Mild-Mannered Canadian Fury

Doug Stephen is Politely Peeved

Mactoberfest


Mon, 15 Oct 2012 Ā«permalinkĀ»

As time has gone by, I’ve become less and less fond of writing about my suppositions on new Apple products. It’s weird. I like talking about it. I have friends that are Apple geeks, and we like to talk about these things. I’m even the Apple “pulse” at work; the go-to guy for questions about Apple rumors. I’m not sure why, then, I don’t really write about them anymore. Maybe it’s because, so often, my opinions line up with those of others who are much more popular writers than myself, so I figure “why bother?”. It’s not a good excuse to not write. Because it puts me in the position of writing for an audience instead of writing for myself; and really, shouldn’t that be what this is about? Me, writing, with my own voice, to say what’s on my mind?

Anyway. Ranting aside. It looks like we’re gonna have an Apple event this month. Jim says so. There will probably be a new product. There might be some old products refreshed. And there will probably be other products that don’t see anything happen to them. After giving it a few weeks for the rumors to simmer and some decent information to bubble up, here are some thoughts on the direction of Apple’s product lines and where we’ll see them going.

New Desktops

There have already been rumors swirling around about new iMacs. Talks about a change in form factor, improved screens, and leaked parts have been swirling around for a few days now. In addition, the iMac has been a hot topic of conversation on blogs and podcasts lately due to Apple’s “Retina All The Things” initiative, with many a nerd becoming damp in the crotch at the thought of a 27” Retina iMac. But the iMac isn’t the only desktop offering. A quick gander at the MacRumors Buyer’s Guide will show you that the entire desktop line is woefully out of date; the pathetic bump to the Mac Pro not even registering and the Mac Mini being left shunned and alone under the stairs since last July. We already know that the Mac Pro will receive some proper love next year. But what about the rest of the desktop line?

Mac Mini

I wouldn’t be surprised if we didn’t see a major bump to the Mini this year. It might get mentioned in passing, or it might receive a silent bump. It’s spot as a beloved home server and media machine means that it’s a prime candidate for a lot of the new tech tricks that Apple is pimping; notably, Ivy Bridge/Intel HD 4000 graphics units and USB 3.0 for external media. I wouldn’t be surprised if a minor spec bump of this nature does occur at the October 23rd event. But I wouldn’t expect any major revamps to the design of the wee fellow. What I would be interested in seeing, however, is if the improved thermal profile of the new chips and a little bit of R&D time refined on the cramped Retina MacBook Pro leads to us seeing more Mac Minis with discrete graphics cards in them; while Intel HD 4000 is more than usable in something being billed at the entry-level market desktop user, a Mac Mini with discrete graphics and a decent Ivy Bridge processor would be awesome for something like Steam’s Big Picture mode. That last little bit about the GPUs isn’t strategic in the slightest; it’s just something that I’d love to see. I wouldn’t bank on it.

Tangentially, and very nerdily, I’m happy to see Apple and Nvidia back in bed together again. I’ve always felt that Nvidia is a superior graphics card maker (subjectively), and I’m also a gigantic mathematical programming geek, so I’m happy that Apple will be producing CUDA capable computers again.

Mac Pro

I’ve discussed once, in the past, my thoughts on what the holdup is on a proper Mac Pro refresh. But time has passed, we’ve all grown wiser and wearier, and I’ve thought. Interestingly, my opinions on the whole thing haven’t changed that much. I don’t think we’ll get a surprise announcement, seeing the computers moved up to this year. I think 2013 is very much a real timeline for a new Mac Pro, because of the interesting engineering challenges posed.

