Originally Posted by winterspan
Not a chance. The only Nehalem chips out right now are the high-end (high power) desktop "Bloomfield" chips with Quickpath that use the X58 chipset. The wholesale cost of the cheapest i7 and motherboard together would probably be close to $500 alone! The mini will most likely continue to use mobile chips for their low power dissipation, and quad-core laptop "Clarksfield" chips won't be out until Q3 2009. The only dual-core Nehalem chips, both mobile and desktop, are the ones with graphics cores integrated onto the CPU package and both have been delayed until 2010.
It is more a question of what will be out at the time the Mini is introduced. It is agreed that the current chips are expensive but that does have to hold at production time. Frankly All Apple would have to do is to agree to clock a special run of i7s a little slower than normal and have intel put a unique part number on them. Let's face it Intel has bent over backwards for Apple already in this reguard.
As to chip set yeah as we currently understand it the chip selection is limited but that doesn't mean that Apple couldn't roll their own. It is not like they don't have experience here to do so. The could even bridge to hyperteansport to make use of GPUs supporting that interface.
I know i7 is a stretch for early next year but it is the platform to go to for a new long lasting generation of Minis.
- Most likely Apple will simply use the Macbook's single-chip nVidia 9400 chipset in the Mini, as they love to share parts to reduce costs. Also, this is a good chipset, and besides nVidia not (publicly) having anything better than the 9400 for an integrated graphics chipset, the Mini never had better graphics than the Macbook in the past so I don't see why they would start that now.
Nvidia has already said that the 9400M started out as a desktop concept and that Apple refocused efforts on the Laptop market. So I take this to mean that what is in the 9400M is a cut down implementation of a larger chip. It also appears that the execution units are modular enough that a few could easily be added to a desktop chip while keeping power managable. Now this is more hope than anything but it is interesting how strong the collaboration between Apple and Nvidia is.
In any event yes Apple likes to manage costs by keeping parts interchangeable across platforms. That is very important for a low volume manufacture. That however isn't what Apple is anymore especially with the mac Mini.
- SSD in a budget computer? First of all, given the price of the Mini, it certainly couldn't be a decent one. Second, the primary benefits of an SSD apply to laptops (low power, durability) and servers (very high random read speed) -- there really wouldn't be a compelling advantage to having one in a budget desktop.
My guess is they will simply continue to use a conventional 2.5" laptop harddrive, which is the cheapest and most sensical solution.
It is a given that SSD are a bit expensive in the larger size right now but let's look at this as a potential hybrid approach. Say a 32 to 64 GB flash storage system was implemented as a system code storage pool and user and log data went to magnetic storage. The flash drive then only needs to be optimized for reads this controlling costs. Of course this won't work well if you are a user with tons of apps and other code installed but for many users it would be a low cost solution to much better performance. Code segments and apps in general would load much faster. Sure we are talking crappy writes but how often is software installed or updated?
- As I mentioned above, an Intel Atom is a major step back from *any* Core (2) Duo.. even an old ultra-low-voltage Pentium M can blow it away. There is simply no compelling reason to use one, other than for Apple to increase their margins.. and I don't believe even they are that cynical.
Atom would be good for a Mac Nano! Let's face it for certain types of servers Atom would be fine. It might also be fine for task specific Mac applications. It won't however deliver the performance that Mini users want.
- i7/Nehalem will also never happen, for the reasons I outlined above. A Nehalem fit for use in a Mini (cheap, low-power, and integrated graphics) won't happen for at LEAST another 12 months from now.
Twelve months is a long time. There are however several things that Apple could do here. One is roll their own support chip something they have had a lot of experience with on PPC. The could also cut a deal with intel for a "special run" which would be nothing more than a relabeled special run that is undercoocked with respect to the high performance chips on the market at the time.
Maybe i7 doesn't mean a $500 computer but it doesn't mean impossible either.
- Including an expresscard slot is the best thing they could do for the Mini, even a current generation expresscard 1.0. This expansion slot would allow easy expandability in the future for Firewire 800/3200, USB 3.0, eSATA, Digital TV tuners, etc. This would be especially relevant if they decided to remove the standard firewire port.
Express Card is certainly a possibility but frankly I see it as less than perfect. Mostly because of two things. One is the rather fragile mechanical housing that the cards are often inserted in. The other is the lack of lanes. A mezzazine card can be more robust if a bit fiddly to install.
What ever they go with it needs to provide a quality mechanical and electrical interface.
Well technically Nehalem/i7 was really about increasing the performance of everything BUT the processor core. The processor core itself has been enhanced with hyper-threading, but it's really not much different than the core from Core 2. It is everything else in the platform that has been changed. The memory controller has been moved onto the processor die, all the cores are on a single die (like AMD K10), there is a new 3-level caching system, and the ancient FSB has been replaced with Quickpath.
Which from Apples standpoint makes i7 a lot like AMD and PPC processors. Also quickpath ought to be easier to work with as all the memory traffic gets it's own path. Frankly quickpath is exactly why I could see Apple adopting i7 early. They have slot of experience themselves and they have PA Semi. If anybody could successfully introduce an i7 chipset for relatively low cost machines it is Apple.
Moreso if they are working with Nvidia we could see Apple being the first to deliver a two chip i7 solution. Now that may seem like grasping at straws but there are Apple excutives on reccord as saying the will soon have hardware on the market others can't compete with. I'm fairly certain we haven't seen that hardware yet. So is it really that much of a stretch to think that Apple would go i7 in a big way.
the Mini, as I mentioned above, it will not see a Nehalem/i7 CPU for some time. The first mobile CPUs (that is, the first CPUs cool enough for a Mini) are quad-core and should be out Q3 2009. Sometime after that, perhaps in 2010, the dual-core graphics-core-on-the-CPU version will be out.
You are making an assumption here that cool desktop chips won't be out early next year. You also assume that Apple would want to run them flat out in the Mini, it is not like Apple doesn't have options here.
As to the integrated GPU version I don't think Apple is all that interested as they look like they have gone all in with respect to OpenCL. As such they will continue to leverage the better performing solutions.
One more thing, because the Nehalem platform is very different than Core 2 it uses a completely different socket. Therefore, there is now way for a new Core 2 Mini to be able to have future drop-in capability for Nehalem.
Well that might be useful but still I think an under clocked i7 running at say 2.5GHz would give Apple a better more salable machine. As cooler chips come out they would be able to easily bump up the clock speed while otherwise keeping the execution environment constant.
I know it is a stretch but a Mini replacement that leads off with i7 would create a massive storm of interest.