blew it.
ARM's strength, and only strength is in power to performance ratio. Intel is terrible at this (Note how the average intel desktop is a 95 TDP watt target, not something reasonable like 30 watts, and the only parts that actually achieve low-power, are heavily neutered i3 and "Celeron" Atom type of parts which are an insult to use.)
Actually by some measures Haswell does very well on performance per watt. The problem is it isn't a low power part by any measure even in Intels best offerings. As to ATOM, ATOM proves that Intel just doesn't understand the SoC market.
8 quad core ARM parts may bring the performance of a ARM based desktop to that of a Intel based desktop, but I dare you to find any software that has been compiled to magically use as many threads as possible.
Just about every app on the Mac these days uses threads in one form or another. There are a few that don't but it is damn few.
No UNIX-land still programs like it's 1979 (and many windows apps too, just look at Google Chrome,) forking processes, not threading.
There are several ways to make use of cores in Mac OS based machines, running separate processes are one of them. Just because a technique has been around for a long time doesn't invalidate its usefulness. The reality is an app can at times make use of multiple techniques to leverage cores such as starting processes that use threads.
It's pretty useless as a desktop when one ARM cpu core is the same power as a 15 year old intel CPU. if you had 4x8 cores (eg 32) threads, 31 of those threads would be idle, just like how 7 intel cores are idle on a i7 hyperthreaded CPU, and those other CPU cores are only turned on when a process is explicitly thread-aware (eg x264)
No, more cores doesn't work unless the software out there is rewritten for it, and only Apple's OS and Apple's software will ever do that.
Baloney! Use some real apps and you will most impressed when you have lots of cores to leverage. Beyond that have you ever once ran more than one app at a time. Maybe updating from App Store while surfing the net, cores help when you do actually use the machine as multitasking machines.
Adobe has enough problems just getting their flash player to synchronize across cpu cores. If a large company can utterly fail on their flagship products, that doesn't bode well for everyone else.
Flash is so thing that has been crap for years so I'm not sure why you even reference it.
Most software that isn't outright number-crunching software (eg SETI@home) just fails to take advantage of multiple cores.
Again baloney. Come up with a list of apps that absolutely never use multiple cores for us. Some aren't well threaded of not threaded at all but the vast majority of apps these days employ at least some threading.
It's just "too hard" for developers to not screw up. I've played countless games, and all of them still do everything single-threaded, and when they do take advantage of threads, it ultimately is too little.
You do realize that some of us use our machines for something besides gaming right? Besides you just have damned you anti threading flame here by admitting that even some games are threaded.
Dedicated consoles have a minor advantage here because games designed for them know exactly what they will have, while the average desktop or laptop can't be relied on having more than one cpu core.
Thinking from the past, most hardware of the last couple of years has been at a minimal dual core.
This is why every time I see a story about "switching to ARM" in the context of desktops I groan and don't see it happening.
It is pretty clear from this post that you don't have a clue and have been woefully misinformed about the value of cores. However I'm pretty much convinced that cores per say are not what Apple is interested in but rather it is the freedom to be able to innovate in the PC arena into the future. It is pretty much the same logic that forced them to develop their own cell phone processors, SoC are the fur ute and that is where your innovation will go.
At best/worst Apple is keeping their options open for a time when the ARM parts actually have comparable single thread performance. That is not this year, and not likely until at least 2016 when Intel runs out of die shrinks.
Apples ARM parts really aren't that bad. Give those cores faster RAM, more cache and similar clock speeds and you might be surprised by their performance. Will they be as fast as Intels fastest cores of the moment, of course not, but Apple doesn't use Intels fastest chips in the Air nor the Mini. The point is even today's A7 core, in the right configuration, would produce a machine with a very interesting performance profile. That doesn't even consider the possibility of a new core beyond Cyclone.
Honestly you seem to be obsessed with single core performance, possible because of your gaming interests, but that is seldom important when it comes to modern computer usage. Crank up apps like XCode, Eclipse Safari, Mail, office, Numbers, Pages, keynote and any other common user app and convince me that they don't use threads at all. This doesn't even consider how many processes are active at anyone time on a machine. Cores are very important when it comes to machine responsiveness.
Comments
Perhaps it isn't Macs they are looking at.
New products for new markets or niches.
Lot's of other variety of apples to use for names.
Some apple varieties and possible products.
Beacon - tracking device
Beautiful Arcade - gaming computer
Pilot - navigation aid
Revival - CPR device
And the Yellow Transparent apple is also known as the Glass apple.
