• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

MacBook Air 2011 Refresh

OnePingOnly

Senior member
What say you, AT Apple forum?

I really like the current (2010) MBA but I'm resistant to purchasing anything these days which still has the Core 2 Duo as its mainstay CPU. Does anyone have a clue which mobile ULV Sandy Bridge processor may be a candidate for the next refresh? I'm looking at going for an 11" model.
 
Most likely either the Core i5 2537M (1.4GHz) or Core i7 2657m (1.6GHz), which will have the same TDP and power consumption roughly, but you're exchanging GeForce 320M for Intel HD graphics.

Battery life might not see a boost, but processing power will undoubtedly soar to near MacBook Pro 2010 13" level at 2.4GHz or 2.66GHz.
 
Most likely either the Core i5 2537M (1.4GHz) or Core i7 2657m (1.6GHz), which will have the same TDP and power consumption roughly, but you're exchanging GeForce 320M for Intel HD graphics.

Battery life might not see a boost, but processing power will undoubtedly soar to near MacBook Pro 2010 13" level at 2.4GHz or 2.66GHz.
Do you see the exchange from the Geforce 320M to the integrated Intel HD graphics as a bad thing (e.g. detrimental to performance in other aspects)? I wouldn't be gaming with it.
 
I think it could be detrimental in certain cases, but the added processing power far outweighs that. For the most part, Intel HD is equal to GeForce 320M where it's due, and it also supports OpenCL for added computing acceleration, so not all is lost.

If anything, I think it makes more sense to use Intel HD since the footprint should be less than GeForce 320M, and the added space can be used either to add in more chips (RAM or SSD), or expand the battery a bit. As it is, I think a MacBook Air 11" or 13" with around 7 hours of usage would be perfect. With backlit keyboard and they can start calling the thing a MacBook Pro.
 
Are there any SNB Core iN chips with the TDP that the MBA11 uses? IIRC it is lower than the MBA13s TDP, hence why you can't get the 1.86/2.13 in it.
 
Are there any SNB Core iN chips with the TDP that the MBA11 uses? IIRC it is lower than the MBA13s TDP, hence why you can't get the 1.86/2.13 in it.

I believe the TDP of the MacBook Air 11" chip is 10W (SU9400 and SU9600), while the MacBook Air 13" chip might be 17W (SL9400 and SL9600), unless Apple used parts not included in Intel's database.

The closest Sandy Bridge chips that come close are:
Core i7 2657M 1.6GHz
Core i7 2617M 1.5GHz
Core i5 2537M 1.4GHz

They are all 17W. Considering they integrate a bunch of chips inside one package, perhaps the entire TDP of the whole board will level out to be slightly higher than the current MacBook Air 11". But the extra space gained by eliminating board footprint may allow them to fit a bigger battery in to level out the TDP difference and still get roughly the same or slightly better battery life. The MacBook Air 13" will benefit the most from this, I think. It's not far-fetched to believe they may be able to reach 10 hours in the 13" with this if they use the denser battery in the iPad 2 coupled with the lower power Sandy Bridge chip.

how's intel hd opencl working these days? up to snuff with cuda?

GPU not supported is the current status.

It's weird since they sang about OpenCL support for so long, but I suspect it might be due to a driver issue or something along the line. Not surprising since Intel has always been bad wih drivers.

Another crackpot theory I have is that... the new Sandy Bridge chips are fast. Fast enough that enabling OpenCL support on the Intel GPU wouldn't make a difference, so maybe that's why they disabled it.
 
I believe the TDP of the MacBook Air 11" chip is 10W (SU9400 and SU9600), while the MacBook Air 13" chip might be 17W (SL9400 and SL9600), unless Apple used parts not included in Intel's database.

The closest Sandy Bridge chips that come close are:
Core i7 2657M 1.6GHz
Core i7 2617M 1.5GHz
Core i5 2537M 1.4GHz

They are all 17W. Considering they integrate a bunch of chips inside one package, perhaps the entire TDP of the whole board will level out to be slightly higher than the current MacBook Air 11". But the extra space gained by eliminating board footprint may allow them to fit a bigger battery in to level out the TDP difference and still get roughly the same or slightly better battery life. The MacBook Air 13" will benefit the most from this, I think. It's not far-fetched to believe they may be able to reach 10 hours in the 13" with this if they use the denser battery in the iPad 2 coupled with the lower power Sandy Bridge chip.