The Mac Pro utilizes (and, in my opinion, should continue to do so) Intel’s “server” brand processors. The processing units themselves are nearly identical to the processors used in high-end i7 chips in the current generation’s parlance. What is different about them are a few small, subtle things that are closely related to “binning”. Computer processor manufacturing is an extremely imprecise practice. The components are so small and so fickle (physics is a bitch), that manufacturers have resorted to a process known as binning; setting different performance tolerances and then selling different tolerance tiers as different products. Two i7 chips manufactured in the same process might respond totally differently to strain and heat, so the less tolerant one will have it’s clock speed tuned slower and sold as a slightly less performant version of an i7, whereas the one that can take the strain has its clock tuned higher and are sold as higher end chips. Server-grade processors are typically of the best bins; they need to be extremely fault-tolerant to the environment and their workload since servers often perform difficult tasks and are meant to stay powered on for months and months and months at a time without a reboot or power down. In addition, Xeon chips are tuned so that they can be used on motherboards that feature more than one physical CPU unit; not just cores, but actual processors. The “12 core” Xeons that Apple sells are in fact computers with two separate 6 core Xeon chips on them. These provide a ton of power, but the high binning requirements are what make them so costly. They also provide support for other server-y things like special RAM, but I won’t get in to that.

As I mentioned in the linked post, the Xeons based on Sandy Bridge don’t feature native USB 3.0 support. In addition, they didn’t introduce an overwhelming amount of performance increase over the Nehalem chips. The Ivy Bridge versions were even less impressive, basically refining on the power consumption of the Sandy Bridge models. It’s obvious that Apple is going to want processors that can natively handle the newest technology that they want to include in their devices; I’m nearly 100% positive that this is the reason they held out on USB 3.0 until a chipset supported it natively, of which Ivy Bridge was the first. And I think this is still a driving factor. The short of it is that the current Mac Pro is still incredibly powerful for what it does, and until they can release a new one that supports all of their newest whizz-bangs then they simply have no need to; it was never a product that sold in massive quantities anyway so keeping it super “fresh” just isn’t super “important”.

USB 3.0 support aside, there’s also that whole Thunderbolt thing. There are a lot of interesting engineering problems that come along with implementing Thunderbolt well in a server-class device; the main one being that these machines are expandable and capable of using a wide variety of graphics cards. Video is only half of the Thunderbolt equation; the other half is data transfer. Any port that is billed as a “Thunderbolt” port needs to be capable of transmitting both a video signal and a stream of data over PCI. As it stands, there are exactly 0 consumer graphics cards on the market capable of connecting to external hard drives. Think about the following: The Mac Pro is not only capable of running removable PCI discrete GPU’s, but it usually needs to. Xeon chips typically don’t offer an integrated graphics unit. Not only are these devices removable, but they are capable of being utilized in parallel, multiple cards in one machine.

We know that Apple can’t “cheat” by making the Thunderbolt controller hook up to the “integrated” graphics and then customizing the chipset for data; there is no integrated graphics controller on the server chip. There’s no way in hell that Apple is going to release a modern Mac that isn’t capable of dealing with Thunderbolt. So they still have to solve the Thunderbolt+Discrete GPU problem. The way I see it, there are a handful of ways to solve this problem. One way is to force GPU OEM’s to start making special graphics cards that handle data and conform to a proprietary Apple bus on the motherboard. This also forces consumers to buy special Mac Pro only cards. There’s already a little bit of a divide when it comes to Mac Pro graphics cards; they usually need special firmware. But at least the more adventurous people out there can get bog-standard cards and attempt to reflash them. This would be impossible if a proprietary bus was introduced. This is probably the “easier” solution for Apple, but I’d also call it suboptimal.

Another solution is the introduction of some sort of centralized Thunderbolt IO controller, with the PCI lanes being passed through to the IO controller. The controller itself has the data portion built in, but receives its video control from the cards being bussed upstream. This introduces its own problems, like separating PCI video cards from non-video PCI devices, but it also seems like it might be the least-hostile way of moving the Mac Pro to Thunderbolt, allowing people to use existing Mac Pro compatible graphics cards in their rigs, as well as keeping the challenges mitigated on the GPU OEM side of things.

Lastly, is the case. Yes, the design of the Mac Pro has become almost iconic. But it’s old. Ignoring the subtle tweaks to the back and insides, the overall look and feel of the current Mac Pro tower debuted with the Power Mac G5 in 2003. The design is running on 10 years old, and it’s time for an update. The demand for a revitalized look as well as accommodating the new technology that the tower will need to grapple with is a perfect excuse for the Industrial Design team at Cupertino to rebuild this sucker from the ground up. And that doesn’t happen overnight.