Has to be something with the Apple stores.
Maybe Apple will test the market by releasing 2 versions of mbp or imacs: the current Intel based and the new ARM based.
...
Imagine touchscreen MacBook Airs that run everything (OSX &iOS) except windows for the price of iPads.
If Apple can really sell MacBook Air Touch at the IPAD PRICE, MS won't exist anymore.
I do video training for Autodesk Revit, at least until there is a real, MODERN BIM tool that is Mac native, and not a second longer! ;-)
I often wonder why Autodesk doesn't port some of their major titles to Mac.
I suppose the AutoCAD community is entrenched in Windows and offering a Mac version probably wouldn't be particularly profitable, but 3DS Max? Wouldn't the people who use that rather use OS X?
Apple is iterating through new ARM CPU upgrades faster than intel: intel's tic-tok model is a two year cycle and they are falling behind that, which is why new Mac announcements are stalled and overdue: the relevant Intel CPU upgrades aren't available.....
Quote:
It would also be easy to use ARM in the consumer space and for server products, and use x86 or an x86/ARM combo for pro laptops and desktops. Apple already has multiple CPUs in the system so this is totally doable and allows Apple to go places where that competition can't easily follow.
Not remotely likely on the MS/Apple front (see more below)
Also, Intel's delays are mostly (certainly somewhat) related to fabbing up for the next round of process/die shrinks (28, 22, 14 nm)... ...and no one spends more on this tech than Intel, so unlikely Apple's fabs could keep advancing any faster.
Quote:....So, if Apple uses ARM for their Mac line. What happens? They will:
a) have to move the entire Mac line to ARM (from the Air to the Pro) or
b) have OSX run on two different architectures simultaneously. Which is a complete mess.
If Apple moves to their own CPU (sort of like the A series) then they will have to constantly upgrade and develop an entire CPU/GPU system just for their Mac line, which doesn't sell enough to support such an investment.
So all in all.
No. Just no.
It doesn't make any sense now to switch from Intel to ARM (in my opinion) because ARM is pushing slowly into "computer" territory and doesn't offer a wide enough array of CPUs (to my knowledge) to cover the Mac line from "Mini" to "Pro".
An A series targeted at notebooks/desktops does seem a big ongoing engineering effort for product lines that hold a static shrinking market share of computing devices - and Apple's left building servers (another ARM target as noted by rcfa) behind for some years, so don't see them going back there.
On the other hand, there's some indication that tablet sales growth, at least in the consumer sphere, may have a lower ceiling than PC's historically had as many to most are finding a phone (or phablet) sufficient for ALL their needs. We're not all prolific content creators, writers, number crunchers, etc.
Another factor is Intel's massive investment in Atom processors. Playing from behind, their cumulative effort could at least catch up and make the full X-86 instruction set available at ARM power consumption levels. Windows 8 laptops running Baytrail are already out and reviewers note they're plenty fast enough.
So might we see an Atom Mac.....??
Quote:
MS already engaged in this "devolution" with Win RT, a colossal flop that's now being merged with WinPhone into likely some "tablet-esque" version of WinPhone 8 (or 9) - that won't have a "desktop stub" to run PC office.
However, MS can clearly build a Win that runs on ARM (tho' the market for Win on Mac seems far too small to justify the software engineering). And the phones are super-snappy in my limited playing around, if nothing else...
An A8 may already have closed the performance gap, and given the low power consumption of ARM chips, you can double or triple the CPU core count compared to x86 CPUs meaning anything that can run in parallel will be significantly FASTER than on equivalent x86 platforms with similar power consumption.
Also, while CPU emulation is a factor of two-four slower than native execution, the same does NOT apply to Rosetta-style on-the-fly recompilation, which has only a moderate overhead of maybe 25% or less, IIRC, and which in some cases even ran code faster than on the native CPU.
You'd first have to close the "coding for parallel processing" bottleneck. My understanding's that software developers by and large still aren't taking advantage of the multi-threading and cores available today.
And even if the OS can split up Apps and higher level chunks to use 'em between processes, they can do little (or nothing) about the code running within 'em to take advantage.
[Said the layman who's neither engineer or coder... ...so correct away if I'm wrong.]
Maybe Apple will test the market by releasing 2 versions of mbp or imacs: the current Intel based and the new ARM based.
That didn't work out particularly well for Microsoft with the Surface and Windows RT... I don't know if consumers are savvy enough to understand that an ARM/iOS version isn't just a less expensive version of the Intel machine.
