Yeah I saw this yesterday, but it is not like I lost faith. I always knew it was coming even though most enthusiasts didn't believe it existed on the grounds of not using the socket for an "X" consumer chipset. In otherwords, typical enthusiasts tend to think it is all about their POV.
About the only concept that stands out to me is the Pentium 1400 series branding of a 1.33GHz dual-core Sandy Bridge with triple-channel; why not call it a Xeon like the rest? Simple, Intel has done their homework. I may never know, all I can figure is Intel is attempting to net an audience apart from typical Xeon customers.
But a pair of pentiums in a dual-socket board strikes me as odd, except if the Pentium 1400's might have been originally called "Xeon E5-1400", signifying the single-socket version of Xeon E5-2400 (akin to E5-1600 for X79 while E5-2600 for dual 2011 systems), and rebranded.
Here's Intel's logic
for which brand of CPU can be used or which server application (because server is a use and not a function limited to certain hardware).
It remains my strong belief that had Intel not overestimated Bulldozer, X79 would have been LGA1356 and not LGA2011. They reacted to Bulldozer's abysmal performance by postponing the cheap 3820 and focused all X79 sales to expensive 6-cores to make up for the expensive platform and pushed a more costly mainstream unlocked 2700K. Historically, Intel always reacted to AMD by dropping prices, not increasing. So LGA2011 for X79 was a mistake, it was overkill for them where they loose premium 8-core models with 2-cores disabled by selling them as i7's that aren't popular because Intel pushed mainstream first with Sandy Bridge. It is a mess.
LGA1356 sounds like a cross between LGA2011 and LGA 1366
Close. Intel used to have two separate sockets for dual-socket boards, Xeon 5500/5600 in LGA1366 (triple-channel) and Xeon 6500 for LGA1567 (quad-channel)-- except the latter was orginally made for Intel's 4-way systems. I think Intel attempted some consolidation and segmentation where the traditional Xeon split by letter (W, X, E, L) was separted by socket, higher end of dual-socket went to LGA2011 while mainstream went to LGA1366, while both sockets have access to 8-core models (killing the myth that more cores need more pinouts), 1356 should have cheaper models compared to 2011.
.... this is what happens when competition is severly limited,
...change sockets every few months.... 1366, 1155, 1156, 2011, 1356
No, you're just thinking one-dimensionally.
Intel always had multiple market segments that, per their massive share of the market, they are obligated to cater to each one; best with a unique platform and socket. In otherwords, they don't replace or succeed each other.
The past few sockets line of succession was as follows:
- 1-way mainstream: mPGA478 > LGA775 > LGA1156 > LGA1155
- 1/2-way server: mPGA478 > LGA771 > LGA1366 > LGA1356
- 2/4-way server: mPGA604 > LGA1567 > LGA2011
- mobile: mPGA478/479 > mPGA988a > mPGA988b
- Embedded can use any socket combination per order, i.e. Xeon C3500/C5500 series are 45nm LGA1366 soddered permanently to their boards.
Intel has four major segments they cater to: Mainstream Client, Server, Mobile and Embedded; each with their own sockets. Yes there was a time when all Intel sockets were the same, that changed to better cater to each segment. AMD and all other non-x86 companies have done the same with each market having their own socket, this isn't new.
It is just that many so-called "computer hardware enthusiasts" are ironically clueless about everything but what they get for themselves. Not that anyone is obligated, but it makes them look like frauds when they don't know what else is going on.
Interesting server socket chip but why only a dual core? Can you give us some insight as to why these would be better than our current gen Xeon?
FWIW, there is only a handful of dual-core models, and it has been that way for a while. A few Xeon E550x
and E560x were dual-cores for dual 1366 systems, they have slow specs and are cheap. Oddly enough, there is one LGA2011 Xeon E5-2637
fast dual-core model deliberately placed in higher TDP to justify costs for what it is: Intel monopolizing a quad-channel 32nm dual-core processor. Triple-channel version will be cheaper, maybe.
If you had a business where your applications benefit from high memory and/or PCIe bandwidth, but not the x86 processing power or associated power/cooling costs, then dual-cores are perfect. Intel can't just ignore a market thinking customers should move on to more cores even though they don't need the extra power and cooling to go with it; plus why allow AMD to fill that gap? Especially in the server market, customers here aren't as guilible or pacified as with desktop/laptop, so Intel doesn't mess around and makes sure just about every combination is available. This is the reason Intel maintains their 95% share of the server market because they have covered all bases. Businesses don't think like enthusiasts
-- for one, they can spend between 3-7 years between overhauling their systems, sockets will change no one is getting suckered-- they are only interested in maximizing their profit
and they might not need the best to do that or upgrade often. Depends on their applications and who is competing with them.
<message edited by lehpron on Wednesday, May 02, 2012 9:26 AM>