Nope. That' not it. They just don't want to give you the SKU because otherwise you could easily figure out how much you are overpaying.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Thunderbolt iMac Torn Down and Benchmarked
- Thread starter MacRumors
- Start date
- Sort by reaction score
Nope. That' not it. They just don't want to give you the SKU because otherwise you could easily figure out how much you are overpaying.
Because so much of the cost of a computer is the cost of the CPU? Of course not. Everyone can see there's more to an iMac than just the CPU.
Scrolling through some of your previous posts, I have to wonder why you bother with this site, you're never going to see something you like.
While the Mac is more the sum of its parts, lilo is perfectly right with his assertion. Apple doesn´t want transparency that detracts from whatever they preach. Nevertheless top product though somewhat underspecced for 2k.
180 markup for i7 upgrade? Plzzzz
Well, it's not jut CPU. Apple will not tell you SKU for any of the components. All iMac components are regular PC components so if a buyer knew the part list s/he would make an educated decision and iMac would lose every time.
I don't think it would be a surprise to anyone to hear there's a generous mark-up on Apple products. It's for the consumer to decide if that's worth paying for or not. There are market forces after all, if it was too expensive, not enough people would buy them - but that doesn't seem to be a problem they're having at the moment.
Still, they are more than the sum of their parts, other companies are free to pick those parts off the shelf and put together something the same or better than the iMac for less. None of the competitors have impressed me with their attempts so far though. I'd be interested to see the cost of the 27" IPS display as well, doesn't strike me as a bargain bucket component.
It's not a bargain. Dell UltraSharp U2711 is a 27" monitor with exactly the same panel as iMac. Costs - $950. And given that monitor tech is not progressing anywhere near as fast as CPUs and other components (so it makes sense to keep these nice monitors for 5...8 years), to me, buying all-in-one computer makes no sense.
Because you were becoming irrationally exuberant - I thought that you needed some balance
.I also wonder about this Intel diagram:
If this applies to the Imac TBolt ports, then the two TBolt ports in fact share bandwidth from the single PCIe x4 connection. The combined max throughput of the two ports would be no better than that of a single TBolt port.
Another moot point, though, since no TBolt devices can be purchased.
Last edited:
On their webpage, apple specifically says that the two ports are on separate busses and have double the throughput of a single bus. I guess whether you believe them or not is up to you...
Sorry, but could you provide a link please? I looked through the Imac/TBolt pages, and didn't notice that statement.
I saw that...
I saw that - but TBolt has always had two 10-Gbps channels per port.
I'm looking for a statement that says that each port is independent - so that you get 4 channels (two ports with two channels each) with an aggregate throughput of 40 Gbps (each direction).
That's not possible with a PCIe x4 link to the controller as is shown in the Intel diagram. Since the teardown shows only one TBolt controller - either the Intel diagram is incomplete (it doesn't show the possibility of a PCIe x8 connection), or the two Imac TBolt ports share the bandwidth of a PCIe x4 connection and cannot simultaneously run all channels at the rated bandwidth.
This entire Thunderbolt hypefest gives the term "vaporware" a whole new definition.
I saw that - but TBolt has always had two 10-Gbps channels per port.
I'm looking for a statement that says that each port is independent - so that you get 4 channels (two ports with two channels each) with an aggregate throughput of 40 Gbps (each direction).
That's not possible with a PCIe x4 link to the controller as is shown in the Intel diagram. Since the teardown shows only one TBolt controller - either the Intel diagram is incomplete (it doesn't show the possibility of a PCIe x8 connection), or the two Imac TBolt ports share the bandwidth of a PCIe x4 connection and cannot simultaneously run all channels at the rated bandwidth.
This entire Thunderbolt hypefest gives the term "vaporware" a whole new definition.
but if it's 'per port' then each port has two 10-Gbps channels I can't see it getting clearer than that.
If the ports weren't independent, it would have to say 'Two 10-Gbps data channels between them' or 'Two 10-Gbps data channels but under full load they're 5-Gbps'
Besides, you are aware that the iMac model in the teardown was the 21.5" with one TB port? There's every possibility that the 27" with two TB ports has two TB controllers. In fact I would say it's more of a likelihood than a possibility
Apple are masters of spec ambiguity
Apple's spec sheets have always been masterful at implying more than is actually available. The sheets will say "five USB 2.0 ports at 480 Mbps" without mentioning that all five are on one USB hub and the 480 Mbps is shared among the ports.