So yes, the Mac Pro is due for a reboot. But to be honest, I wouldn’t be surprised if they’re waiting for Haswell, which means that we may not see new Mac Pros until as late as next summer. Don’t expect anything about the Mac Pro on the 23rd. I’d put money on it.

iMac

This seems to be the area in which we have the most information regarding imminent refreshes. I already linked to it in the intro, but there are claims of parts from a revamped 21.5” iMac that is going to be unveiled this month at the iPad Mini event.

This is totally plausible, and I think seeing an iMac refresh along side a Mac Mini bump makes sense. And if I had to only pick one of the two to get some limelight alongside the iPad Mini, I think it would be the iMac getting a little bit of love while the Mini either gets nothing or is spec-bumped quietly in the background with no more fanfare than the “new” indicator on the store. We have no visual proof that the new iMacs will sport a new design, but lots of anecdotal conversation about a new “tear drop” form-factor and an improved screen. Nobody has bandied about the word “retina” with a lot of authority, but of note is the fact that we’ve only received information regarding the 21.5” model of the “new” iMac.

Obviously, people want HiDPI iMacs. If only because it’s such an insane proposition, especially on a 27” display that already sports a mind-blowing resolution. I use a 27” iMac at work, and it’s damn near retina enough for me. But anyone will tell you, especially if they’ve played with a new rMBP, that the allure of these displays isn’t solely in the pixel density; the sharpness and detail and color reproduction that becomes possible by making those pixels so small is truly remarkable. It’s also computationally expensive.

Ignoring the fact that getting a reasonable yield on 27” 5120x2880 IPS panels is going to be a nightmare, pushing out that many pixels from a graphics card is also fairly non-trivial. Driving that many pixels alone is an incredibly daunting task and it will require a very, very powerful GPU. Especially if they want to let people do other tough things like play games or do complex graphics renders. Consider that, not only does the iMac have to power its own display, but that it also has to be capable of powering at least three displays, all of which can have the same possible resolution, and the math gets scary. If Apple is capable of producing an iMac with that level of visual beauty, then they are also capable of releasing a Thunderbolt display with the same resolution. And the iMac has always been a machine with the ability to power external displays. I realize that this next statement refers to such a tiny minority of real-life situations, but a 27” iMac should (at least on paper) be capable of powering at least two additional external displays with equivalent resolution. An iMac, flanked by two Thunderbolt displays.

I’m here to tell you that there isn’t a single graphics card on the market that is capable of powering 15360x8640 pixels that would also fit inside of an iMac. Because of how cramped their internals are, iMacs actually utilize laptop GPU’s, and there is definitely not a single laptop GPU capable of pushing that many pixels. Desktop setups that drive that many pixels all utilize multiple GPU’s to do so, and they are a LOT larger, hotter, and more expensive than the single GPU’s used inside of iMacs. And I just don’t see Apple releasing a 27” iMac only to have to turn around and say that you aren’t allowed to connect it to an external display or that you can only connect it to displays with shitty resolutions.

I’m not saying it won’t happen, and I’m not saying that it’s outside of the realm of possibility that Apple can’t figure out a way to power external monitors even though most of the GPU’s resources are being taken up by the main display. Hell, maybe Apple will drop support for external displays; I think it would be a stupid move, but not outside of the realm of possibility. But then we see that the only iMac that we have proof of a redesign for is the 21.5”. The current model in this size has a resolution of 1920x1200, and we already know for a fact that the Kepler GPU used in the rMBP can push out that resolution in HiDPI mode because, well, it does if you ask it to. And we also know that the Kepler is capable of handling that kind of resolution while also powering up to three external displays as long as one of them goes over HDMI (which is capable of hi res but currently caps out at 1920x1200). I wouldn’t bank on the performance being all that great, but it’s doable. 3840x2400 is a little bit less scary to deal with, especially when being helped out by the integrated graphics. It might sound like a small difference, but it’s still a difference.

If we’re going to see a retina iMac, it’s going to be the 21.5”. However, if they push out updates to both the 21.5” and the 27” on the 23rd, don’t be sad if there’s no retina. I’ll bet that we’re still a year or two off on that sort of technology.

Speaking of retina.

