We sure install a lot of Cat 5 and Cat 6 cable and data jacks in new businesses, stores, schools, hospitals, etc for ethernet to be going away like some people think. Sure we install more wireless nodes now but we haven't seen any business willing to trust wireless completely.
Cat 5.....why are you still installing Cat 5, Cat 6, I've heard some building going all fiber (not economical).
Cheap is the reason for it. When I said conservative in these things, I was referring to their actual behavior. They're slow to adopt anything new unless it will push large numbers of units. Given that Ivy is a softer refresh, things such as the rMBP and usb3 help push sales in what might otherwise be a slower year. While they could be more proactive about adoption, I kind of expected this. They'll have to engineer some kind of driver stack either way for at least one third party chipset if the thunderbolt display will take on usb3. It may not happen immediately, but it will happen.
While that may be disappointing, it would be in line with a general spec bump to bring the current model to 2012 hardware. USB3 and the HD4000 are compelling there. Two TB ports would be great, but I doubt it. this would allow your choice of display on its own port without more costly hardware to split the connection. I mention that as this is somewhat of a budget machine in Apple terms. I wonder if the FW800 chip really costs much. In terms of usb3, I don't expect them to add anything beyond what intel provides. They will be forced to support some third party chipsets to keep the thunderbolt display current. Beyond that I don't see it happening.
Apple is notorious for calling anything they don't want to include "legacy technology". They intentionally spin it that way to keep people from considering whether they might use the port at some point. If you absolutely need it, it's kind of an issue, especially as dongles carry a much higher failure rate and can be flaky. I went through some issues finding a good mini displayport to displayport cable a while back. In the case of dongles, some display manufacturers reported issues. I prefer a solid cable and one less connection in the chain. My hatred of dongles can be summarized by many of the Apple Store reviews on the various adapters. Some just don't work with certain things, and I did this research in advance as I hate returning things. If mini displayport had better overall adoption, this would be an easier task. The Apple thunderbolt cables don't work in many displays.
Next year should be a similar effort in terms of integrated graphics. I'm curious of their direction with the 13" mbp. If they're going for the high resolution display concept across that line too, a discrete gpu in place of the ODD could be a good move. In the case of the 13", it takes up a much larger percentage of the total space than it did with the 15". In terms of what will fit, some ultrabook manufacturers have pulled it off as Marvin mentioned. Ivy also has a single 35W quad cpu. It's more expensive than Apple's current 13" choices. Next year they max expand upon this, although I don't fully expect such a thing for at least two years. Apple doesn't like things that may decrease battery life due to higher peak power consumption (even if tdp is listed as the same value), so they may be conservative on that one. Overall the 13" could see a lot of cool options if they don't intentionally hold it back.
They are conservative on things that do not add coolness factor, such as usb2 to usb3. The machine will appear the same either way, and not everyone will know. If they really wanted it, they could have pushed it into an update at least by last year. As to external chipsets, they will update the thunderbolt display at some point with usb3, and that will also require a third party chipset. It's possible that they could push that further out to a point when thunderbolt sees a more definitive update and implement a display based on whatever panel comes out then. Most of the panels currently on the market in that size are from 2009-2010. I think the generation officially launched in 2009. Anyway resolution is a good way to go right now as contrast makes little sense until you can get to at least 10 bit displayport levels. Otherwise you're just spreading out the addressable colors further and further thus becoming more reliant on dithering to give the impression of visual continuity.
I get your desire for an Xmac. You want a decent modern stationary computer as the ipad can take over a lot of mobile needs, and a stationary one could offer more than a docked laptop if someone actually brought out such a thing. Apple with the ipad is in a decent spot to do this.
They aren't high in quality. They always have terrible reviews at the Apple Store. The chargers and batteries are the same, but a percentage of that comes from the fact that those are expensive replacement parts. People will naturally be irritated when they have to buy these things. In the case of dongles, they're spare parts. In my experience the seams are always uneven where they are put together. The connection is never snug. Many of them are not as reliable as a real cable, so sometimes you need a backup if they support a critical function. In the case of thunderbolt on my laptop, I ordered a specific mini displayport to displayport cable in case I wanted to view something from my laptop on the larger display (other computer still has dvi... ugh). A thunderbolt cable isn't necessarily backwards compatible with a displayport monitor (it has multiple outputs, but displayport is best), and dongles add an unnecessary and sometimes unstable connection to the chain. It's almost like they're only carried more as a sign of assurance in case a buyer is unsure about purchasing a mac with missing ports.
