Whither the PowerMac?
So, even Apple's best quarter in years didn't extend to the PowerMac.
Apple confessed that they're no longer expecting 200K units per quarter. They said further that they're not concerned with the performance of individual lines, only with aggregate sales. I'm sure that's if it's true, it's true up to the point where the line costs them money—and I'm not really convinced that it is true. I think they're putting a brave face on the wildly variable, but ultimately favorable, results of the last quarter.
This is actually something I predicted way back when I first joined the board, although I was off by a number of years: The tower is a hangover from 30 years ago. Most users simply don't need one. They bought towers because, for the performance they required, there was no other option. Now there are options, and tower sales are in the toilet. Not because pro sales are down, but because a whole lot of pros are better served by something else in Apple's product line.
But clearly, and especially in some markets that Apple is just breaking into, the need for absolute, pedal-to-the-metal power remains.
Since gross margin (and every other kind of margin) actually went up, it's clear that Apple has weaned itself of its long-standing dependence on the towers' profit margins. They saw this coming. And, contrary to the profits vs. marketshare opposition set up so commonly on these boards and elsewhere, they've managed to shift the burden of maintaining the bottom line to precisely those models that are flying out of the Apple Stores. This gives them a lot of flexibility with the high-end models that they've never had before. The pro models can become more specialized, and there's much less pressure on any of them to sell in large quantities.
So, why not break the PowerMac == tower equivalence?
Back in the halcyon days of Future Hardware when I thought we'd have FireWire 3200 by now, I postulated a modular pro line. Need a quiet workstation with a choice of monitor(s)? Get this little box here. Need absolute computational power, and nothing else? Get this bigger box (but still much smaller than a tower) here. Heck, get three of 'em. Need hard drive expansion? Add this little box here. Need more? Add another. Need PCI? This box over here... and so forth. The Mac mini points the way, design-wise, and the connective tissue my vision required is finally here in the form of PCI Express (which can run over a wire) and HyperTransport (which can also run over a wire). There is no longer any particular reason to offer a tower, although of course they could keep one around for people who actually do need everything that a tower offers, and who'd prefer it in one package.
This really could redefine the landscape, completely. Pros could build systems à la carte, swap out the parts the need or don't need without cracking a case open, add new things, etc. As with the mini, there would be fertile ground for resellers, VARs and third parties to offer accessories and solutions. And, as of right now, they could be enhanced with Xsan, racks of Xserves, etc. for file storage and extra muscle for batch work. Apple could accommodate everyone from the lone pro who wants everything crammed into one tower, to the pro in a shop who just wants a quiet little box powerful enough to preview her work before she saves it to the SAN and sends it to the render farm.
This is all, of course, 100% blue sky, and I reserve the right to be horribly off track.
I'm offering all this mostly to stimulate discussion on the future of the PowerMac. Because as it is, it's not prospering.
Apple confessed that they're no longer expecting 200K units per quarter. They said further that they're not concerned with the performance of individual lines, only with aggregate sales. I'm sure that's if it's true, it's true up to the point where the line costs them money—and I'm not really convinced that it is true. I think they're putting a brave face on the wildly variable, but ultimately favorable, results of the last quarter.
This is actually something I predicted way back when I first joined the board, although I was off by a number of years: The tower is a hangover from 30 years ago. Most users simply don't need one. They bought towers because, for the performance they required, there was no other option. Now there are options, and tower sales are in the toilet. Not because pro sales are down, but because a whole lot of pros are better served by something else in Apple's product line.
But clearly, and especially in some markets that Apple is just breaking into, the need for absolute, pedal-to-the-metal power remains.
Since gross margin (and every other kind of margin) actually went up, it's clear that Apple has weaned itself of its long-standing dependence on the towers' profit margins. They saw this coming. And, contrary to the profits vs. marketshare opposition set up so commonly on these boards and elsewhere, they've managed to shift the burden of maintaining the bottom line to precisely those models that are flying out of the Apple Stores. This gives them a lot of flexibility with the high-end models that they've never had before. The pro models can become more specialized, and there's much less pressure on any of them to sell in large quantities.
So, why not break the PowerMac == tower equivalence?