The MacBook Air, and all the non-removable battery MacBook/Pros have the same battery tech as the iPad 2.

And the concern for me isn't the battery life, though I do get better battery life on my MBA than I ever did on my old MacBook, even though it is less than half the weight and like a third the volume. The concern is HEAT. With a smaller logic board they could fit on a larger heatsink, something more in line with what the MBA13 uses.

I am just waiting for when the Pro and Air lines merge. Have just one line of aluminum unibody system (assuming they don't switch to a different material, carbon fiber is lighter after all) that run from 11" to 17".
 
I think heat is always going to be a concern with a powerful Core processor.

I'm more waiting for the day when Apple officially switch to ARM for their Mac lineup, and then heat won't be a concern anymore.
 
I think heat is always going to be a concern with a powerful Core processor.

I'm more waiting for the day when Apple officially switch to ARM for their Mac lineup, and then heat won't be a concern anymore.

I don't see that happening any time soon. I think we have a better chance of Intel developing specialized, ultra low power X86-64 before we see ARM based CPU's on the Mac lineup.
 
Well, it probably won't happen any time soon, but at the rate Apple is renovating their mobile hardware, it'll only be a matter of time before the iPad can catch up to the 2010 MacBook Air in terms of processing power. Maybe in 2 years.

Also you have to consider the fact that Apple has an in-house chip development team dedicated entirely to developing ARM-based SoCs. Couple that with Apple's preference of assuming total control over hardware and software development, and it's not really that hard to see.

Meanwhile, Intel is still struggling to bring x86 down to the same thermal and power package as ARM.

Another factor to consider would be that the ARM codepath for Apple's OSX, which is what iOS really is, is arguably much more optimized than x86 for performance. iPad 2 beats a lot of x86 Mac in iMovie, for instance. The merge of iOS and OSX also have already begun with OSX Lion taking a few elements from iOS into its own. I'm not surprised if iOS 5 "borrows" a few elements from OSX. It's just bound to happen.
 
Last edited:
I don't see that happening any time soon. I think we have a better chance of Intel developing specialized, ultra low power X86-64 before we see ARM based CPU's on the Mac lineup.

I wouldn't be so sure on that one. The iPad 2 gets 10-12 hours of battery life. It has a 25Wh battery which means that it uses a mere 2-2.5W of power, and that is with everything included.

The MBA11 gets anywhere from 2.5-7 hours of battery life. iFixit's teardown reveals that it has a 35Wh battery, which translates to 5-14W of power usage. Best case scenario it uses twice as much power, worst case 7 times as much power.

ARM is going to keep getting faster and more capable within the same power window, Intel has to keep working harder and harder to get their power usage down while maintaining their performance. I would argue that it will be easier for ARM to go up than intel to go down.
 
ARM is going to keep getting faster and more capable within the same power window, Intel has to keep working harder and harder to get their power usage down while maintaining their performance. I would argue that it will be easier for ARM to go up than intel to go down.

If I may add, I think it's easier now for Apple to super-optimize their softwares to perform better on ARM than it is for them to redo everything for x86. So it may not be that easy for ARM to go up, but Apple has proven itself capable of bringing its software down to match ARM and x86 performance.
 
If I may add, I think it's easier now for Apple to super-optimize their softwares to perform better on ARM than it is for them to redo everything for x86. So it may not be that easy for ARM to go up, but Apple has proven itself capable of bringing its software down to match ARM and x86 performance.

I would say that they are more than capable of optimizing their software for x86. Up to 7 hours! of battery life on the 11" Air. I mean cmon, they made a deal with Satan or someone to get that. The 13" gets something obscene like 11 hours on light usage, which just defies belief considering its size.
 
Well, the 11" Air gets something like 3-5 hours realistically with continuous usage, and the 13" gets something like 5-7 hours. Either of them only gets that obscene battery life when your usage comes in short bursts of 5 - 10 minutes during the entire time.