I can play the Marvin pricing game. Keep in mind a degree of rounding in the numbers will always exist. Most of the time they go in even hundreds rounding up or down to the nearest $99 mark. I don't know which side of that they're on with the discrete mini. The cto option on the dgpu model is the 2620m. If they can go up $30 (less wholesale unless I'm reading it wrong) they can use the 3612QM at 35W and maintain a discrete gpu. I'd be surprised if they can't cool that in a stationary form factor. Currently the server is their only quad option. This would give you quad + dgpu. The cpus are expensive as they're mobile parts, but Apple has used cpus in the upper minis that cost more than some of the mac pro cpus in the past. This is true with two current models. As to the desktop, Wizard's concepts are always interesting. I know desktops aren't doing well overall, but Apple always goes in wherever they think they can make a difference. Most television margins suck, yet we constantly read stories that Apple will build one. I'm a little surprised the current Apple TV hasn't morphed into an IOS device.They obviously have their reasons. It might cost too much for suitable hardware or be a support headache. Overall I don't think the success of those devices lies solely in hardware sales. They must drive itunes purchases too.
Server software should be a non-issue. I am not up to speed on Mountain Lion, but Lion included a number of server licenses along with the regular OS. When you take that out of the equation, the $999 price point can include vastly improved hardware compared to the ones that were factoring in the price of server software...OK, I realize that Apple actually has to pay something for additional hardware whereas the server software did not actually cost them anything and so any price they charged for it was essentially pure profit, notwithstanding amortized costs.
The problem with the present Mini is the form factor. It just does not have adequate cooling for decent hardware. People who want "small" don't necessarily want absurdly small. If there were to be a separate form factor for a server mini (Mini-Max?) it should be possible to use a 3 1/2 inch HD which will provide better performance (and capacity) at a more competitive cost than a 2 1/2 inch HD and space for a PCIe slot graphics card. Perhaps even a 65 watt CPU (and its cooling). Maybe even a few other goodies.
Found that geekbench entry yesterday, the Mac mini would be much better with desktop cpus.
The Core i7-3770T is a 2.50GHz 4C/8T 45W desktop cpu ($294 vs $378 for the cheapest mobile quad-core cpu from Intel) and it has HD 4000 graphics. So even just for the server model, it would be great. There are a few other T series chips from Intel, maybe Apple could use the 35W 2.90GHz 2C/4T Core i5-3470T ($184, it has HD 2500 graphics only), in the regular model along with a 20W nvidia GT 640M LE gpu. The savings on the cpu (the cheapest mobile dual-core cpu from Intel cost $225) could be used towards the dedicated gpu.
Apple doesn't need many models of Mac mini, just good ones:
$599/699 2.90GHz 2C/4T Core i5-3470T, nvidia GT 640M LE gpu, 4GB RAM, 500GB HDD
$899/999 2.50GHz 4C/8T Core i7-3770T, Intel HD 4000, 8GB RAM, dual 7200 rpm 750GB HDD
BTO: RAM, storage
FYI the mobile 2.90GHz 2C/4T Core i7-3520M costs $346. And mobile quad-core cpus from 2.1GHz to 2.6Ghz cost at least $378 (depending on the other specs).
I don't see Apple using more than a 20W gpu along the 35W cpu (the previous 6630M was a 15W gpu at full speed), so people should forget about the 32W 640M (let alone the 45W 650M) in the current design.
The "server" model is a bit awkward overall. I don't think it's necessary. It may benefit them from a marketing standpoint, but obviously I don't have their internal data to review. Mjteix makes a good point in that low power desktop variants are not a bad option, although I'm not sure what that would change. I'm also not feeling like looking up reference micro-ATX boards or researching peak power draw between the two at the moment. My example was to fit seamlessly with their current design.
The "server" model is a bit awkward overall. I don't think it's necessary. It may benefit them from a marketing standpoint, but obviously I don't have their internal data to review. Mjteix makes a good point in that low power desktop variants are not a bad option, although I'm not sure what that would change. I'm also not feeling like looking up reference micro-ATX boards or researching peak power draw between the two at the moment. My example was to fit seamlessly with their current design.
