But we're comparing to products for people that might actually need those 8 DIMMs.
No. In what I quoted, the comment was that Apple's 2013 version of a 12 core box was going to be more expensive than other vendors dual cpu offerings. That dual sockets was better on price than a single socket limitation.
This is kind of an odd possition. Your're saying this will be cheaper than products with 8 DIMMs because it doesn't have 8 DIMMs. But IT DOESN'T HAVE 8 DIMMS, so why would someone shopping for a computer that needs 8 DIMMs look at this Mac Pro.
It is cost, not DIMMs slots that is at issue in what I was commenting on. Saying that 8 DIMMs are necessary to be in the equation is just circular with respect to system costs.
Or conversly, why would a shopper looking for a computer that only needs 4 DIMMs look at computers with 8?
errr, because looking for a cheaper system? Most folks have a budget. Econ 101.... lower price (or better $/performance-value ) , higher demand.
The top end 12 core will likely get you 2x of the 2850's, which will be 8 cores each for 16 total cores at 2.6 GHz, compared to 12 cores at 2.7.
Err. no. The current 2650's are 2 GHz not 2.6. ( presuming typo above 8 for a 6). The leaked (in another thread ) 2650 v2 is 2.6. However, the current 2650 is $1,107. 2x that is $2,214 which is more than anything on the current E5 list. The 12 core will be higher than were those prices top out at : $2057 , but Intel still has about $200 to play with before doubling. +/- $80 ( on a $6,000) system the 12 core price is probably pretty close double 2650. To get under you'd have to 2640 and lower range where going to start to trade off clock as I said before.
More DIMMs slots, more x16 PCI-e slots (and usually correlated larger power supply ) typically leads to bigger infrastructure costs was also eat into any +/- $80 difference.
Not sure what exactly your point here is, but at the same CPU cost you can generally get better performance with DP systems, except at the very low end.
And that is exactly why Intel's pricing scheme for the E5 is not exactly linear.
But they also have single processor systems they will let you stick the 2687W in, if you like. So...huh?
So this won't work at all for Apple, but it does work for the current system vendors? That is the "huh" ?
This isn't a new scheme that gets partial coverage that Apple is trotting out here. Almost every one of those dual system vendors will sell a configuration of their dual processor system hobbled with only one E5 2600 in it? Why because some folks budgets don't go high enough to fully socket the machine and then
can get productive work done on the machine now.
Is lack of the empty processor f going to loose the subset of "just buying for future proof sake" folks who buy those kinds of sytsems. Yep. Was Apple looking for customers who are primarily focused on capacity they are not going to use in the immediate-medium term future ? No.
Indeed, but for the forseeable future mid range DP systems are going to be a better value than high end SP systems.
???? You are waaaaaay out in specialized workloads now. For folks who have a mix of mild to ultra parallel apps higher clocks are going be more generally valuable. Intel's nosebleed pricing to get high core count and high clock speed shrinks that market, not makes it bigger.
If partially talking about filling multi ranked DIMMs slots with cheap last generation RAM .... well that is going to hit a brick wall with E5 v3 ( Haswell) and DDR4. Apple's designs focused on single ranks is 100% aligned with the future. The focus on just 4 DIMMs slots has been 100% aligned with the present for over 4-5 years.
Furthermore, there are more than a few folks switching to high as possible x86 cores for scalar constrained work and throwing embarrassingly parallel work at GPGPUs. Those folks are going to fit in just fine on the new Mac Pro's general system architecture.
If by going embedded GPU Apple can get their dual GPU set-ups prices more competitively price that "roll your own" workstation set-ups ( e.g., Maximus ) then they do have an advantage here they can use to sell a sizable number of these systems.
Are they going to sell one to every one of the old Mac Pro buyers? No. That isn't material. What matters is even folks just buy them. If they can replace 1.1 new buyers for every 1 that stomps off in a huff then this new machine will be a success. That is better growth than the old Mac Pro was doing.
