What is Intel\'s future socket type

[jlamonsnwa]

What CPU socket holds the most chance of being used on future CPU releases. I always seem to pick a Mobo socket which is destined to be redesigned in the next 2 years making my MOBO an instant relic. You can include both Intel and AMD...

 

[amuffin]

socket 1155 is ready for next years ivybridge cpu's.

 

[jaguarskx]

Ivy Bridge will be the final socket 1155 CPU middle of next year.

I assume that Ivy Bridge-E will be coming out for socket 2011 towards the end of next year. Not sure since I have not done the research.

Intel's next CPU architecture, Haswell, will be coming out mid 2013 and it will use socket 1150. It will replace Ivy Bridge as the new consumer oriented CPU.

 

[blazorthon]

AM3+ socket motherboards support AMD's recent Bulldozer processors and should support AMD's upcoming Piledriver processors that come out next year or so and will might support AMD's next processor architecture after Piledriver, I think it's called excavator but I don't remember.

 

[jaguarskx]

Piledriver will be AMD's final CPU for socket AM3+. They will be dropping CPUs entirely and devote resources towards the APU (their version of a CPU and GPU core in one chip) like Llano and the upcoming Trinity.

 

[CaedenV]

these days sockets tend to only last for a few years. It's not your fault, it is just the way things are.

I agree with Jaguarskx with the exception of IB-E chips. Normally Intel will come out with a new socket for each high end line. Supposedly this is because they are 'so different' from one another, but really it is to sell more chip-sets, and to make high end builders feel good about having 'the best' available to them. There may be some truth to it, but something tells me that it is mostly marketing.

 

[amuffin]

Piledriver will be the last cpu that amd will make. Best option would be switching to intel imo.

 

[blazorthon]

No chip is so different that it can't use a previous chipset if a chip maker wanted it too. The differences between Itanium and x86 Xeons is huge yet current Itaniums and Xeons use the same chipsets.

That the socket would need to change, that is reasonable. But there is no way you can't take a chipset and use the exact same chip(s) with a new socket if you wanted too.

Neither of the above mean that there is no good reason (for the consumer) to have different chipsets, far from it. Having new chipsets means that you can have new features or other changes that the previous chipset lacked or move some parts of the chipset to the CPU (such as the northbridge).

 

[blazorthon]

these days sockets tend to only last for a few years. It's not your fault, it is just the way things are.

I agree with Jaguarskx with the exception of IB-E chips. Normally Intel will come out with a new socket for each high end line. Supposedly this is because they are 'so different' from one another, but really it is to sell more chip-sets, and to make high end builders feel good about having 'the best' available to them. There may be some truth to it, but something tells me that it is mostly marketing.

IB-E will probably use the same motherboards and chipsets as SB-E just like IB uses (some of) the same chipsets and motherboards of SB. That is unless Intel does more changes to SB-E than they did to SB, in which case you may be right but I bet Intel will keep compatibility between SB-E and IB-E.

 

[IntelEnthusiast]

The biggest reason for the change is sockets is the change in technology to go with it. 775 was a great socket that lasted a long time. When we released the 1st generation Intel® Core™ processors not only were some of these the first processors with a IGP (Integrated Graphics on Processor) but we also moved the memory controller onto the processor. Between these two changes it is understandable that we would have to change the socket to deal with these changes. Two years later with the release of a new microarchitecture the sockets would change again. So while I havent been given a crystial ball to allow me to see the future but it looks like the change in sockets seems to be based off the advancement in technology.

Christian Wood
Intel Enthusiast Team

 

[blazorthon]

The biggest reason for the change is sockets is the change in technology to go with it. 775 was a great socket that lasted a long time. When we released the 1st generation Intel® Core™ processors not only were some of these the first processors with a IGP (Integrated Graphics on Processor) but we also moved the memory controller onto the processor. Between these two changes it is understandable that we would have to change the socket to deal with these changes. Two years later with the release of a new microarchitecture the sockets would change again. So while I havent been given a crystial ball to allow me to see the future but it looks like the change in sockets seems to be based off the advancement in technology.

Christian Wood
Intel Enthusiast Team

I agree. Socket and chipset changes reflect changes in the technology that are beyond architectural changes and along the lines of integrating new features into the CPU and/or moving features from the chipset to the CPU.

 

[87ninefiveone]

No chip is so different that it can't use a previous chipset if a chip maker wanted it too. The differences between Itanium and x86 Xeons is huge yet current Itaniums and Xeons ise the same chipsets.

