I'm sorry but are you campaigning for a Xeon all-in-one?
No, I am dispelling the notion that an iMac with substantially more performance and accessibility can't be created without destroy the basics of the design.
Yes, they could do it. But why? The whole point is to keep any semblance of modularity (if that is a word).
The arguments about how the iMac cannot implement every feature of the Mac Pro are circular. If pile up all the the features of the Mac Pro as design requirements you will end up with something very Mac Pro like when done.
The "why"? If Apple decides the Mac Pro class of machines are not a viable market for them then the "why" is extremely simple. With the Mac Pro gone, there is now an
empty gap in the over $2K Mac Product line up. Apple can peel off a higher number of the ex-Mac Pro users with a Xeon E3 iMac that has more of the features Mac Pro users grumble about ( 'better screen', 'easier replaceable drives' , 'better GPU' , etc.) than just letting all of those folks leave. The objective is not to get all of them. The objective is to get some of them while maximizing reuse of another Mac model to keep tight control on R&D cost increase. The bulk of the R&D savings from the cancellations would be plowed into another Mac Model that has higher chances of average-to-above-avg growth
and profits. At the same time they are still buying the more expensive versions of CPU/GPUs from Intel to maintain that relationship at similar levels as they currently have. Same with LCD vendors (more lucrative parts being sold through to the customer)
An iMac with:
Ivy Bridge Xeon E3 quad 3.4, 3.5, and 3.7GHz (
http://www.cpu-world.com/news_2012/2012020701_Prices_of_Xeon_E3-1200_v2_CPUs.html )
27" high gamut screen with antiglare options
2 Thunderbolt (TB) ports
one 10GbE port and one 1GbE port
two drive sleds/cages that are accessible being a panel
a GPU power budget of around 75-80W (highest end mobile, entry desktop range )
would be competitive with the entry level E5 1600 quad (1620 3.6GHz) on the market. Same number of cores and similar clock speeds (give up some cache and substantive amount of memory bandwidth which will impact some workloads). There is a high speed network port built in so don't need to use slot. Similarly, TB compensates for some x2-x4 PCI-e cards.
Is that going to work for folks would need two card GPU+GPGPU combos? No. For multiple 10GbE ports? No. Internal RAID 5. No. Need 12 cores. No. It doesn't have to.
Should Apple cancel a viable Mac Pro platform solely because they could make the above "performance" iMac? No. With some improvements:
E5 1600's so that have both a high clocker (3.6GHz) and sub $3K 6 core box.
Better case so that not gratuitously rack hostile.
More cost effective paths to 32GB RAM.
built in 6Gbs SATA channels
they could almost certainly sell more than they have over the past 2 years where the entry models have been hampered, increasingly had to cover some XServe contexts, and increasing have to deal with 64-bit apps that will use large blocks of memory. Those are far greater issues/blockers than lack of Thunderbolt. Between pushing the prices closer to the $2K border and adding TB ..... pushing the prices closer to the border is more important to long term model survivability.
There are a couple of external factors that Apple didn't have 100% control over that have suppressed Mac Pro sales over last 2 years. It is almost certainly worth one more shot at trying to get a better picture of whether it is a viable growth+profits market before terminating it. Probably a better option than a MBA 15" (where flush performance and sockets just to do yet another maximally thin box. ). However, if Apple releases an updated Mac Pro and the sales don't jump off the historical charts then Apple certainly has options. An iMac derivative device to fill the $2-3K range is an extremely viable option if that gap opens up.