Retina MacBooks

There’s already been some rumors about a retina 13” MacBook Pro. I won’t go in to this too much, suffice to say that I’d mark this as plausible. What I find interesting, though, is that this might be the beginning of new 13” MBP’s with discrete graphics cards; something that a lot of people I’ve spoken to have been clamoring for. If I had to make predictions, I’d say discrete graphics with 512 MB or 1GB of video memory, quad core i5 processors, and 4GB-8GB of RAM instead of the 8-16 of its big brother.

And the MacBook Airs? Unlikely. Not just because of their size and the probable need for a nice GPU; but because I think the Air will be poised to take over as the “low end” MacBook. The product line will be moving towards the MacBook Air with non-retina, low-cost displays and the MacBook Pro, powered by the gorgeous Retina display. I think that, over the next few years, the Air will eventually go retina. But it’ll be years down the road, when the displays become so cheap that posturing on the low-end won’t be hampered by the cost anymore. The Airs already have some of the best non-retina pixel densities of any Apple device, so it won’t be too tough to get by. They may just improve the overall quality of the screens in general. It’ll also give integrated graphics technology time to advance far enough that it can handle the displays with aplomb. Either that, or major process shrinkages on discrete graphics.

So what will we see relating to MacBooks on the 23rd? I think the 13” rMBP is a good bet. I don’t know that we’ll see anything happen to the Airs. I think they’ll get bumped at the typical MacBook bump timeframe, in March/April.

Are we really going to see all of this crap alongside an iPad Mini?

Maybe. Gun to my head, I really do think we can expect the following on the 23rd: iPad Mini, major improvements to iBooks, new iMac bodies without retina displays, passing-mentions to 13” rMBP and spec-bumped Mac Mini.

Gun to my family’s heads, if I had to go the conservative route and hedge losses, I think we’ll get desktops and iPad mini, with the laptops punted til the traditional Spring refresh date.

This brings us to the iPad Mini

There isn’t a ton to say that hasn’t already been said. When the rumors first cropped up about a new, smaller iPad, I was a little confused and doubtful. I’d even go so far as to say that I was mostly wrong.

I think I had the right idea. There have been statements that the event on the 23rd will spend a lot of time on iBooks. And this makes sense; I stand by my statement that devices of this form-factor are geared towards consumption and that a strong content ecosystem will drive the success of this market. It’s all of the 7” tablets released up until now, regardless of how much people love them, have all been rather tepid.

The iPad Mini Junior1 will be a full-scale iOS device, because it won’t need to be “hobbled” in order to make the content a first-class citizen. I now believe, pretty firmly, that the form-factor itself begets this artifact. The only two thing that people seem really confused on: price-point/role/niche, and screen.

A lot of people are confused as to what Apple’s game would be by releasing an iPad mini and then pricing it in a way that overlaps with the iPod Touch. And to this, I say: The form-factor begets the role. The geek argument is “Why would anybody buy an iPod Touch when they could buy an iPad Mini for the same price?”. And the answer to that is, they do two different things. Not on paper. On paper, they are both iOS devices. But their form-factors lend them to very distinct roles. Listening to music on an iPad is an awkward notion; when’s the last time you saw somebody at the gym with an iPad strapped to their arm and running earbuds in to their skulls? Conversely, while reading on an iPod Touch isn’t the worst thing in the world, I’d much rather read a book on an iPad than an iPod/iPhone. In other words, fuck the price. Who cares if they cost the same? They don’t do the same thing. The Mac Mini Server is priced similar to an entry level MacBook Air, but they do totally different things.

And the screen? Will Apple really release a new device that isn’t retina? That’s actually a pretty good question. The original bet was that Apple would be cutting iPhone 3G/3GS screens to a new size, and I still feel like this is a good candidate for “Canonical Truth” if only because it will keep the costs down. But there’s very much a part of me that wants to agree with something Marco Arment said on a recent Build and Analyze; I don’t remember which episode, and for that I’m sorry. It’s possible that they could just cut retina displays to a new size, maintaining the same aspect ratio and effective HiDPI resolution. It’s just so weird to think that Apple will release a new iOS device that lacks a retina display. But then, the iPad 2 is still a supported and actively sold device that is non-retina. It’s a tough call to make. But I do think that the Tim Cookiest decision would be to use the 163-ppi screens for the iPad Mini.



  1. With apologies to Dan Benjamin.