In comparison, the iPad with its 25Whr battery can go for 12 hours continuously, or days if your usage is, again, in short bursts. My 2010 MacBook Pro 13" will say 15-18 hours with that same usage, but I can never go beyond 7-8 hours with continuous usage.

So even though it might be true that Apple is more than capable of optimizing for x86, I think it's fairly obvious that their optimizations for ARM go much deeper than x86. And if they have to choose between the two, I believe they'll choose ARM since they already have a very optimized and stable platform that they can just build upon. Plus ARM hardwares likely cost them less than x86 to manufacture.

But fun fact: my 2010 MBP rarely gets hot. Very rarely. I can put the whole computer inside my backpack and have it running all that time without overheating issues.
 
Last edited:
yes but your MBP has very thick heat sinking material compared to the paper thin macbook air. i walked slowly around the corner and clipped the wood door to the bathroom (reading material when i take a dump,tmi) and it dented the screen metal and dented the wood. i was shocked since i bought the macbook air pre-dented but took it apart to straigten the sheet metal - this was a dent where no man has gone before and i swear to god this thing is cheap.

the mba acts like the old 2007 (non-unibody) as far as heat and collects lint quite well on the fan. The 2008 unibody macbook pro's are totally different and my 2009 17" macbook pro does indeed stay extremely quiet.

remember battery life quoted is using SAFARI (firefox eats more) and no flash since that is the TRUE DEVIL. lol. i suspect html5/javascript can be just as taxing.
 
I'll agree that battery life on iOS is good with the ARM chips, but I wouldn't say they're near-ready for full fledged PC's. The argument that they produce good speed and battery life on iOS is based on an unequal playing field.

Keep in mind that iOS is totally restrictive and doesn't multi-task. With concurrency out of the question, that's a giant chunk of overhead that iOS saves on. Also, the operating system denies all forms of garbage collection in apps, which also saves on battery life. It doesn't let the user run extra services. It's only a single-user environment. So yea, iOS is extremely optimized, but it's for a limited domain. I don't think ARM chips of today are anywhere near where they need to be for general computing on a fully capable operating system.

Maybe in a couple years? Even so, I don't think the "Pro" lineup will ever switch. There's also the issue that all software must be recompiled for the different architecture, which is a pain. Although, some years ago, Transmeta made a multi-core CPU (VLIW architecture) that used one core to translate and schedule the X86 instructions on the other cores. They called it "code morphing" and it was a cool idea. That could be another possibility with ARM, at least for the transitional stage.
 
Last edited:
Well, I think some of your statements came from misunderstanding of technical data, but here are some points I would like to elaborate on:

- iOS does support multitasking. What it doesn't do is allow excessive background processing of any application or service. This holds true for all applications, even for those from Apple. It does mean, though, that an app may always try to finish its current task before suspending itself to background. Certain services are given to allow certain tasks to process in background without inhibiting performance of the whole OS. These are location services, network services, and audio services. So at the very least, it is still barely enough multitasking for most users. Just that the device itself isn't multitasking, much. At least not in terms of running multiple applications at the same time, but there is a good reason for that.

- Garbage collection does happen in apps. In fact, it happens very regularly, and apps are always given low memory signals to start flagging their memory space for garbage collection. This has to happen very frequently, or otherwise iOS won't have enough memory for anything. Both the CPU and GPU in iOS have to access the same memory space, so while 256MB looks like a lot, there are times when less than 96MB is available to the CPU, and about 30MB of that has already been reserved for system services, leaving only around 60MB for apps to use. That was the memory strain on the first iPad.

- Users can run extra services when jailbroken without any inhibitive penalty on performance. In fact, "real" multitasking may also be done after jailbreaking without much of a penalty to performance either way. The main issue is still memory shortage. iOS doesn't allow swap file with its virtual memory system, which in turn means that however much RAM the device comes with, that's the absolute in terms of memory usage for everything. You'd quickly run out of memory if you start letting applications process things in the background and then all sorts of crash would happen because something tries to access more memory than it should. But Apple stubbornly refuses to enable swap file despite the fact that the service to manage it is right there. Hopefully they'll do something about it in iOS 5.