That the socket would need to change, that is reasonable. But there is no way you can't take a chipset and use the exact same chip(s) with a new socket if you wanted too.

Neither of the above mean that there is no good reason (for the consumer) to have different chipsets, far from it. Having new chipsets means that you can have new features or other changes that the previous chipset lacked or move some parts of the chipset to the CPU (such as the northbridge).

This is just blatantly wrong and rates right up there with people who think that just because two products are made in the same factory that they'll be equivalent to each other.

The reason for Itaniums and Xeons being the same is obvious. Business who use those products buy 100's or 1000's at a time and need to them to be forward and backwared compatible to avoid huge capitol outlays. The consumer market however is driven by individuals and allows for more flexibility since 99% of people could care less about backwards and forwards compatible socket design. They need a new PC they buy one.

As the Intel rep explained, adding new technologies require the use of different sockets. If they need 1155 pins to do the job, there’s probably just not much of a chance that 33% of those pins are just sitting there doing nothing aside from making you buy a new socket type over your old LGA 775 setup. Granted 1155/1156, and 1366/2011 are pretty annoying there’s an obvious reason for it with the 2/4 core support vs. 4+ core support.

 

[blazorthon]

This is just blatantly wrong and rates right up there with people who think that just because two products are made in the same factory that they'll be equivalent to each other.

The reason for Itaniums and Xeons being the same is obvious. Business who use those products buy 100's or 1000's at a time and need to them to be forward and backwared compatible to avoid huge capitol outlays. The consumer market however is driven by individuals and allows for more flexibility since 99% of people could care less about backwards and forwards compatible socket design. They need a new PC they buy one.

As the Intel rep explained, adding new technologies require the use of different sockets. If they need 1155 pins to do the job, there’s probably just not much of a chance that 33% of those pins are just sitting there doing nothing aside from making you buy a new socket type over your old LGA 775 setup. Granted 1155/1156, and 1366/2011 are pretty annoying there’s an obvious reason for it with the 2/4 core support vs. 4+ core support.

The reason for current Itaniums and Xeons being the same is obvious? They aren't the same they just both support the same chipset. Itanium and Xeon use extremely different architectures that aren't even as similar as 32 bit x86 and 64 bit x86. Itaniums use the EPIC architecture that is closer to VLIW than x86 and it is still pretty far from VLIW.

Not only are you wrong about the first parts of my post but you completely ignore the last part which pretty much says that I agreed with the Intel rep and the post right above your own post where I say I agree with the Intel rep. Yes there is good reason for using new sockets but it isn't because we need to use a new socket, but because we want to.

If Intel really wanted to they could have made Nehalem CPUs use the LGA 775 socket but that would have meant that the on-die memory controller and the IGP (for those CPUS that had one) would have been inaccessible and that they would have needed to put the circuitry to make Nehalem compatible with the socket on the CPU, essentially making it an architectural upgrade to the Core 2 series instead of a radical change in the CPU/chipset features and what chips those features are on. That would have made it like the difference between current Itanium and Xeon products or between AMD's AM3 and AM3+ CPUs. They have different architectures but can both use the same sockets.

My whole point was that we don't need to move to new chipsets/sockets but that we do so because it allows us to advance the technology.

 

[blazorthon]

This is just blatantly wrong and rates right up there with people who think that just because two products are made in the same factory that they'll be equivalent to each other.

The reason for Itaniums and Xeons being the same is obvious. Business who use those products buy 100's or 1000's at a time and need to them to be forward and backwared compatible to avoid huge capitol outlays. The consumer market however is driven by individuals and allows for more flexibility since 99% of people could care less about backwards and forwards compatible socket design. They need a new PC they buy one.

As the Intel rep explained, adding new technologies require the use of different sockets. If they need 1155 pins to do the job, there’s probably just not much of a chance that 33% of those pins are just sitting there doing nothing aside from making you buy a new socket type over your old LGA 775 setup. Granted 1155/1156, and 1366/2011 are pretty annoying there’s an obvious reason for it with the 2/4 core support vs. 4+ core support.

Ok I found my mistake in my post. I forgot to add "and sockets" to most of the places where I mentioned chipsets. I now see how my post can be misinterpreted as me saying there is no reason for new sockets, which yes is blatantly wrong. That is not what I meant and for that I hope this clears up any confusion.

Regardless of this mistake I did fix it in the post directly above yours so I'll assume you started your post before I submitted it.

EDIT: I also think it should be noted that the sockets don't have much of anything to do with maximum core count supported on a platform. Maximizing the core counts of CPUs made for a certain socket are for other reasons.