Which architecture has the best potential to run fast, ARM or x86? Are there things about ARM that make it better other than using less power?
That didn't work out particularly well for Microsoft with the Surface and Windows RT... I don't know if consumers are savvy enough to understand that an ARM/iOS version isn't just a less expensive version of the Intel machine.
Microsoft launched Window RT / Surface RT a year before Intel became a viable option for low price tablets.
Honestly you sound just like the CP/M and Apple 2 guys complaining about the new IBM hardware and then the new Mac. Times change and you either get left in the dust or you work to keep up.
Or you could simply say intel is dishonest in their marketing. I find their current naming conventions to be extremely confusing and without a product map in front of you it is hard to tell what you are actually getting. Even AMD does better with the naming of generations of processors.
No, actually, I'm complaining that applications we rely on, may not be supported for some time, by the software vendors. Similar to what happened with some of the Adobe apps, and what happened to Quicken, when the OS could no longer support PowerPC apps. What were we told in that year or two before they re-supported (or rather supported) the newest OS10? The vendors told us to switch to windows. Doubt that would happen these days. However, we are stuck (for good or bad) using what works, until the rest of the world jumps forward to support any new OS based on ARM. Now, on the other hand, if no "PC" emulation or connection is supported under an Arm processor, us "corporate" types will simply not be able to use a new Mac, until a workaround comes about. Could take years, and thats lost productivity, if nothing works. Thats why we would have to stick with old technology until things fall in place.
The reason they switched from PowerPC to Intel was that the CPUs were becoming a limiting factor in making devices smaller and thinner (i.e. the Powerbook G5 that never happened). And certainly a switch from Intel to ARM could be justified on the same grounds. The thing I don't believe about the report is that Macs would have 4 or 8 of these CPUs. Surely that would require some nasty motherboard logic, with 8 CPUs all access the same memory? Doesn't sound believable.
In terms of having a Rosetta type solution on the ARM Macs to let them run old Intel apps, I don't think they should do that. Rather they should do the opposite, and have a Rosetta solution on Intel Macs to let them run ARM apps go forward. The ARM Macs should take on the policy of the iDevices, and only be allowed to run apps from the Mac App Store. And the only apps that should show up for downloading are ones that developer have uploaded as pure ARM or ARM/Intel fat binaries.
One out-there thing I wonder is whether Apple could possibly do static in-place Intel to ARM translation to the apps on the Mac App Store. Would that even be possible, or are there some things that simply can not be done until runtime?
Anyway I would be first in line to buy one of these new Macs. But mainly for software reasons not hardware. Switching to a whole new architecture is a golden opportunity to get rid of a lot of cruft and declare a lot of old frameworks deprecated and will not be supplied in ARM versions.
I often wonder why Autodesk doesn't port some of their major titles to Mac.
I suppose the AutoCAD community is entrenched in Windows and offering a Mac version probably wouldn't be particularly profitable, but 3DS Max? Wouldn't the people who use that rather use OS X?
Most of the film and TV stuff from Autodesk has Mac versions already, because they where all Mac first stuff I think, and Autodesk just bought them, and continues to develop for that entrenched market. AutoCAD actually did get a Mac port a few years ago, and my understanding is it is actually a really good translation of (most) AutoCAD functionality to a familiar but still very Mac UI. 3DS and Revit likely never will because the code base is such crap it would require a ground up rewrite, and Autodesk doesn't do that usually. Acad had a much cleaner codebase due to the major rewrite it got around the turn of the millennium. In general Autodesk is not even a Fast Follower, so much as a JIT Buyer. They let others actually think big thoughts and lead, and when something is beating them in the market they buy it and milk it till the next better idea comes along.
As for 3DS users preferring Mac, I really doubt it. Anyone who likes the craptastic UI and workflow that is 3DS probably doesn't even notice that Mac is elegant and easy.
I have no doubt they have experiments in the labs, but there is absolutely no way such a device would sell except to a power-conscious "green" customer. No Desktop will ever sell like this except maybe macmini "servers" relegated to NAS/Servers. An iMac would certainly fail, and a Mac Pro would be dead before it hit the store.
This is not the same argument as the PowerPC problem. With PowerPC, IBM was the sole CPU supplier to Apple, and IBM was much more interested in getting Nintendo/Sony/Microsoft to make consoles using their IP than they were building CPU's that could be dropped into a desktop or laptop. Part of the reason the PowerPC platform held on as long as it did was because those same parts were also being used for the Commodore Amiga. IBM makes servers with the POWER parts now. Sticking a server cpu in a laptop is a joke, and Apple knew it.
