Two bits of info to upset your assumption. First, Asus has 37 for Sandy Bridge alone (LGA1155). They also currently have Socket 1366, Socket 1156, and Socket 775 for Intel, and AM3+, AM3, AM2+, AM2 and FM1 for AMD - as well as Atom and E-350/450 boards. Newegg has 128 Asus mobos in its current catalog.
I went by your figure at the time of that post. Going back and counting, I came up with 140 from ASUS's site (all boards, including those with CPU's soldered to the board).
I suspect your confusion is with the systems engineering aspects of board design.
Lets say you're working with a particular socket (socket count = 1), and have 3 chipsets available for that socket. It's possible to create a single PCB design that can handle all 3 combinations (and is intentional by the CPU vendor - think Reference Design for a CPU socket - where the different features are located in the chipset). Simple example, but this is the basis (i.e. chipsets for a socket share the same package = same pin-outs).
Now let's expand on this a bit.
Board vendor has the following design elements to chose from (some from CPU vendor Reference Designs, others they create on their own; BTW, values are arbitrary):
- CPU sockets = 12
- Chipsets = 30
- Voltage regulators = 8
- Peripheral chips (additional semiconductors to add features) = 15
By creating board designs via a mix-and-match methodology of these elements, they can produce a number of different boards over every socket they've chosen to sell, while sharing R&D for these various elements over multiple boards. Some can even be used in boards for competing CPU vendors, such as peripheral chips, and basic voltage regulator designs for other areas of the board, such as the PCIe slots which are standard regardless of CPU used.
The MP can't do much of this, as it's the only system of it's kind offered by Apple (i.e. share the same FW800 chip as say the iMac).
Hope this finally get the point across as how product lines and total volume matter vs. a single product in terms of reducing R&D cost per board.
But, the "even split" assumption is absurd. Asus has a huge OEM business - you'll find Asus motherboards in HP, Dell, Gateway and other branded systems. So millions of those mobos may be different models from the 128. (And low-end integrated graphics models probably outsell the higher end boards by a huge margin.)
I was just trying to keep it simple (KISS).
If the boards actual sales volume per P/N were equal, I'd be absolutely dumbfounded.
As per their profitability, they've taken that into consideration. For example, Yes, the cheaper boards may sell a higher volume. But they also generate less profit per board than the upper scale variants (when available). So they've set MSRP's that balance it out = each is profitable. They also have control of the production volumes (since they actually manufacture what they sell), so they're less likely to be stuck with excess volume, unlike system vendors that use OEM/ODM methodologies to get their products.
As per boards made for HP, Dell,..., I'm not sure if the 21.6M figure includes those or not, as they're actually sold through their Pegatron arm (technically, a separate company that's owned wholly by ASUS - it's their OEM/ODM division, just as Foxconn is the retail division/label of Hon Hai Precision).
However, we're talking about technologies (X5690, USB 3.0, SATA 6Gbps, TBolt, BD, ...) that have been around since last winter.
Yes, they've been available. But it wouldn't have been in Apple's best interest on multiple levels.
- They would likely be left with too much of the previous systems when the new ones arrived.
- Purchase contract volumes and delivery dates for CPU's were set before the newer CPU's were available (they don't buy as many electronic components as other vendors for their workstations - not just CPU's = affects purchase prices).
- New CPU's you're interested in are too expensive to keep the MSRP in line with Apple's idea of margins.
- R&D would increase per board due to reduced sales volume (has to do with time frames they can sell the new systems before the SB-E5's begin shipping to system vendors).
It all adds up to a bad business decision for Apple due to marketshare and timeframe of sales.