The server model resulted from a shock Apple received. Customers (corporate and EDU among others) were making use of the Mini in ways Apple had never envisioned and were reportedly buy substantial numbers of them to do so. The server model was simply a response to customer request.
The use of mobile components results in greater costs without any particular benefit. If a revised Mini form factor were developed to deal with the greater thermal envelope, the less expensive components could easily be accomodated.
I get your desire for an Xmac. You want a decent modern stationary computer as the ipad can take over a lot of mobile needs, and a stationary one could offer more than a docked laptop if someone actually brought out such a thing.
Now if only Apple would understand this.
Depends if you consider the PowerPC 7447A a desktop processor.
A year ago people were doubtful over whether Apple would even release usb3 drivers and suggested that thunderbolt was the only thing needed.
If you're referring to me, what I suggested was the Thunderbolt plus USB2 would suffice and that Apple might not support USB3 to further the strategic objective of wider Thunderbolt adoption. My prediction was wrong. On the other hand, my prediction that the 2012 MacBook Pro would have the RAM soldered directly on the motherboard was correct.
I'm hoping that Apple will drop the SO-DIMM slots on the Mac Mini and offer 8GB as the baseline soldered directly onto the motherboard with 16GB as a BTO option. I'm also hoping for an SSD slot. The rest of it seems relatively predictable: Ivy Bridge, USB3, 2nd Thunderbolt port, no FireWire.
If you're referring to me, what I suggested was the Thunderbolt plus USB2 would suffice and that Apple might not support USB3 to further the strategic objective of wider Thunderbolt adoption. My prediction was wrong. On the other hand, my prediction that the 2012 MacBook Pro would have the RAM soldered directly on the motherboard was correct.
I'm hoping that Apple will drop the SO-DIMM slots on the Mac Mini and offer 8GB as the baseline soldered directly onto the motherboard with 16GB as a BTO option. I'm also hoping for an SSD slot. The rest of it seems relatively predictable: Ivy Bridge, USB3, 2nd Thunderbolt port, no FireWire.
I wasn't referring to anyone specific. Lots of people shared your usb3 sentiment. Regarding soldered ram, what real benefit is granted if you aren't constrained by thickness? Assuming you're just soldering regular sodimms to the board as is the case with the rMBP, you would occupy more surface area on the board. Right now it mounts in a perpendicular manner to the board, yet surface area is more constrained than a notebook. They wouldn't be able to fit it in a similar manner unless I'm missing something. If you know more than me about board design, which is extremely probable (I'm eyeballing the layouts of the past couple models as well as the Air, rMBP, etc.). In both cases the soldering has come into play where the machine is constrained by height. The mini is constrained by logic board surface area, so even if they wanted to seal them in, I can't see Apple laying them down flat as they do with the rMBP. The Air doesn't use standard sodimms at all.
Regarding soldered ram, what real benefit is granted if you aren't constrained by thickness?
We've been over this before. The advantages of soldered RAM are: price, performance, reliability, and space. I grant that space is less of a constraint in the Mini compared to the MacBooks.
Assuming you're just soldering regular sodimms to the board as is the case with the rMBP, you would occupy more surface area on the board. Right now it mounts in a perpendicular manner to the board, yet surface area is more constrained than a notebook. They wouldn't be able to fit it in a similar manner unless I'm missing something.
The rMBP does not have SO-DIMMs soldered to the motherboard. Take a look at the motherboard of a Mini. The connector for the SO-DIMMs takes up nearly as much space as 16 DRAM chips soldered directly to the motherboard and even more space if the soldered DRAM chips were stacked two high. Integration of other chips would probably leave enough space on the motherboard to avoid stacking.
We've been over this before. The advantages of soldered RAM are: price, performance, reliability, and space. I grant that space is less of a constraint in the Mini compared to the MacBooks.
The rMBP does not have SO-DIMMs soldered to the motherboard. Take a look at the motherboard of a Mini. The connector for the SO-DIMMs takes up nearly as much space as 16 DRAM chips soldered directly to the motherboard and even more space if the soldered DRAM chips were stacked two high. Integration of other chips would probably leave enough space on the motherboard to avoid stacking.
I found a picture of the macbook pro 13" logic board. It doesn't have a lot of free space on the logic board. You'd pick up some dropping the optical drive in a non tapered body, but those dimms are still significant. They would have to be laid out side by side for Apple's soldering as you see in the rMBP.
