The industrial design of the new iMac is expected to be nearly or completely identical to current models, with the possible exception of port placement/arrangement, and there have been some questions as to whether Firewire 400 will remain or be scrapped in favor of a single 800Mbps port.
Beyond its enclosure, however, very little of the new iMacs’ internals will be familiar to those who have looked behind the proverbial curtain with past models.
Sporting a dual-core 2.66GHz Core i7 processor at the entry level and quad-core versions at 2.8 and 2.93GHz, the top-end model will technically give up about 133MHz against the current iMac but with the gain of Nehalem technology and two additional cores as well as 8MB of on-chip L2 cache versus 6MB in current models….even without the other advances, the leap forward in CPU technology alone will push the new iMacs far ahead of even their closest ancestors.
Among many other technical advances which mostly involve the way the processor is put together and components that used to be separate (like the memory controller) that are now on-chip, eliminating major bottlenecks which existed in Core 1/2 based & earlier designs….Nehalem also re-introduces Intel’s “HyperThreading” technology that allows two threads on each core — essentially allowing a quad-core chip to handle eight simultaneous threads which would otherwise require eight physical cores.
This doesn’t always make a large difference in single tasks, but when running many applications and multi-tasking aggressively, the difference is obvious. Finer-grained juggling of threads makes a huge difference in these situations.
Like the Core i7 based Mac Pro that will follow it to market by a few weeks (if our reported timetables hold true; historically, these are the most malleable of all facts “confirmed” at the time they were reported….but in this case, they’re backed up by Intel’s own publicly reported roadmap), the new iMacs will have not simply two memory channels/busses as current Macs do — but rather, three.
In the compact enclosure of the iMac, this is a potentially significant issue; already, 1GB entry-level iMacs only take advantage of one of their two memory channels because Apple decided users would be happier only having to buy a single 1GB SODIMM to get to 2GB instead of having to buy two (as in a 512×512 entry level config that existed in previous generations of Intel iMac)….and adding a third only deepens the issue.
The triple-channel RAM system in Nehalem is one of its key advantages; shipping entry level models with only one of the three channels utilized would be nothing short of crippling it. So….does Apple move up to 1.5 and 3.0GB of memory (512×3 or 1GB x3) in its stock configs or do they go the same route they have with current models?
If so, even the “high end” version, unless they had three tiers (1, 2, 3GB) of stock memory configuration, would still only have two of its three channels utilized.
This is a big deal — a very big deal.
Apparently, just as with the Mac Pro’s huge 12-slot memory banks (3×4; we saw 3×3 9-slot versions in prototypes, but since the new larger enclosure was added to the later protos, we’ve only seen 12-slot versions), there has been a lot of debate between the engineering and industrial-design teams about this issue….and the executives who make the final call have only made things murkier & more confused according to sources.
Thus far, we’ve seen the same 1.5/3GB (in some cases, the full 6GB — 3x2GB — capacity) configs in the iMac prototypes, but that could be merely for debugging of the full three-channel configuration by internal & external developers/testers….it doesn’t necessarily give us any clear idea of what Apple will choose to do with the final product.
Despite the uncertainty about whether the new iMacs will make full use of all three 1066MHz DDR3 SDRAM channels in all stock configs, very little else remains unclear about these remarkable new machines that will be the flagship of Apple’s desktop offerings.
The sources who generously allowed us the opportunity to have a look at this latest prototype via streaming-video conference, also performed a limited number of “stop-watch” benchmarks for us which we will be posting this weekend in a follow-up to this article; please stay tuned for that.
Click through to the next page to see a more detailed list of technical specs beyond what we’ve already mentioned….and if you have questions (other than exact benchmark numbers which we won’t be able to publish for at least another 7-10 days; but “stopwatch” type tests against current machines are nonetheless a possibility), please feel free to send them in whether it be by email, IM, Twitter or web comments here on the site; we’ll do our best to answer them as quickly as possible.