No. I'm assuming that anyone already with a functioning MacPro can probably afford to wait a little while to see if TB is going to come soon, and if it's going to be a major improvement or not.
For a few, this will be the case. But the only users that will be for the moment, are those that need to share their devices with both their workstation (desktop workhorse) and portable systems (i.e. location data gathering = hauling an MP and any necessary external peripherals isn't really feasible).
I don't see any other viable reason TB would be a better solution for workstation/desktop use only ATM, as there's other ways to get fast system I/O (perhaps not quite as convenient, but are faster, and usually offer redundancy for the specific requirements). Specifically for fast networking and storage, as this is what needs to be addressed in workstations and servers.
In the case of transferring GPU data to a monitor, TB is a compromise since some existing video connection standards already offer higher bandwidth.
As it's been stated to be capable of up to 100Gb/s, this may change with some future revision. But I'm dealing with the current revision (Gen. 1.0).
But most importantly, these alternative solutions I speak of are actually available right now (InfiniBand, FC, 10G Ethernet, RAID cards, Flash Drives <unfortunately, none of these work in a MP ATM>, ...).
They already have a system and they have their storage systems.
Maybe.
Fast, redundant storage can reach or even exceed the cost of the MP without much effort, and not be overkill for the requirements. It's also an area that seems to be where independents and SMB's compromise on their hardware in my experience (just take a look at all the RAID threads). That is, they spend their budget on the MP, and don't have sufficient funds for the necessary upgrades needed for what they're doing with it (i.e. under capitalized and/or don't truly realize they haven't filled all their needs as the system comes OTB).
But I believe most MPs in use right now will stay in use for the foreseeable future.
In this regard, I absolutely agree, but it seems for different reasons.
TB for a pure workstation environment (no sharing devices with portable equipment), doesn't offer any advantage over existing PCIe based solutions.
I agree to a point. If Apple is still interested in the MP, then current pricing would seem to indicate that they think the market can bear the price. Of course they have access to data we can only guess at. On other hand, Apple may have decided to abandon the product is now just trying to squeeze the last bit out their investments. Only time will tell.
Absolutely.
Unfortunately, there's a perfect storm of fewer units sold due to rising prices, higher CPU prices from Intel (just look at Intel's roadmap for enterprise parts; larger sockets = more $$$ per part), and other technologies, such as TB, that mean other solutions would be considered a viable alternative to Apple IMO. For example, an iMac equipped with TB would work for software development (what Apple needs internally), modest video editing, and audio production. This would by no means ideal, but it's cheaper to make, and would allow Apple to keep their margins high (which they dearly love, and there's no indication so far they're willing to reduce it).
Other alternatives would be to cease using Xeons, but at that point, it's an iMac with slots (headless Mac users have been begging them for years over). I doubt they'd go this route, as it would certainly cut into the iMac's sales, unless they intentionally priced it high enough to prevent this, and then the same situation would exist with this form of MP/headless Mac (too few sales to be profitable).
I've seen 20Gbits. Or 2 to 4 times faster than USB3. But its also connected to PCI bus, if I understand correctly, which gives it a real-world advantage as well. The advantage is it's simplicity (from the user's point of view).
This is my point.
Systems with PCIe slots can benefit from additional bandwidth not available via TB ports (current part requires 4x Gen 2.0 PCIe lanes; BTW, those 4x lanes are actually good for 2GB/s, but the TB chip can't utilize all of it). Worse yet, some of that is lost in overhead (800MB/s usable out of 1250MB/s, so 36% is overhead).
Now imagine using an 8x or even a 16x lane slot with a card that requires that many lanes... You end up doing this
so much, your face hurts (just make sure you don't drool on the keyboard if you ever get to experience this).
Another reason why MacPro sales may be stalling. If a user is now using a portable with TB, and is using TB devices, they aren't going to be as willing to fork out for a desktop system that they can't use with the TB devices.
For those that don't need more than what a fast laptop can do, you're correct.
But those that do (i.e. need a faster system for the real work), will as they require it (i.e. earning a living, and less time per job = more jobs per unit time = increased bottom line). This will be a small market (compared to the overall computer market), but there's enough independent film makers for example out there that will fuel it (niche market will exist).
Yes. To a point. But those storage devices will also need to be plugged into the workstation at the office.
As it's not portable, and there's no need to be able to carry it around, users' as a whole will accept this for the additional speed and redundancy. So this is a non-argument. Remember, we're talking about workstation users here (part of the enterprise market), and they earn a living with this equipment. Not the avg. non-commercial/home user.
Fair enough (the fewer systems argument). But really... it's just a logic board. Which I know is not trivial to muck about with. But whatever work they do for TB now will also be usable in future versions, so they can spread that cost out.
It's just a logic board, but the changes are more expensive than you imagine (there's additional design time, hardware verification, software validation, new PCB's that have to be manufactured, the existing assembly line has to be re-tooled <such as new programming for the pick n' place systems, and solder paste systems if using ovens for soldering>, and final product verification).
Worst yet, your idea that this will transfer over to Sandy Bridge is false (totally different socket = total redesign).
So figuring all of those costs over a precious few units (and we're not even getting into the issues with existing stock of non-TB equipped MP's), it's not financially viable, as users won't be willing to pay for it.
Better to include it as part of a total redesign than an incremental one given the unit sales volume.
On the PC side, this is a bit different, particularly for vendors that just sell boards (as they're only dealing with their boards, not all of the rest of what's involved for a complete system vendor). They can do it with completed systems, but it's usually part of the 2nd part of the Tick Tock cycle from Intel in order to deal with the costs (use scale of economy to reduce the cost per unit as low as possible, and their production would be higher than MP sales figures).
No arguments here. However, sometimes companies keep a brand going for bragging reasons, as well as a profit centre. Or a loss leader. If Apple were merely breaking even on MacPros, but sold high-margin SW and services way above the average system they might keep the MacPro anyway.
I realize your point.
But I'm don't see Apple doing this. Their focus has clearly shifted to consumer devices, and that's where their "switchers" are coming from these days. So that particular reasoning doesn't hold in this particular situation.
Nor does Apple make a bunch of money on their software products. In fact, that was their loss leader product in order to generate sales of their workstations (FCP for example; at $1k a license, they're not making much if anything on it, and I'd be surprised if they did better than break even with their development costs).
Apple's business model is based on hardware sales, not software. This is why they're not as likely to offer support as long as other vendors, as it helps sell new systems (those that either require or think they can't live without some new feature).
I have to take your word for it. I will just say that whatever work is done to just get TB verified is not lost when they do the SB release. Also, I think Apple has had TB samples for a while now, and well before the official announcement.
To be done with the current model, it is (see above). But by waiting to implement it with a new system design, it's financially viable (why the earliest we'd see it is with the Sandy Bridge equipped MP's).
Putting it with the DP was brilliant. Didn't have to physically shift things around much to make room for another port. I'd forgotten that they had twinned up on the DP.
This is why it's aimed at portable systems (laptops and devices) in it's current form. You may have noticed there's advertisements for TB equipped devices that run on 4G cell networks, so they're coming...
