Originally Posted by jomama22
No offense, but the way silicon based chips are produced, you have to sell cut down chips to be profitable.
If every chips had its own wafer and own specific production line, no one would be in business.
I understand the thought process, but semiconductor manufacturing doesn't follow the process of other manufactured goods. Do you expect Intel to throw away the xenons that can't run 8 cores at their specific bin but can run 6 core effectively? Of course not. Why? Because there are us enthusiasts who want to gobble up those chips.
Here's a good analogy: if you are a farmer of corn and you plant an an entire acre, let's say only 75% is actually edible and fully ripened, would you throw the left over 25% that didn't make the cut away? Hell no, you grind that up and sell it as animal feed or corn meal or to Frito lay for their cheetos.
Sane goes for a cook, if I buy a fish for $10 a lb, but only get 75% yield in flesh out of it, should I just throw the bones away? Hell no, you make that into a fume (fish stock) and make a volute (cream based fish sauce made with fume) and make money off of it.
This goes beyond binning for the sake of harvesting mostly-functional die.
Do you think $1,000+ is a fair price to pay for a hexa-core? Not exactly much of a discount for lacking 2 functional cores.
With the way Intel sells their Xeons for way over one-grand, corporate customers can effectively subsidize SB-E/IB-E and Intel would still likely be making a profit. Also, Intel is the undisputed king when it comes to maximizing wafer yields. They don't have too many partially defective die to begin with.
The point is, given the age of SB-E, Intel should be selling these chips for less (like Ivy Bridge currently), especially the top-end 1,000 SB-E model.
Oh wait, this chip does not have any other real alternatives (or IB-E successor for that matter) out in the market apart from other Xeons. Intel can charge whatever it likes because it doesn't have to answer to any other corporation/competition but themselves. Never mind.
Besides, my argument is about 'more features and performance for you money'.
Intel likes to force consumers into a choose '1 feature set or the other, but not both' scenario ('business' features like TSX for non-K CPUs, overclocking for K CPUs but not both together). Or weird artificial constraints just for the sake of product differentiation (why don't all desktop CPUs have unlocked multipliers).
Better yet, why did Intel switch from solder to TIM in the first place? Only Intel knows the real answer to that question.Edited by elreyhorus - 6/23/13 at 5:00am