Good point about the 21.5" teardown, but in fact I would say that I anticipate the 27" teardown photos.
Apple's spec sheets have always been masterful at implying more than is actually available. The sheets will say "five USB 2.0 ports at 480 Mbps" without mentioning that all five are on one USB hub and the 480 Mbps is shared among the ports.
Good point about the 21.5" teardown, but in fact I would say that I anticipate the 27" teardown photos.
Good point? I would say it was more like a fundamental point. Anything said before actually seeing the insides of a 27" model is pointless speculation, including me speculating there are 2 controllers/new controller design.
OWC did a lower quality teardown of the top of the range 27" model http://blog.macsales.com/9995-2011-imac-unboxing-teardown
Looking at the 21.5" from iFixit, the TB controller is marked with the black rectangle here: Image 1
and this: Image 2 gives you an idea of the size of the chip.
Compare this with OWC:
Image 1
Image 2
Image 3
There is what looks like a white passive heatsink covering a chip(s) that isn't there in the 21.5". That heatsink is in the place you would expect to find the TB controller, and it's definitely wider than the TB controller shown in the iFixit photos. Eyeballing it it looks like it might be wide enough for two of those little TB controller chips side-by-side... Yes, the two MLB's aren't laid out identically but all the connectors appear to be in a similar position and I can say that in most cases the different models have similarly laid out MLBs.
Again, more speculation but it looks like there's something there. Perhaps the potential for 40Gbps of throughput in that one area of the MLB is enough to warrant a passive heatsink. Guess we'll see.
Wait, seriously?! Not that I'm doubting your credibility ('cause I'm not), but mind linking me to your source on that?
For what it's worth (I didn't know this either):
The graphics card of the 2011 MacBook Pro is connected on PCI-Express 8x
Unlike in the 2010 models, the graphics card in the new 15" and 17" MacBook Pro is not connected through a PCI-Express 2.0 16x, but only an 8x.
From a practical point of view, it doesn't change anything and there is no decrease in the performances, since the 4 GB/s bandwidth is more than enough to get the best out of the graphics card. Instead of getting angry at Apple, we tried to understand the reasons of that choice.
To begin with, the Sandy Bridge processor has 16 PCI-Express 16x lines available and 8 others that are used to connect the Cougar Point controller.
There are also 4 PCI-Express 2.0 links on the CPU and the controller (DMI link), used to connect the two components with a 2 GB/s bandwidth, which is plenty enough to connect the CPU to the hard drive(s) and all the non graphics components.
Even if all the PCI-Express links offer the same data transfer rate, the ones that are integrated to the processor offer the advantage of never being limited in their transfer of data to the processor.
Here is part of the internal diagram of the 2011 MacBook Pro.
We can see on this diagram that Apple decided to limit the connection to the graphics chip in order to get 4 PCI-Express links for the Thunderbolt. This decision was probably made to maximize the performances of the connector and also prevent from making the DMI link a bottleneck.
That decision will have a limited impact on the performances of the THunderbolt concerning file transfer but it could become important if the data transferred has to be treated directly by the CPU.
In the end, at first it looks like Apple limited the graphics card, but that is not the case, and instead we get a more coherent architecture, which will allow the Thunderbolt to show its best.
TB is PCIe (among other things).
Et tu, Aiden
. We agree on many points, but this one… j/k
Last edited by a moderator:
To clarify, I'll wait for the "iFixit teardown" of the 27" model to make a comment. Guessing about what's underneath fuzzy shots of heat sinks isn't useful.
In any case, though, it's more to know what you're getting. Oversubscribed interfaces can be very useful - most of our connections are oversubscribed.
I'd just like to know up front whether and how the bandwidth is shared between the two TBolt ports. Apple can make it very hard to understand those details.
Nothing new, it's in my 4 year old 24" iMac.
Don't know why they made a deal of it at FixIt.
I still wonder if they could have made it thinner... Seems that Apple does not really want to change the iMac and Macbook Pro physical dimensions.
