This is mostly likely a misinterpretation.
This is mostly likely a misinterpretation.
A man must have code -Bunk
IF this does happen feature rich mobos will be restricted to high end cpus and vice versa.
Mobo manufactures will probably have a smaller feature set range and offer a few cpu options for each set.
just a guess. This is all a rumour though SemiAccurate seemed pretty sure on it
SA seems pretty sure (and equally pessimistic) about everything lately.
As I said on another thread, funny how they only post after everyone already knows about it, claiming to have known it for months.
Everything leads back to that single Japanese article which, lets face it, isn't entirely intelligible through Google translate. Someone on Xbit IIRC posted a proper translation in the comments, and it's completely different to how most people are interpreting it.
The way I read that chart is that Broadwell is BGA/PGA. and that LGA is 'unknown' at this point but there is still a plan there.
PGA is pin grid array which is well what we had before LGA!
Does look like Broadwell is a system on a chip so bringing whats left of the chipset onto the cpu package ?
I would like to see maybe up to Core i3 BGA. you get it with Atom already and could give HTPC or home server builds more powerfull SOC solutions.
Intel is releasing its own SOC the NUC.
The only real issues I can see are:
Poor cooling leading to the same issues we get with the XBOX and PS3 with the solder.
I have had a few motherboards die in my time but not a single CPU. Unless warranty times increase dramatically you may have to bin a perfectly good CPU if the board dies.
It would be interesting if there was a move to CPU daughter boards but I thought we were trying to get to reduced distance between components and how many pins would you need in a PCI style slot to take that amount of data?
Intel gets a payday..
I've more than a few times swapped MB or CPU independently of each other - it's harder to do given intel's aggressive socket replacement program no doubt but it does happen. One of the biggest boons of desktops is that things are interchangeable..
That is what I mentioned in the other thread. Basically,expect to see Haswell probably extended for desktop use,ie,maybe a refresh in 2014,with Broadwell going to mobile and AIO computers on 14NM. These are most likely to have soldered CPUs. Socketed CPUs will be on 22NM with Haswell and remember IB-E has not even been launched yet. By keeping socketed desktop CPUs on 22NM,they will need to have less 14NM capacity initially and use this exclusively for mobile devices. The desktop market still takes up a large percentage of what Intel sells.
Remember,Intel had the original Core only as a mobile CPU,and kept the Pentium 4 for desktop.
"Though I have never actually upgraded a CPU on any PC I have made since 1990,"
Seriously? Are you made of money or plain stupid ?
I've been thinking the same thing really, Intel and AMD have gone from releasing top>bottom to bottom>top e.g. with Nehalem we had 1366 Bloomfield followed by Lynnfield roughly a year later. Then it was around the same time with Westmere; Gulftown and Clarkdale in Q1 2010. It flipped around with SNB, with a much larger gap between mainstream and enthusiast, even more so for IVB.
Intel have a load of idle 22nm capacity and are delaying the 14nm Leixlip plant.
So all these reports are based on one dodgy slide which also shows intel not having any CPU with more than 2 cores beyond haswell.
In short, its all bunk until we see something substantial.
That is until the whole board needs replacing due to a fault of a component. Of course there is a lot to be said for surface mounting components that increases reliability.
I cant wait to return and entire unit because one small thing has failed though and replacing faulty boards is a fun exercise Intel seems to like repeating, lets hope they can decrease the likelihood of recalls if they go this route.
There are currently 1 users browsing this thread. (0 members and 1 guests)