Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Apple's ARM Plans in 2009? NetBooks, Tablets, Other?
- Thread starter MacRumors
- Start date
- Sort by reaction score
*confused* at its most basic level, ARM assembly is 16 instructions and 16 conditions, with fully conditional execution and a no-penalty barrel shifter.
Compare:
http://en.wikipedia.org/wiki/Arm_instruction_set#Design_notes
http://en.wikipedia.org/wiki/X86_instruction_listings
I am approaching this from the design POV. Apple does low cost. This is how they protect their margins.
Apple has an ARM OS X with 10,000 apps already written. Is it a stretch to see a game console with a 400 or 600MHz ARM?
You say Apple won't do this, but Apple already does in the form of the
TV. If the TV could also natively run all the iApps, you'd effectively have a convergence device TV/iTouch/games console. Apple has shown a preference for selling hardware that is tied in to services like iTunes or the iApp store, and this would allow them to monetize the TV and open up new markets for iApps, whilst creating a moderately competent game platform that would have similar performance and stature to a Wii.It's really easy to sit there and say "it won't happen." I'm not saying it will. I'm saying these are the technological/hardware elements in place, and these are the easy steps forward from this position. Nobody knows what Apple WILL do, but we know from the technology they have available what doors are closed and what doors are open - everything else is product packaging and marketing. And that is something Apple does very well.
"Objective-C Instruction Set"
This could make sense if you assume the author really meant:
Objective-C Runtime Hardware Acceleration, which could take the form of a *new* dedicated instruction set.
I could see Apple wanting to speed up the ObjC runtime. Since ObjC is dynamic, it is both more flexible and significantly slower than C++ in an apples-to-apples comparison.
For example, Objective-C methods are compiled into C functions, with the method name (its "selector") stored in a giant lookup table along with information on which objects have it and the memory location of the function itself. Then calling a method involves looking up the method's name in that giant table, figuring out which version to call, and then running it. The time it takes to perform that lookup is the main reason Objective-C is slower than C++.
If a future chip had a built-in support for objective-c selector lookups, that'd *significantly* speed up anything written in Cocoa.
This could make sense if you assume the author really meant:
Objective-C Runtime Hardware Acceleration, which could take the form of a *new* dedicated instruction set.
I could see Apple wanting to speed up the ObjC runtime. Since ObjC is dynamic, it is both more flexible and significantly slower than C++ in an apples-to-apples comparison.
For example, Objective-C methods are compiled into C functions, with the method name (its "selector") stored in a giant lookup table along with information on which objects have it and the memory location of the function itself. Then calling a method involves looking up the method's name in that giant table, figuring out which version to call, and then running it. The time it takes to perform that lookup is the main reason Objective-C is slower than C++.
If a future chip had a built-in support for objective-c selector lookups, that'd *significantly* speed up anything written in Cocoa.
Netbooks are a true solution today. Not everybody wants to edit a video or build a webpage, I own 4 Macs, One Tower, One iMac and two laptops. Last month I bought a Netbook, erase the Windows XP OS and install Ubuntu OS (better that Windows more Mac feeling) and it´s been great since that day. Today My hard work is made on a Mac but the everyday tasks (office, email, music and web browsing) are made on my Netbook... 8 inches of pure pleasure. Apple need to build this little cheap machines, they do the job.
I am approaching this from the design POV. Apple does low cost. This is how they protect their margins.
Apple has an ARM OS X with 10,000 apps already written. Is it a stretch to see a game console with a 400 or 600MHz ARM?
You say Apple won't do this, but Apple already does in the form of the
It's really easy to sit there and say "it won't happen." I'm not saying it will. I'm saying these are the technological/hardware elements in place, and these are the easy steps forward from this position. Nobody knows what Apple WILL do, but we know from the technology they have available what doors are closed and what doors are open - everything else is product packaging and marketing. And that is something Apple does very well.
Finally! Someone who is as smart and cogent as I! Seriously, though - I am both smart, and cogent
.On a more personal and less sarcastic note - your posts have brought light to this thread and I, for one, thank you. A thoughtful post is hard to find on these forums most days (except for mine, of course).