Back in the halcyon days of Future Hardware when I thought we'd have FireWire 3200 by now, I postulated a modular pro line. Need a quiet workstation with a choice of monitor(s)? Get this little box here. Need absolute computational power, and nothing else? Get this bigger box (but still much smaller than a tower) here. Heck, get three of 'em. Need hard drive expansion? Add this little box here. Need more? Add another. Need PCI? This box over here... and so forth. The Mac mini points the way, design-wise, and the connective tissue my vision required is finally here in the form of PCI Express (which can run over a wire) and HyperTransport (which can also run over a wire). There is no longer any particular reason to offer a tower, although of course they could keep one around for people who actually do need everything that a tower offers, and who'd prefer it in one package.
This really could redefine the landscape, completely. Pros could build systems à la carte, swap out the parts the need or don't need without cracking a case open, add new things, etc. As with the mini, there would be fertile ground for resellers, VARs and third parties to offer accessories and solutions. And, as of right now, they could be enhanced with Xsan, racks of Xserves, etc. for file storage and extra muscle for batch work. Apple could accommodate everyone from the lone pro who wants everything crammed into one tower, to the pro in a shop who just wants a quiet little box powerful enough to preview her work before she saves it to the SAN and sends it to the render farm.
This is all, of course, 100% blue sky, and I reserve the right to be horribly off track.
I'm offering all this mostly to stimulate discussion on the future of the PowerMac. Because as it is, it's not prospering.
Comments
As for the future of the towers, I think Apple realized this last March when IBM first showed signs of delivery problems. They probably went to a crash iMac mini development program as a result of this an dramatically ramping iPod sales. That doesn't mean they think it'll last forever though -- what it is waiting for is a compelling improvement to reinvigorate sales. Same applies to the PowerBook, and this boils down to 970FX replacement(s). Give the towers dual dual processors, and intro the PB G5 and the flagging sales may just well climb back up to their traditional levels and possibly beyond.
I don't see any indication that the market has changed, and neither have the PowerMac and PowerBook changed significantly in roughly 2 years which is plenty of time for people who want those products at their current price/performance levels to upgrade. Bump their price/performance and you'll likely see another round of buyers plunk down their credit cards.
Sounds a lot like the SCSI days to me. Except back then the cables were big thick things with big annoying connectors. Today we have FireWire and USB which aren't nearly as obnoxious, but the two kind of compete with each other. A motherboard bus speed external bus could change things.
What might make this workable is if they can link CPU modules with HyperTranport wires. G5 clusters are showing themselves to be very good performers. What if Apple took care of all that connectivity for you? Buy CPU modules for $500-1000 a pop and build your own cluster; Apple technology handles the details.
The obvious stackability of the Mac mini may well point to the future. It will be interesting to see what the Mac mini landscape looks like in 6 months. Will it be a hot little machine with lots of stackable goodies? Or a slow CPU with a lame GPU and limited expansion that folks ignore in favor of laptops?
Nice thoughts, though. Got the speculative juices flowing.
- Jasen.
Many Pro's already have to augment their systems with large network storage and RAID boxen. Why not make the actually physical desktop computer small and efficient and quiet!
Originally posted by Programmer
Interesting notion, but I'm not convinced its terribly feasible. The current tower could probably be divided in half -- seperate the drive bays out into modules which "stack" on top of the main logic section. That main section will has to hold the motherboard with its memory, ports, processors, power supply, and cooling system. The drive bays connect via SATA or FireWire800+, so they can be "at arms length"... and this would also address current concerns about the lack of drive space in the towers.
FW was my connector of choice back in 2000, but at this point PCIe and HT look better to me—depending, of course, on the details about how scalable they are once they're on a wire, which I don't have access to.
That doesn't mean they think it'll last forever though -- what it is waiting for is a compelling improvement to reinvigorate sales. Same applies to the PowerBook, and this boils down to 970FX replacement(s). Give the towers dual dual processors, and intro the PB G5 and the flagging sales may just well climb back up to their traditional levels and possibly beyond.
