LGA775 is having a good run. While AMD has introduced three desktop sockets since 2003, Intel has been largely sticking with LGA775 on the desktop for the past couple of years, and leaked roadmaps suggest this strategy might continue until well into next year. However, VR-Zone reports that Intel will finally ditch the socket come 2008. Together with the company’s upcoming “Nehalem” architecture, which will supplant Core, the site says Intel will introduce two new sockets dubbed Socket B and Socket H.
Both LGA775-style land grid arrays, the sockets will feature 1,366 and 715 pins, respectively. Socket B’s increase in pins will leave room for Intel’s Common System Interface—the company’s answer to HyperTransport—and an integrated memory controller, which will supposedly both be featured in Nehalem chips. As for Socket H, VR-Zone suggests it will complement Nehalem-based processors without an integrated memory controller.
From chip-sets deliberately lacking any forward-compatibility (anybody remember the 915 and 925, incompatible with dual-core and effectively obsoleted <6 months after introduction ?), to incompatible voltage regulators, to incompatible pin-outs on the same socket-matrix, to fully-incompatible sockets, Intel sure knows how to keep the motherboard ( er, I mean chip-set) business running red-hot. Need to get that $half-billion or so that they are investing in upgrading their chip-set production capacity back as soon as possible.
You didn’t have an extractor?? Pfft you fail at nerd.
AMD should be crapping on their pants from here on… 🙂
I miss the days of using an expansion card cover to pry out CPUs… 🙁 Ahhh, those were the days. Pushing in a CPU thinking it was resisting because of the barb on the pins when in reality it was an actual pin neetly folding over under the cpu. good times…
q[< It's not really shooting in the dark if Intel releases the VRM specs ahead of time. Given that the 45nm CPUs have already taped out Intel must know the specs for these CPUs.<]q There's a considerable lead time between when a motherboard is designed and when its available. When the current gen of board was designed, 45nm MPUs were unavailable, testing couldn't have taken place for that simple reason, and final specs need real silicon to verify what models might tell you. q[
It’s not really shooting in the dark if Intel releases the VRM specs ahead of time. Given that the 45nm CPUs have already taped out Intel must know the specs for these CPUs. I am doubtful more because it’s not like the mobo makers to design for specs ahead of time, but if there isn’t a new VRM standard for them then it doesn’t matter and even if there is there are some historical cases of older mobos unofficially working with new CPUs.
Anyway, personal guessing isn’t what I was looking for I can do that well enough myself 🙂 I was looking more for confirmation or denial from semi-reputable sources.
And relax with the Intel fanboy-ism, your post#11 is just a troll in response to what was a perfectly fair, balanced, and accurate statement by moshpit.
q[
q[
I think he was referring to the current LGA method of insertion with ‘slap the retension clip over it’ and mistakenly called it ZIF although he probably meant it in terms of the acronym and not the socket description. I suppose one could say LGA sockets are technically still ‘zero insertion force’ since it’s not like the really old mounting methods where you actually did have to exert force to get the CPU to mount.
The new sockets = LGA, because ZIF becomes problemic with 1,000+ pins.
Socket F2 is also LGA, becuase it has 1227 pins.
I never understood the negatively with LGA. It is just a different and more pracitcal solution when you are dealing with tons of tiny pins. Do you have any idea how easy it is to bend a pin on modern CPUs?
I suspect it was also partly a business decision because an IMC makes the chipset much less of a factor. Intel sells lots of chipsets and having a chipset-based memory controller seems to give them an advantage versus their chipset competitors. It will be interesting to see whether Intel continues to license the bus for these CPUs to other companies.
On a side note, I haven’t been able to find any rumors let alone hard info about something that I want to know. Will current C2D LGA775 boards support 45nm CPUs? Usualy a new process involves a new VRM standard but since the current boards support a very wide voltage range down to 0.825-ish it’s not clear.
I think that some people’s sarcasm meter needs readjustment. 😉
as long the new socket remains zero insertion force i think it will be not a big deal
pop it in and slap the retension clip over it and bingo takes like 3 seconds
espcially since its at least a year away
I think Krogoth was talking about the socket change, not CSI and IMC.
I did. Intel has been as happy to copy AMD tech as AMD used to be doing so with Intel tech. CSI and IMC were a eventual given and both together certainly would require a new socket. I applaud Intel for getting with the program. Too bad we have to wait for 08 to see this stuff from Intel. It gives AMD some advantage for awhile longer in servers at least if not the desktop anymore.
Finally….
Wow, I didn’t see this coming……………….