D
No, I said they don't care that it is an ARM processor and is different from an x86 processor. They may care about speed of code, and they might care about code size (although I find that unlikely, because code size is quite irrelevant compared to the size of everything else). They don't care one bit whether it is an ARM processor or a bunch of crazy monkeys moving bits around at enormous speeds.
yep
so true. I could see apple using this to easily take a low end product to the market with in one quarter (Q1 2009). but i agree that they will probably do it mid 2009
so true. I could see apple using this to easily take a low end product to the market with in one quarter (Q1 2009). but i agree that they will probably do it mid 2009
I agree - although I installed OS X on my netbook (please - I know how some of you feel - but I bought a valid, legal copy - and I know, Apple says I am using it in an incorrect fashion, but make a netbook and I will gladly restore it to factory software). I think I use the netbook for most tasks such as web-surfing on the go with WiFi and e-mail at the house - and iTunes.
D
Long History
This makes a lot of sense considering Apple's long history with ARM, the company, and the CPU. I wonder if Apple retained any rights, licenses, at all from when they helped found ARM Holdings back in the 90s. I remember talking up the StrongARM used in the Newton when I was an Apple Student Rep.
This makes a lot of sense considering Apple's long history with ARM, the company, and the CPU. I wonder if Apple retained any rights, licenses, at all from when they helped found ARM Holdings back in the 90s. I remember talking up the StrongARM used in the Newton when I was an Apple Student Rep.
I'm not saying they aren't becoming popular, but I really don't think this is where Apple would be headed. The desktop space (this includes netbooks, notebooks, and desktops) are basically dead, in a sense. The two big things right now are web and mobile. Apple seems to be clearly focused on mobile.
They've now got a pretty strong mobile platform with 2 devices so far, and I expect that to grow. So I don't think they are going to bother with netbooks, as it's not a way for them to push forward.
I think they'd be smarter putting a mobile OS (iPhone OS) on a device than a desktop OS (Mac OS) in a cramped fashion (let's face it, if you put Ubuntu or XP on a 5inch netbook, it's going to feel super cramped). Apple upsizes the user experience, not downsize.
I don't think "days" of use is the holy grail. But the expectation is there for a handheld device (e.g. the iphone) to last at least a full day's use without charging. The iPod was able to do 8-10 hours of playback from day one. That to me is the appropriate threshold.
Back in the day (late to mid 90's I think) when I was working at an ISV that was an Apple Developer Partner, we got access to really early Mac Os 8 (Copeland) and CHRP hardware - remember that?
Anyways, the CHRP box we were provided had a Java VM built in on a chip. I think it was JDK 1.0.2 if I remember correctly. The machine was fast all around, but the JVM screamed (until the VM had heap issues).
To my knowledge we were one of two ISV's (for very specific reasons) that were provided these boxes. It would not be a far fetch to consider that they may put objective-c on a chip as well. This would accelerate almost every Apple app on the current Mac.
Food for thought - let me see if I can dig up some old shots of the box and the innards....
Anyways, the CHRP box we were provided had a Java VM built in on a chip. I think it was JDK 1.0.2 if I remember correctly. The machine was fast all around, but the JVM screamed (until the VM had heap issues).
To my knowledge we were one of two ISV's (for very specific reasons) that were provided these boxes. It would not be a far fetch to consider that they may put objective-c on a chip as well. This would accelerate almost every Apple app on the current Mac.
Food for thought - let me see if I can dig up some old shots of the box and the innards....
Cheap and Mac?
What's wrong with the above statement. If anything, cheap is always followed by "crippled" or "missing "standard" features" if anything at all. I mean, now look at the Macbook vs Macbook Pro - same chasis (unibody) with exception of a .50 cent piece of silicon (firewire), an express slot, $39.00, and a few extra inches and worth $700? I don't think so. Maybe it was worth more when you were comparing plastic (previous macbook) to 50+ screws that had to be screwed and un-screwed, now it's apple that's doing the screwing. LOL.
But seriously, does anyone in their RIGHT mind think we'll ever see a workable "inexpensive" piece of hardware that has features that we ALL need? Right now they can't even get PUSH or FLASH to work on the iPhone and FLASH is 80% of the internet.
