goMac, I think you are mistaken on this. Apple introduced FireWire 800 on the January 2003 Power Mac G4s. These had faster processors across the line, a new graphics card option in the Radeon 9700, AirPort Extreme (b/g rather than b), the introduction of bluetooth and the superdrives were faster.
Doing it in a single update cycle makes sense from a cost POV, which system vendors are more prone to having to heed than just board makers (those that have total control from design to production, as they have their own engineers and production facilities).
System vendor's don't, as they've done away with that (if they ever had them) in favor of an OEM philosophy, and more recently, ODM (laid off design engineers). All in the name of cutting costs. The "price" of course, is a lack of control, including the ability to offer mid-cycle refreshes that require any PCB rework. They're now limited to things like GPU's, memory capacity, and HDD capacity.
Why? No one has explained to me why this is no longer valid.
I figured you either knew the history since Intel or have at least gone back and looked.
Here's the short version:
- 2006 = Intel switch (system firmware = EFI32; CPU = 51xx series)
- 2007 = First Octad system (new CPU's <53xx series> that were a drop-in replacements for the 2006 systems = no new boards). As it's the same firmware, 2006 owners can drop in the newer CPU's and function.
- 2008 = First EFI64 based system (built on new Xeons <54xx series>, different chipset, and ICH), so there were speed gains due to bus technologies (faster FSB, introduction of PCIe 2.0 lanes for slots 1 and 2).
- 2009 = Nehalem (35xx and 55xx series = first systems to require different CPU's for SP and DP versions)
- 2010 = Westmere (same socket as Nehalem, and only needed a firmware adjustment to use the newer parts which had different steppings). Unfortunately, 2009 owners can't get the new firmware as Apple won't allow it as a means of locking the systems (i.e. forced upgrades down the road).
The situation with Nehalem and Westmere is a result of Intel's Tick-Tock cycle (one socket meant to last 2 years; just need a firmware update to run the newer CPU's when they release). They're also going to continue with this methodology, as it cuts costs on their end as well as system vendors (costs more to deploy a new socket every year than stretch it for longer). The LGA775 and to a lesser extent, the LGA771 were anomalies, not the norm.
You may also want to note, that the time between releases has increased, despite the seemingly 365 day cycle (MR's information should provide the details here). This is a result of the complexity of Xeons vs. consumer grade parts used in all the other systems Apple produces (less complex = shorter development cycle). Keep reading, as I'll go further into this.
I'm simply using the last time Apple added a new I/O port as past experience.
I don't know for sure if this was truly the case (think otherwise actually, but declined comment as I was too lazy to look up G4 and G5's, as I didn't monitor the MP market all that closely back then due to my software requirements meant PPC and OS n wasn't a viable solution to my needs then).
It still isn't, but that's to do with Apple, not the actual hardware (firmware limitations and costs to configure it as I need - and Windows makes the most sense for my software as it allows a single OS).
No one can seem to tell me why that reasoning isn't valid now.
See above.
Apple refreshes their computers every year, otherwise sales begin to dry up, even if it's only a minor refresh. That's how they work. Why would we suddenly have an exception to that rule, when even the Macbook Air and iMacs are getting Thunderbolt?
You can't compare consumer products with enterprise products (laptops, iMac and Mini all use consumer grade CPU's, while the MP and former XServe use/used Xeons = enterprise grade CPU's). They even wait for a normal cycle for their consumer products as well. It just isn't as long a wait, as they're on shorter cycles.
These differing CPU cycles are the result of the additional complexity of the enterprise parts (and getting longer as the complexity continues to increase; adding more engineers can make matters worse <major mess when trying to integrate the sub-component designs for example>, which is why they can't just add bodies to shorten the development cycle).
Apple also sells a lot more of the consumer products than the MP (we're talking orders of magnitude it seems), as the MP has rather small sales. As a result, there's not enough systems to absorb additional R&D costs due to a mid-cycle refresh resulting from a PCB rework, let alone the additional expenses due to discounted inventory <i.e. add the lost margin to the new systems> and breached contract penalties. It would be horrendous, and why they don't do it (not just Apple, but any system vendor would have to deal with the same issues if they considered it, let alone attempted it).
I think this and the differentiation between consumer and enterprise products is where you're getting hung up on.
Does anyone here actually seriously believe the Mac Pro is going to get Thunderbolt at least 8 months after every other Apple product has Thunderbolt?
Absolutely. And it's actually possible that it won't get it then (if they decide they can't afford the solution that gets the DP data to the TB chip).
Especially when Thunderbolt is targeted at pros? And Apple has a major new pro app on the way this year that is targeted at taking advantage of Mac Pros? And we have evidence that the 6970 is on it's way?
When they stated professionals and TB,
they meant those that use portable computers. The press release, particularly the illustrations, clearly showed this and it's why the first products it's showing up in are only laptops and portable devices (not in conjunction with PCIe cards for desktops/workstations).
Everything points to a new Mac Pro revision this year. The only way one would think there is no chance of one is if you buy into this mantra that Apple is somehow magically bound to only release new Mac Pros when Intel releases new chips, which doesn't have any evidence at all backing it, besides circumstantial. Quite frankly, no one can say Apple won't release a new Mac Pro because we haven't yet had a year when Intel doesn't release new Xeons, so we don't know how Apple deals with that situation.
Look at the released roadmap for the next Xeons. Q4 2010. This means that more time is needed for initial production and shipping (components to the board facility, and the finished systems to the distribution locations), which is a quarter minimum (aka lead time, and is 13 weeks minimum).
So we're talking about March 2012 at best. Unfortunately, if you look at Nehalem and Westmere's release, Apple lagged behind other vendors, which doesn't bode well for even hitting that date.
They were the same model of processor, there was just a higher clock rate available. Same situation as now.
They did rev the GPUs, add Bluetooth, and add 802.11g. It still was not a major revision.
It's possible
if there was no PCB rework involved (i.e. all new parts use the existing PCB layout = new parts use the same component packages as their predecessors, won't exceed power draw limit of the voltage regulator, or generate too much noise it interferes with other parts on the board). Unfortunately, this doesn't always work out, and gets much harder when you're talking about multiple parts (odds of all of them fitting the existing layouts are quite a bit lower than that of a single part).
I'm not convinced of this yet, so if it proves to be true, take the above information into careful consideration.
Worst case, if this did happen, and there was a PCB rework, the odds of them loosing their proverbial shirt were good, and would now be seen as a major blunder (Apple likes high margins these days, and haven't given any indication they're going to suddenly change this position).