Note how it's laid out.
I guess you could squeeze it in there if the width shrinks, but it's still silly. You mention reliability. It has one of the lowest failure rates of any part of the system. If you're having ram failure issues, stop buying cheap ram. Those components may well outlive the rest of the system. What signs have you seen on price and performance? Memory benchmarks from geekbench are lower by 100 points or so comparing the top ones on the retina macbook pro to the 2012 cMBP. It provided no distinct performance benefit in drag racing volatile storage. It won't save you space as you'll have to lie them side by side. It's more like they could probably cram them in if they want, and since when has the memory contact area been a common reason for a logic board replacement? Dimms rarely fail with quality memory. If they do with soldered, you can't replace them anyway. Looking at Air + rMBP, they haven't stacked soldered memory. It's laid flat across the board.
I'm reiterating the same few points here, but you haven't shown any real advantage other than Apple trying to sell ram upgrades :P. Also note that in both prior cases vertical restrictions could be noted as the reason for the use of soldered ram. It likely wasn't a design priority when determining placement and case height in those regions.
You mention reliability. It has one of the lowest failure rates of any part of the system. If you're having ram failure issues, stop buying cheap ram. Those components may well outlive the rest of the system. What signs have you seen on price and performance? Memory benchmarks from geekbench are lower by 100 points or so comparing the top ones on the retina macbook pro to the 2012 cMBP. It provided no distinct performance benefit in drag racing volatile storage. It won't save you space as you'll have to lie them side by side. It's more like they could probably cram them in if they want, and since when has the memory contact area been a common reason for a logic board replacement? Dimms rarely fail with quality memory. If they do with soldered, you can't replace them anyway.
Switching from SO-DIMMs to directly soldered DRAM means fewer components and, more important perhaps, fewer manufacturing steps. These improve both price and reliability. Future JEDEC specs will offer higher bandwidth and lower latency by requiring direct soldering of DRAM chips onto the same board as the CPU. The difference in space is negligible compared to the advancements in integration. Just dropping FireWire support probably frees up enough marginal space on the Mini motherboard.
Switching from SO-DIMMs to directly soldered DRAM means fewer components and, more important perhaps, fewer manufacturing steps. These improve both price and reliability. Future JEDEC specs will offer higher bandwidth and lower latency by requiring direct soldering of DRAM chips onto the same board as the CPU. The difference in space is negligible compared to the advancements in integration. Just dropping FireWire support probably frees up enough marginal space on the Mini motherboard.
They really need to add another thunderbolt port. The current one port thing is asinine given that one may be dedicated to displayport functions. I realize it doesn't support the bandwidth or functions of displayport 1.2 either way. I'd say that belongs in place of a firewire port if peripherals really start to roll out. It sounded like you were allocating a current performance advantage to soldered ram. Future specs do not mean much in terms of performance relative to a 2012 mini. It's silly that one hasn't surfaced for the current year. Do they plan to wait for bumped ivy bridge chips and roll it out at the end of the year for some bizarre reason? I find Apple's logic to be irritating at times.
If you're referring to me, what I suggested was the Thunderbolt plus USB2 would suffice and that Apple might not support USB3 to further the strategic objective of wider Thunderbolt adoption. My prediction was wrong. On the other hand, my prediction that the 2012 MacBook Pro would have the RAM soldered directly on the motherboard was correct.
I'm hoping that Apple will drop the SO-DIMM slots on the Mac Mini and offer 8GB as the baseline soldered directly onto the motherboard with 16GB as a BTO option. I'm also hoping for an SSD slot. The rest of it seems relatively predictable: Ivy Bridge, USB3, 2nd Thunderbolt port, no FireWire.
There is no sound reason to use USB 2 over USB 3 (it is backwards compatible).
Why on earth would you want RAM soldered to the motherboard?
My point was that his suggestions aren't reality today. He later explained that they would be advantageous when measured by future standards. In terms of space, it takes up at least slightly more space in a mini. In terms of performance the rMBP doesn't exceed the cMBP in memory performance. In reliability, that's debatable. I've never seen the connection there become a primary point of failure. If it's ram, stop buying cheap dimms. Most 3rd party ram carries a lifetime warranty anyway. If it goes bad with age, just take advantage of that. I'm relatively certain McCarling knows more about this stuff than me, but none of those suggestions reflect anything that would be implemented today.