So you are right, partially, in that iOS is restrictive. But not in execution method but in memory management method. It's insane to think about how much Apple is holding back iOS by forcing measly amount of RAM on their devices. But at the same time, this strategy also preserves low memory usage for future applications. When they have enough apps, that's when it's safe for them to throw gigs of RAM at the equation and have tons of apps running in the background without having to worry about any of them bogging I/O access for swapping files.

I'd also argue that Apple's way is much better as each app understands best what it needs to preserve in RAM to keep itself from crashing. Swap files would swap everything indiscriminately.
 
Last edited:
It's not true multitasking. Sure, a few services can run in the background, but the *user* cannot perform two tasks at the same time. Multitasking is about the user, not the machine. This scheme works really well on a phone, but not so much on a fully featured operating system. As soon as you enable true multitasking and the user wants to do some real work, you need more resources. More resources = more power consumption. And if we take into account multi-user environments, such as Mac OS X, then the problem grows even larger.

And I can't find any mention of garbage collection anywhere in the iOS SDK. In fact, all I could find is specifically there is *no* garbage collection on iOS. I think you're mistaking autorelease pools with garbage collection, which are not the same thing. Where did you get that information?

Fact is, there's a lot more going on with OS X than iOS. It does require more resources, and more resources equals more power consumption. I'm not saying ARM is completely out of the question; all I'm saying is that we can't take the perceived benefits of ARM due to its success in the iPhone/iPad at face value. There's more to take into consideration.
 
I think you are confusing multitasking with multi-window. In iOS, it is very possible for a user to accomplish two tasks at the same time. You can leave the web browser to download your document while you go and work on another document, all the while listening to music. On any computer, that should be about the limit of what you can do. You only have a pair of eyes, a pair of ears, and a pair of hands, with one brain to control it all. You can't play a 3D game while you work on your document at the same time. You can play your 3D game then switch to your document, though, which is what iOS does, just fine.

Also, you seem to be confusing automatic garbage collection with manual garbage collection. What iOS doesn't have is it doesn't have an automatic garbage collector, but it very much has a manual garbage collector which gets triggered on a per flag basis. It just doesn't swipe until more memory is needed. I guess using the term "memory management" is more proper since "garbage collection" somehow substituted for "automatic garbage collection" where it shouldn't.

There is no denying that there are a lot more going on with OSX, but at the same time, if you view both iOS and OSX as interfaces to achieve your tasks, which is what users do, then there is virtually no difference as long as it's regular goals such as listening to music while working on your essay or something along the same line.
 
ARM is going to keep getting faster and more capable within the same power window, Intel has to keep working harder and harder to get their power usage down while maintaining their performance. I would argue that it will be easier for ARM to go up than intel to go down.

You also have to keep in mind that ARM isn't even attempting to compete with Intel on performance. It's easy for ARM to make big strides in performance when they are starting from so far behing since all of the hard work has already been done so to speak. Therefore they may make progress narrowing the performance gap with Intel but actually catching up with or overtaking them won't happen for a very long time.

To put things in perspective a single core low power Atom is still faster than a dual core 1ghz Cortex A9 processor and Atom hasn't changed significantly since 2008 and it wasn't a very fast part back then.
 
I'd like to see what OS X could could do with a 16 core ARM. Let's go ahead and assume all software would be multithreaded within reason. It could be a pretty nice experience, actually. If anyone's going to do it, it'd be Apple.

So yea, if it's going to happen, I'd definitely like to see a MacBook Air with an ARM solution. Maybe an 8 core or something. They require a LOT less hardware than X86, so 8 to 16 core chips aren't farfetched at all. Power savings could be amazing and we could still have some sweet speed. As for X86 support, as long as we've got some decent emulation, I'd be okay with it. It's kind of like when Apple initially switched to X86, OS X had Rosetta, the PowerPC emulation layer. Eventually it was phased out and all was well.

As for ARM across the board, I'm not so sure. Who knows what'll come out, but as for now, it looks like X86 will be the performance king for quite some time to come. In other words, even if the MacBook Airs switched to ARM, I don't expect to see the high end products to switch soon after... or even ever. But who knows; something cool might get developed. Intel could have something up their sleeves - we don't know 🙂
 
Do you think the next hardware refresh will happen after the release of Lion? It seems logical on Apple's part to make sure the next refresh has the next OS.
 
Back
Top