They might climb up to something, but I really don't see them climbing up too far. At this point, they're basically targeting the UNIX workstation market, which is small in absolute terms (but absolutely crucial, strategically). Consider that, before the iMac, Apple sold over 600,000 PowerMacs per quarter. The more options they've offered, the fewer PMs they've sold, and that's because there are as many professional needs as there are professionals. Not even the initial release of the PMG5 really changed that. It goosed the sales back up from moribund, and more importantly shifted the sales mix from heavily favoring the low-end model to heavily favoring the high-end model, but the general downward trend has continued steadily.
Even if Apple releases the PowerMac Godzilla at, say, WWDC, there will still be people who want the power but not the expansion, or the expansion but not the power, or one kind of expansion but not the other, or none of the above, but they do still want to use their Cinema Displays. Those whose needs aren't served by other lines don't all need a big, hot tower, and I think the pieces are falling into place to diversify the line somewhat.
I remember the head of a graphic design firm saying that the Cube would have been the ideal workstation but for the price. He'd have bought hundreds of them. Now, some of that demand has been soaked up by the 20" iMac, but I'd bet that pros wouldn't begrudge a PowerMac mini...
To the best of my knowledge HyperTransport is not suitable for cables, from the way things have gone it does not even look like it is a good technology for internal expansion card interfaces. Even the company that is behind it (NVidia) has only ever proposed it as a method of connecting systems on the same piece of PCB. NVidia has used it to connect the North- and South-bridges on their motherboards and as an intra-component buss on their video cards.
My guess is that we will not see this style of expansion, at least not in the next 10 years. I think we will see the trend of less-and-less need for expansion cards altogether as things don't need that much bandwidth (in comparison) to communicate with the "host". Most of the old uses of expansion cards have either moved onto the motherboard or onto "connection" busses such as USB or FireWire.
The last hurrah of the expansion slot will be the replacement for PC Cards that is being demo'ed at CES, but I don't think that it will ever see widespread use.
Your point is well taken that internal expansion itself is on its way out, which simplifies things somewhat. But in the near term, there are still some very expensive and very crucial capture and processing cards that some people simply need.
So, what are the available connectors? I see now why Programmer mentioned FW, because it's suddenly a leading contender, and he also mentioned SATA? The other options (10Gb Ethernet, Fibre Channel) are too expensive and rare right now. Apple does have a history of pulling such technologies down out of the stratosphere and into the PowerMac, but I don't see them doing that with either tech quite yet.
So, let's say that anyone who wants PCI expansion will end up with a tower. Done right, they're the easiest things to cool anyway, and since PCI represents a fairly open-ended cooling challenge, that's important.
So we're left with computing boxes and local storage boxes. Anything else? Maybe lower end boxes that support single monitors, and higher end boxes that support duals? Or onboard video vs. AGP/PCI-e slots? Anyone else more inclined to believe that the tower will remain the way for the PM to go?
I think this is ultimately going to be more about marketing than good technology sense.
Most people buy more computer than they need... this is clear. I help set up medical practices from time to time, and they are always thrusting computer advertisements about and asking how much computing power they will *need*. Need? They use notepad, IE, and a telnet application! They don't need a computer, they need a $300 workstation. Video card? Ghz? Modular tech solution? Bah! Yet they talk endlessly, balancing prices and evaluating features with sales people. They buy professional machines. They spend 1.5k-2.5k, happily.
I guess I'd argue that business success is only tenuously tied to meeting the technical needs of computer buyers. People (like me) buy PowerMacs because we want the power, not because we need it. Even the fancier stuff I want to do (like play the newest games and edit movies) could be done just fine on an iMac. (I'm guessing that's what you meant above when you said professionals are meeting their needs in other ways). And really... I want the PowerMac because it has been succesfully marketted to me. I don't just want to edit movies... I want to edit movies quickly! Like a pro! With the best consumer movie editing machine on the market! Yeah!
I love your idea of modularity. It would be extremely cool to mix and match technical needs with greater control or figure out ways of adding multiple processors to my existing rig when and if I needed them. But a cool technical idea may or may not lead to success in the desktop marketplace. Not when my grandmother buys a 2Ghz Pentium IV so she can write letters in Notepad.
Marketing. If Apple wants the PowerMac tower to make a comeback (and their latest intiatives are definitely, and perhaps wisely, pointed elsewhere!), it will be all about Marketing... not meeting the needs of the relatively few customers technically savvy enough to understand them.