ARM's strength, and only strength is in power to performance ratio. Intel is terrible at this (Note how the average intel desktop is a 95 TDP watt target, not something reasonable like 30 watts, and the only parts that actually achieve low-power, are heavily neutered i3 and "Celeron" Atom type of parts which are an insult to use.) 8 quad core ARM parts may bring the performance of a ARM based desktop to that of a Intel based desktop, but I dare you to find any software that has been compiled to magically use as many threads as possible. No UNIX-land still programs like it's 1979 (and many windows apps too, just look at Google Chrome,) forking processes, not threading. It's pretty useless as a desktop when one ARM cpu core is the same power as a 15 year old intel CPU. if you had 4x8 cores (eg 32) threads, 31 of those threads would be idle, just like how 7 intel cores are idle on a i7 hyperthreaded CPU, and those other CPU cores are only turned on when a process is explicitly thread-aware (eg x264)
No, more cores doesn't work unless the software out there is rewritten for it, and only Apple's OS and Apple's software will ever do that. Adobe has enough problems just getting their flash player to synchronize across cpu cores. If a large company can utterly fail on their flagship products, that doesn't bode well for everyone else.
Most software that isn't outright number-crunching software (eg SETI@home) just fails to take advantage of multiple cores. It's just "too hard" for developers to not screw up. I've played countless games, and all of them still do everything single-threaded, and when they do take advantage of threads, it ultimately is too little. Dedicated consoles have a minor advantage here because games designed for them know exactly what they will have, while the average desktop or laptop can't be relied on having more than one cpu core.
This is why every time I see a story about "switching to ARM" in the context of desktops I groan and don't see it happening. At best/worst Apple is keeping their options open for a time when the ARM parts actually have comparable single thread performance. That is not this year, and not likely until at least 2016 when Intel runs out of die shrinks.
I'll blame Microsoft for just screwing up the marketing. the RT was a device looking to solve a problem that didn't exist. It was trying to run "full windows" on hardware that would normally be tablet hardware. The Surface Pro on the other hand is a laptop in a tablet form factor. The latter is utterly incapable of doing anything CPU intensive because Intel parts are bad (AMD are worse) in the side of device. The Surface Pro 3 just proves this point that Microsoft is trying to make a tablet using laptop parts, and doesn't give a care about the parts lasting very long. They removed the Wacom parts and replaced with inferior nTrig just to make it lighter. Let's not even get started on WindowsCE/Mobile/Phone
See in the Apple world, the difference between OSX and iOS is some UI API layer and a recompile. On Windows, you can't do this, because recompiling a windows app designed to use a two-button mouse doesn't work on a tablet that doesn't have a right-click button. Apple software however has always been able to work without a right-click button, as they traditionally only ever had a 1-button mouse.
Apple could switch to ARM for their desktops and laptops but that would be forgetting the mistake Microsoft already made too quickly. Software has to specifically be designed to take advantage of the many lower-performance cores, and most software aside from bechmarking software itself doesn't lend well to that.
The poor sales of things like AutoCAD are likely due to the software costing way too much for what you get vs what people need. Let's face it CAD was modern technology in the 70's & 80's, these days it isn't anymore difficult than a modern word processor. So if an excellent word processor can be had for $300 why not a CAD program. It should be noted that Dassault offers its basic CAD system for free.
I'm not sure what the problem is. I'm left with the impression that AutoDesk is on greedy company.
However, as far as ARM goes this would not be a justification for going to ARM. The best reason is that Apple would have no choice if it wants to continue to innovate in the desktop/laptop computer space. This is the same reason they had to start designing their own cell phone chips, in a nut shell silicon is the printed circuit board of the last century. In order to innovate in PC space you either need to design your own chips or get Intel to truly go the custom route. Intel seem highly resistant to custom chips so in the end it is either AMD or ARM.