Only a failed economy and slow sales would force the above scenario to happen, unless of course, Jobs steps down and the marketing department takes over the way it should.
Here come all the "defending" why should Apple accommodate (fill in the blank statements) - hey, I'm not the one who told Mac Specialists that they don't do a good job if they don't sell Apple care and One to One with Macs then turn One to One (which used to be PRO CARE) and training for Logic, Shake, FCP into a place where barking dogs, baby's, and people that don't know what an attachment is try to learn iLife with out FW support for their camcorders while pushing the creatives (those that kept apple alive for years) making the decisions.
Have a nice day.
Of course it would still be as ARM, just as it is today after several extensions and a nearly complete different instruction set have been pasted on (arm4, 2 thumb ISAs, Jazelle, VFP, & etc.)
CPU architects study things such as the best optimized LLVM output and Obj-C runtime, and look for opportunities to decrease average instruction path length by greater amounts than the required additional hardware. Having not done that (recently), I can't say whether it's a good idea or not.
.
Actually, Intel used to make ARM processors (XScale), but sold that operation off to Marvell a couple years back.
.
"Battery life measured in days, much like smartphones."
You mean like the iPhone?
The iPhone is only rated for around 3 to 7 hours if you watch a running app continuously. The graphics display refresh and backlight take up a huge chunk of the power. So, without major display improvements, battery life would not go up to days even if the CPU took ZERO power. So maybe they have some clever display technology (as has the OLPC) in the works as well.
.
The dumbest move they every made. Talking to colleagues at Intel they all agree its the dumbest move and their answer, Atom, pales in comparison.
The Java VM isn't the Java Runtime.
The equivalent would be putting the LLVM on a chip and leveraging optimizations with CLang to be tuned to the best optimizations for C, C++, ObjC, Java, etc.
I'm guessing that this is why Steve is going to make everything iTunes Plus in the near future, according to 1 rumor. He doesn't want this announcement to take away from the stuff he's sure to announce at MacWorld Expo in just a few short weeks!
BJ
BJ
Ok i maybe completely dumb as well here.
But what about LLVM that Apple have been hard at work using and improving to make a lot of new stuff in core libraries in the past and SL into the future.
Could this custom ARM be a Low Level Virtual Machine that wasn't so Virtual.
At least in part a Low Level Hybrid Machine if you will.
If they use the Touch OS (for want of a better term) for this so-called Netbook, then using the Air as a point of comparison is not partially apt.
Apple NEEDS a netbook category computer. They are a very quickly growing segment of the laptop market. And as Jobs has always quoted Gretsky "you skate to where the puck is going to be." (or something like that). If Apple doesn't start making relatively low cost netbooks soon, Apple is gonna start to lose marketshare again.
Apple can sell netbooks at a nice profit too, because people expect to pay more up front for Apple hardware and OS X. If Apple sells a new netbook type device with hardware equivalent to the $400 netbooks already out there, and then sell it for $600, they'd sell like hotcakes and Apple would make a boatload of money. At over $200 profit on each device, Apple would sell a zillion of these and MORE than make up the profit that they'd lose in MB Airs. Apple probably makes much more than $200+ profit on each MBA, but Apple would sell SOOOO many $600 netbooks that they'd wind up making a LOT more money.
Apple can sell netbooks at a nice profit too, because people expect to pay more up front for Apple hardware and OS X. If Apple sells a new netbook type device with hardware equivalent to the $400 netbooks already out there, and then sell it for $600, they'd sell like hotcakes and Apple would make a boatload of money. At over $200 profit on each device, Apple would sell a zillion of these and MORE than make up the profit that they'd lose in MB Airs. Apple probably makes much more than $200+ profit on each MBA, but Apple would sell SOOOO many $600 netbooks that they'd wind up making a LOT more money.
yeah there were a slew of horrible decisions during that time frame.
We never understood why the arm chip wasn't ported over to the 860/861 process. They could have ported it over to 120nm then 90nm then to 65nm years ahead of others.
The atom may be bloated but on a 32nm platform it will be interesting to see what it does for handhelds and notebooks.