I talk too much. And repeat myself.
-bb
I am one of them.
I was excited when the G5 was first announced after watching the demo, but after the independent tests were done I was glad I waited until I saw the results of the test before buying
It's also because of Processor constraints. I went through this with the G4, and a person working in 3D can not wait and watch the everybody's computer get faster like what happened with the G4 @ 500MHz for 2 years.
Now Apple is forced to put a radiator in their PowerMac to keep it cool enough for an upgrade.
The signs of past computing hardships haunt many of us. Buying a new PowerMac isn't in the cards for me right now because I can get a much faster Pro computer elsewhere.
If Apple started treating the PowerMac like a Pro machine in competition it would sell again.
The problem with the current tower is not so much it's size but it's lack of umph for that size. As mentioned many times on these boards the current towers should use the extra space for two more drives and an additional optical drive.
I've never been a fan of external add ons. I tend to like things in one box that sits on or under the table. It's clean, in one place and there are too many cables as is. I am loath to add external peripherals like HD's, Optical drives, etc. For Professional Pixel Pushers, Video, Audio, Multimedia, etc there is a desire and a need to have multiple drives and two optical drives all in one box. I don't see this going away any time soon.
i mean, if you were going to jump on the G5 bandwagon, you might have already done so, so will an imminent jump to dual 3GHz get a lot of sales? i'm not so sure.
this all reminds me that i need to start getting into high-end video stuff to warrant a large budget request increase for our department next year.
I'm also defending the need for towers by some of us as well. Video, and 3D for sure. I believe most 2D work can settle for a regular OpenGL based graphics card like the ones that Nvidia, and ATI already have made available.
But the lack of higher-End specialty graphics cards does leave some of us out in the cold using a Mac. (it's disappointing)
I like to keep my options open. I have not yet bought my PC, (next month after tax return) but I do NOT intend to hook it up to the internet. It's going to be a 3D work machine only.
I know the OS is sh*t, but I do know how to maintain a computer. Even if it runs windows.
Originally posted by rok
...will an imminent jump to dual 3GHz get a lot of sales?
It all boils down to individual replacement schedule and money. You will not completely understand unless you are in need of a replacement at an off time like this. I guess it's like birth? You don't remember until it's time for the next child birth.
Why would we all just thumb our noses at a stupid little speed bump? If I need a new computer today and it coincides with the time of expected revised or upgraded pieces, why would you not understand the dilemma? There's only 1 real answer. MONEY! Either you're going to save money ($500?), or you'll be able to buy more today (Faster CPU/$500 worth) then you did yesterday with the same exact money. There are also benficial tweaks made at every revision/speed bump (PCIexpress, faster ram, faster fsb, radiator deletion for more pci slots or hard drive space.) Maybe it's something less sexy?
Whether it's my employer's money or my own personal, the game is about maximizing investment dollar over a extended period of time. The apple train will always save you up to $500 every 6-8 months.
You'll understand again...
Cheers,
_M_
Prince of run-on sentences.
http://buyersguide.macrumors.com/
Originally posted by mikenap
I'm bumbed about this news that apple does not expect strong sales from powermacs, and by onlookers needing to swith to a PC to get the performance he needs. I work in large format color printing, and use a dual 1 gig quicksliver. The machine is a workhorse, but is not filling the bill performance wise. With photoshops increasing prowes in handle text, we have been getting more and more 2 gig photoshop files instead of quark docs with placed images for poster and trade show booth printing. I am also constantly ripping open quark generated eps files to deal with printing problems. I have saved my pennies and am ready to get a monster PM tower with loads of ram (next update will have to do) but it's dishearning to see my segment, that craves speed and expandability, the old core of the apple universe, seemingly ready to slide into non relevance. Using PC's that freak out and break daily is not an option for me, I have 15 apps going full spead all day long, and my business depends on uptime and billable hours. the pc's in our shop are fast, but geeze, there down more than up. I want Apple to make me a freakin' monster! I can hope at least...
