LightPeak may well change the CONFIGURATION of most desktop / workstations. That would not be an abandonment of the "pro" market but simply applying new technology (LightPeak) to the pro market.
If it were applied to a workstation, you'd be correct. But keep in mind, one of the aspects of a workstation is the expandability (PCIe slots and separate drive and graphics cards which are user upgradable/replacable).
An AIO (iMac in Apple's case), isn't the same. No slots, the graphics are integrated onto the board, and it's not meant to be accessed internally by the user. Now adding LightPeak can allow for a faster interface (as an AIO's expansion is via external devices), but even that's limited compared to PCIe (bandwidth, as PCIe gen 2.0 is capable of 500MB/s per lane, so a 16x slot is capable of up to 8GB/s, as LP is 10Gb/s = 1.25GB/s). That's quite a difference, and may matter in the workstation environment, especially for graphics.
As LP matures, it will get faster (theoretical limit claimed is 100Gb/s), but it won't release at that speed, and will be increased incrementally over time.
As to Intel's roadmap & Xeons - there are going to be low / medium / high end cpu performers. You may be able to wedge a high performing cpu into an iMac like case - but it would still be a high end pro computer - thus the market was not abandoned - just a very different configuration.
Traditionally, workstations use Xeon parts, though I have seen high end desktops substituted for SP models for cost reasons (and no need of ECC RAM).
These two markets are merging though, based on Intel's road maps. And I don't actually forsee problems with this for the most part. Perhaps a few exceptions, where ECC memory is truly needed, but it's not the majority of systems (there's been no choice in the past, as the Xeon line's been designed to work with ECC; standard memory of the same type, wasn't an option). That's changed now. You can get a Nehalem or Gulftown equiped system with standard or ECC memory (UDIMM or RDIMM).
So technically, it's possible to continue on with a tower system (and what other vendors will end up doing).
But Apple doesn't have the sales volume of other vendors in the enterprise market (workstation and servers), which will cause them to look at it from a financial POV (they're a business intent on making money afterall...).
Now keep in mind the fact Apple likes AIO systems (MP's and XServes are the bastard children of Apple, as they don't get the attention they once did). Also take into account the recent bugs in the MP line (not completely solved in the case of the audio issues, and the firmware limitations aren't ever going to be fixed - namely the memory frequency is fixed to 1066, even when CPU's that can run 1333 are used).
They can also save money by moving the worstation users to LP equiped iMacs, as it cuts the R&D spent on MP's and XServes alltogether, and by doing so, increases the sales volume of the iMac. That makes the production cost that much lower, and the margins higher as a result.
And Apple loves high margins.
Also, the pro market will have monitor demands that will not likely be satisfied through a dedicated iMac screen.
I absolutely agree. But it already has the ability to use a separate monitor, and Apple sees this as a compromise for such users (current models). Then with LP, it could be possible to run a multiple monitor configuration as well (built-in screen, second off of the back of the iMac, and from an LP equiped graphics card for example).
There would be bandwidth issues with LP, but it might be possible to include multiple LP ports and team them as you can do with NIC ports. I'd have to go back and look, as I can't recall if this ability has been designed into the specification or not.
I just don't see an abandonment of the pro market. Apple is both creating new markets and competing very profitably in existing markets that they choose.
Ultimately, I do NOT see an LP equiped iMac as a true replacement (mainly due to the bandwidth difference between LP and PCIe slots /= true replacement for PCIe's expandability, even forgetting the mess of external devices and cables). But only that Apple might see it as a potential replacement product, particularly for their own needs for in-house software development.
What I'm seeing, is a focus shift from the workstation users that kept them alive since ~'97 to a portable and device oriented company. That's where the real money is, and thier published sales data and recent product releases support this.
Are there options for them to stay in the professional worktation market?
Yes. "But will they?" is the real question. It has to be profitable, and it's not generating the profits that the device segment does.
Wasn't light peak first demonstrated on a prototype Mac Pro of some sort? Apple has been behind the push for light peak from the start and some rumor mills have light peak coming out this fall. Is it possible the 2010 Mac Pro update is being held back so it can incorporate the new technology as well as be a part of its official launch?
What was shown by Intel publicly was. Now whether or not OS X was the first OS they ever tested it under, I don't know.
Parts are supposed to begin to release to the supply chain Q4 2010 (assuming no delays). So it won't be until 2011 before we see any systems with it (even without any delays).
The history isn't so clear, but the consortiium that developed LP doesn't include Apple. All the work has been done by Intel and the other partners (see Wiki if you're interested as a starting point). So there's some question as to the extent of Apple's actual involvement.
...you don't want to waste a PCIe slot for it?
That's a good enough reason for some, as 4 slots can be a limiting factor for some. More so, as there's no configuration utility as there was in the '06 - '07 models that could assist with specific situations.
