... . However, as far as the vast majority of users are concerned, I think making the system a dual socket Xeon with a single GPU would have been better than the current single socket CPU with dual GPU config.
There is little to no quantitative evidence that two socket systems outsold the single sockets in earlier models. Duals where over represented in the 2007 era because that was the only way to get to 8 cores. Once could get to 6-8 single socket. If the overwhelming vast majority of users bought such systems then Apple would probably still be making them.
"... In the past six years, the SFF workstation has been the fastest-growing machine type, now accounting for roughly 40% of all entry-class deskside workstations shipped (source: Jon Peddie Research). ..."
http://www.cadalyst.com/hardware/workstations/next-generation-small–form-factor-sff-workstations-26262
That's one reason why the single socket 6950X (covered in the review linked below ) costs more than 10 cores than buying two 8-10 cores Xeon E5 v4's. People are going to pay for the more compact form factor ( and the single threaded performance is higher also when just need that).
Apple isn't aiming at 100% of that same SFF market ( smaller still enclosure , but keeping more of the E5 and custom GPU focus). But the notion that the larger collective isn't moving toward smaller ( via laptop or more comfortable fits on the desk" ) doesn't measure out if do broad spectrum sampling.
For folks with "lowest common denominator" software even the single socket E5 has some limitations.
In the adobe benchmarks here:
http://www.tomshardware.com/reviews/intel-core-i7-broadwell-e-6950x-6900k-6850k-6800k,4587-5.html
the Core i 6700K is coming out in front in several of these. If following the "most users usage" then 4 x86 cores is plenty. Also points out why Intel is more than a bit goofy to kneecap the 1620-1630 at probably artificially low base clock rates for those price points. Yes, in the rendering corner cases the cores do pan out. However, if option of After Effects on x86 and after effects on GPGPU where the software quality is the same who is choosing CPU?
For all the noise that's been made about GPGPU computing, it still hasn't really replaced pure CPU muscle in my view. The second GPU in the Mac Pro seems to often sit as a case warmer in the system.
Even the pure CPU muscle though is coming to a single slot. v2 maxed at 12 cores. v4 maxes out at 22
http://ark.intel.com/products/family/91287/Intel-Xeon-Processor-E5-v4-Family#@Server
v5 will have 22 that isn't quite so ridiculously priced. Over time the "2 socket NUMA implementation" is being implemented with a single socket. The number of folks who are sitting on relatively static software that scales past 15+ x86 cores is shrinking; not growing. The individual cores are better and the performance is "good enough" for a larger set of folks. There is a set of folks who "there can be not enough cores", but they are never were in the majority of users and that number is shrinking of whom need a 100% time committed used on only one individual system physically on/under their desk.
That said Apple hasn't done as good a job as they should have in getting more software deploy that takes full advantage of the hardware deployed ( both in Mac Pro and other Mac over last 2 years that were reasonably GPGPU capable. )
[doublepost=1465239789][/doublepost]
Great... so we'll still get Macbook "pro" with intel graphics...
if it is latest generation Iris Pro with eDRAM acceleration the performance isn't going to 'horrible'. ( should be sufficient to drive a 5K TB display/docking station for mainstream MBP work. ) In terms of space and cost efficiencies, Nvidia has nothing to offer for the bulk of the MBP line up. Performance wise AMD still has problems from a holistic x86+GPU perspective.
There is only one standard model with a dGPU for the last year (or more) [ upper end 15" ] :
http://www.apple.com/shop/buy-mac/macbook-pro
The reality is more so on how soon Apple drops that outlier rather than if the dGPUs with take back any design wins they've already lost out on in previous years (or were never in).
That trend wasn't going to reverse with anything Nvidia/AMD is offering the short term. Even with the move to HBM ( GPU+Memory all on a chip to get somewhat competitive in space efficient dimension ) that isn't close to moving into the mobile landscape any time soon.
Intel is the largest GPU vendor that Apple (and most of the market ) deals with. It has been that way for several years now.