I was in the same boat a while back. I never thought I'd see the day I'd build a billboard in Photoshop, but we do it all the time now. I had a 933 not-so-Quick-silver at work with about 1.5gigs of RAM. When the G5 first came out I excitedly went to the local Apple store to test it. Even with additional RAM installed, the G5 in the store didn't perform much better for everyday tasks such as resize, convert to CMYK, and even just the time it takes to save a large file. Yes, applying a blur was faster, some filters were faster, but not enough to justify a jump to an entirely new class of processor.
Does anyone remember the jump from 604/e to the G3s? Now that was a HUGE jump in real-world performance. You'd have been a fool to keep plugging away on an old 603/4 when there were G3s out there. PC computers experienced a similar jump going from 486 to Pentium. Intel, IBM, Motorola, AMD and company need to provide the computer industry with another leap in processing power.
Back to my story, though... What I decided to do instead of getting a new G5 was to add some SCSI drives (it pained me to even think of it - "What a throwback," I thought). I added one Ultra160 single-channel card. On it I put an internal 15K RPM 18gig drive for the OS and applications. I then put a dual-channel Ultra160 card in it and ran a RAID 0 with two external drives in an enclosure (7200 RPM UltraSCSI Barracuda leftovers from our video department's old Avid). I then used the original 60gig UltraATA drive for file storage and added a 160. The RAID is used for the Photoshop and Illustrator scratch disks... I can't tell you how incredibly well this worked speeding up production. The 15K RPM drive with OSX just SCREAMS, and the Photoshop RAID scratch cuts though multi-gig files like a dream. I was impressed by how fast the UltraATA drives seemed to respond once the bandwidth used by the OS was removed from the ATA chain.
The new Apple towers should all be using faster drives - 7200 RPM is just so 2003.
I have some other points, but this post is already too long.
Originally posted by __MaGnUsSoN__
There's only 1 real answer. MONEY! Either you're going to save money ($500?), or you'll be able to buy more today (Faster CPU/$500 worth) then you did yesterday with the same exact money.
...
Whether it's my employer's money or my own personal, the game is about maximizing investment dollar over a extended period of time. The apple train will always save you up to $500 every 6-8 months.
YES!
He's really hit the nail on the head here, for me.
With the PowerMacs being 220 days without an update or price cut, where is the price attrition of the ever advancing computer world? Waiting six months is supposed to improve my investment, not leave me looking at the same models at the same price point. I have to assume other people looking at the PowerMacs feel the same... its a big purchase, and right now it looks like yesterdays speed at a steady price. But I really need to get this computer by early February, so I'm boxed in. I'm begging for buyer's remorse.
Originally posted by Amorph
So, what are the available connectors? I see now why Programmer mentioned FW, because it's suddenly a leading contender, and he also mentioned SATA? The other options (10Gb Ethernet, Fibre Channel) are too expensive and rare right now. Apple does have a history of pulling such technologies down out of the stratosphere and into the PowerMac, but I don't see them doing that with either tech quite yet.
To play the other side for a moment...
There are cables, and then there are cables, and then there are connectors. I'm reminded of a component stereo system that came out back in the early '80s which eschewed all the wires between components for a plug/socket arrangement on the top and bottom of each unit. What is possible across such a design is quite different than what is possible across multi-foot long flexible serial cables that daisychain between devices. While PCIe and HT might not be feasable for the kinds of things that FW can do, it might be possible to design a modular case system where drives, slot bays, motherboard(s), and power supply can be interconnected in this kind of a fashion so that you can build-your-own-tower in a lego-like fashion.
LegoMac.
The trouble with this arrangement is that it is going to fall short of what is possible across the tight high-speed traces of a single high quality motherboard. As a result the multiple processors, memory subsystem, and maybe even the GPU will all live in the main motherboard module for the foreseeable future... unless you are willing to give up the potential for maximum bandwidth and minimum latency. This might not be bad at all since most of what Amorph is suggesting should be modular is the peripheral hardware.
With advances in the OS (e.g. NUMA support SGI-style) and clustering technology it might even make sense to use this mechanism to allow people to build a cluster-within-a-tower by using multiple motherboard modules, mixing and matching different versions of the motherboard module as they go (i.e. start with a single 2 GHz 970FX, add a dual dual 2.5 GHz 975GX a year later, and tack on a quad 8-way 5 GHz MacCell Extreme three years after that).