By the way it seems like AMD has seen the writing on the wall here and is very willing to do custom SoC. In the end the business of computers is changing for the same reason the business of cell phones is changing. It is now possible to do SoC that are truly deserving of the SoC label. Not believable at all actually. How about neither? If they did this I wouldn't be able to support them by buying the hardware. Right now the only reason I have to run a Mac is to run apps not available on the App Store. Apps that will likely never be on the App Store. It would kill the platform for more advanced users of the machines. Given the above, the App Store is key to getting this to work. I could see Apple giving developers an ultimatum, either have the ARM binaries by XYZ date or your app gets dropped from the store. I can see most developers jumping on the bandwagon due to Mac OS being the only real growth opportunity out there for PC type apps. It is possible. What would be more likely is that they leverage LLVM and require apps be built to one of LLVM's intermediate forms and then have LLVM generate the native code on the target machine. It is an option different than fat binaries but a workable one considering how far LLVM has come in the last few years. Such an OS would never be tied to hardware in the future. Yep! Given that I have the same access as I currently do with Mac OS, that is the ability to install third party software and run scripts and such then I'd be all in. I just don't run Windows software anymore. Ideally the platform would also be open enough to run Linux also. Well it is the whole deal really. For me a whole bunch of low power cores would work out far better than the current two core solutions found in most low end laptops. With the ARM cores I can see Apple putting eight cores into a SoC by the end of 2015.
Possibly but Mac OS has been overhauled fairly completely now. My goal or hope would be that we would get a low cost platform, with all of those cores, that is significantly better than the Intel offerings in the same price range. Better means performance that holds up under load, power usage that is absolutely minimal under load and also when sleeping, and finally cool operation.
By the way, I realize that the ARM cores Apple would lead with this year won't be the fastest in the world. That isn't as important has getting access to lots of them.
That is unadulterated bull crap. Most software running on the Mac these days is threaded and does leverage cores. Beyond that if you run more than one app at a time those cores become very handy indeed. People seem to not remember anything around here but a significant number of apps got significant performance boosts when Apple released GCD (Grand Central Dispatch) a few years ago. GCD is just one way Apps can leverage cores. Sure you can go out and find a few apps that make use of no threading at all, but they are damn few these days.
The funny thing here is that the Mini or a replacement for the Mini would be an ideal location for ARM based chips. You could leave it turned on and with it in sleep mode most of the time your electric bill does not get hammered. It isn't being green it is all about economics.
As for the Mac Pro nobody is talking about an ARM based Mac Pro at the moment. A laptop however would be great for the same reasons a Mini would be. Well we can speculate all we want here but in the end IBM blew it. Actually by some measures Haswell does very well on performance per watt. The problem is it isn't a low power part by any measure even in Intels best offerings. As to ATOM, ATOM proves that Intel just doesn't understand the SoC market. Just about every app on the Mac these days uses threads in one form or another. There are a few that don't but it is damn few. There are several ways to make use of cores in Mac OS based machines, running separate processes are one of them. Just because a technique has been around for a long time doesn't invalidate its usefulness. The reality is an app can at times make use of multiple techniques to leverage cores such as starting processes that use threads. Baloney! Use some real apps and you will most impressed when you have lots of cores to leverage. Beyond that have you ever once ran more than one app at a time. Maybe updating from App Store while surfing the net, cores help when you do actually use the machine as multitasking machines. Flash is so thing that has been crap for years so I'm not sure why you even reference it. Again baloney. Come up with a list of apps that absolutely never use multiple cores for us. Some aren't well threaded of not threaded at all but the vast majority of apps these days employ at least some threading. You do realize that some of us use our machines for something besides gaming right? Besides you just have damned you anti threading flame here by admitting that even some games are threaded. Thinking from the past, most hardware of the last couple of years has been at a minimal dual core. It is pretty clear from this post that you don't have a clue and have been woefully misinformed about the value of cores. However I'm pretty much convinced that cores per say are not what Apple is interested in but rather it is the freedom to be able to innovate in the PC arena into the future. It is pretty much the same logic that forced them to develop their own cell phone processors, SoC are the fur ute and that is where your innovation will go.
Apples ARM parts really aren't that bad. Give those cores faster RAM, more cache and similar clock speeds and you might be surprised by their performance. Will they be as fast as Intels fastest cores of the moment, of course not, but Apple doesn't use Intels fastest chips in the Air nor the Mini. The point is even today's A7 core, in the right configuration, would produce a machine with a very interesting performance profile. That doesn't even consider the possibility of a new core beyond Cyclone.
Honestly you seem to be obsessed with single core performance, possible because of your gaming interests, but that is seldom important when it comes to modern computer usage. Crank up apps like XCode, Eclipse Safari, Mail, office, Numbers, Pages, keynote and any other common user app and convince me that they don't use threads at all. This doesn't even consider how many processes are active at anyone time on a machine. Cores are very important when it comes to machine responsiveness.