Yep, this was the case when Apple shifted from PPC to 32-bit Intel X86 CPUs when only a generation later 64 bit CPUs were introduced and when Apple could no longer use the NV 9400M GPU [edit: and had to step back to Intel iGPUs] and that is what this next generation will be-- just "good enough"--a step backwards but close enough that the average consumer won't notice. So much for Apple "excellence." It is not a good time to be someone who needs to upgrade hardware.
I find it a little ridiculous that the mentality is applied to their most expensive hardware, but it is what it is. I'm not sure what percentage of people buy the 15" for its size (as the 13" is small) as opposed to its specs or a combination of the two. In the case of intel Apple those used in the mac pro 1,1 were x86-x64 cpus. They just didn't have a 64 bit efi for that generation of machines. Comparable hardware can run Windows 7 64 bit.
In the case of first-generation
portables, as well as the Mac Mini (and iMac?), Apple used 32 bit X86 CPUs. As I have illustrated, Apple has on several recent occasions taken a step back in performance whether out of necessity or to push a new direction for hardware that Apple has chosen.
Dedicated GPU vs. Integrated GPU.
For my uses, the biggest difference is Dedicated RAM vs. Shared RAM.
It's not the computational power, it's that if I get 8GB of RAM for the CPU, I don't want the GPU siphoning some of that away in applications that are both CPU and GPU intensively and that use all the RAM they can get, like 3D CAD software.
I like the compartmentalized way of a Dedicated GPU with it's Dedicated RAM; one doesn't intrude the other.
Yes. People who haven't used both will not notice or appreciate the performance difference.
Moreover, when Apple makes the upcoming Intel iGPUs incompatible with their OS in three years, people will begin to understand what Apple has sacrificed. Example: My 2008 BlackBook is stuck with either 10.6 or at the latest 10.7. It is incapable of running anything later than 10.7; however my father's 2007 MBP is able to run 10.8 and even the upcoming 10.9. The difference is the graphics chip--my MB has an iGPU and my father's MBP has a dGPU. People championing iGPUs should consider their need for hardware longevity before getting so excited.
Apple what are you doing? PRO without a dedicated GPU? No way i would pay over 2000$ for it.
I wouldnt' pay over $1500 for a MBP with an iGPU. If this rumor comes to fruition, demand for refurbs/used 2012 portables is going to skyrocket, as will their value. There is no real performance difference between Ivy Bridge and Haswell, only gain in battery life. Those who prefer or need dGPU performance will flock to whatever stock remains of the current crop of portables, and I predict that not only a few will find themselves in this predicament. As others have pointed out, Apple's version of the Iris Pro will have to achieve parity with the latest NV and ATI cards, not last year's NV 650M (which the current IP can't even match) or Apple's souped-up 660M equivalent (which uses faster GDDR5 instead of the stock GDDR3 found in the 650M). Time to start looking at the current models before this new line gets released.
Any possibility of 15" retina MBA in the next two years?
Yeah, it is coming in the next generation rMBP.
[Edit: Added Post Below]
I couldn't agree with you more. I was very disappointed when the nMP was announced at WWDC. I'm not buying one and will keep my 2010 Hexacore MP for several more years.
But this is the direction Apple is going. I think in two to three years it will make sense but in this transition period it's very frustrating. I hate external drives, having to worry about losing them, having them become disconnected, power supplies, etc.
Great point. Correlatively, once it is possible to use external GPUs, I think using iGPUs only in Apple's portables will make perfect sense; however, as has been demonstrated by other members in this thread, even TB 2.0 will not have enough bandwidth to make using external GPUs plausible without significantly reducing the performance of the external graphics card. While I can envision a day when this is possible, this is way too premature to attempt this now. Perhaps it will be possible when Intel introduces PCI-e 4.0 with Skylake in 2015.