Maximum PC

I was thinking about this the other day as well.

It will be interesting to see if Intel sticks with their current 8 w/2 disabled, or you'll be able to buy an 8 core chip from them for the 'desktop' side.

n0b0dykn0ws

Because 8 cores for most of the world is unnecessary, let alone 4 cores. Even in games, where you can count the number of games that effectively scale with cores with one hand.

I don't think IVB-E is for us. It would drive desktop users nuts just like the s2011 adopters are doing currently. I've heard it in my daily talk even; "I want the s2011" in which I have to retort, "yeah, if you want 6-cores with 6 more fake cores w/ no upgrade path...that's what you should waste your money on".

4x4 is still the wall us users need to deal with emotionally, let alone 4x1 of the "dated" i5 2500K. Yes, it's very interesting to see 16 native cores on AMD and would love to see 8x8 from Intel...but that's server stuff and not us. Bigger numbers or cost does not mean better and I just feel we should all find equilibrium with our parts finally.

Ivybridge is not a huge step most envisioned. It's just a SB shrink after all. Those that adopted it know already about the silver being gone in the IHS and makes them worry (and ignored years of folks saying "temps are relative"). Now we want to start the IB-E hype? Haswell, Broadwell...I think we should focus more toward the tech and not the MAC crap that "the next is what we must have".

BTW, the big Mac announcement was they now have Ivy Bridge like we've had. USB 3.0 is new to them. Let's not stumble down that lust list.

Last I've seen -- last half of 2013

Except it isn't because of one thing: DDR4 is a vastly different animal from its ancestors. DDR4 is doing away with the traditional memory channels and instead adopting PCI-Express like signaling. So all the sockets have pretty much no life once DDR4 hits (which allegedly is soon).

Besides, Socket 2011 has an annoying requirement: quad-channel RAM. Two RAM sticks is fine, but needing four?

Haswell is firm and will move to s1150, Broadwell is mapped next and the chipset moved from the motherboard to on-chip (maybe on-die, but that's too far out to see) making many boards unusable. s1155 still has legs for the next two years. I say that because Haswell is plotted for Mid-2013, but I see problems with it on the level of P4's introduction. We're kind of seeing that now with Ivybridge and it's relative performance gains and the SB 2600K still hanging in there tightly.

Not to dig too deeply into my thought process and what I envision (and avoiding specifics as much as I can), I have to be vague. Core count growth is one-lined (scratched out) currently for desktops. There's no real data to prove that the 6/12 CPU's out in the wild are doing any better than even native 4-core builds like the 2500K and are difficult to manufacture in numbers. It's a niche market with the only other competition for desktops basically giving up and regrouping. When you beat someone in a drag race by 10 car lengths, you just lift off the throttle while looking at other competition (markets) as you slow down.

Atom is being embarrassed by ARM/RISK, Qualcomm's Snapdragon, nVidia Tigra...the world is currently mobile and that's where the money's at. With MS's Win8 push and the required x86 and x86-64 ability, I just suspect Intel wants in on that far before they make some 8/16+ desktop CPU. They can ride off Apple for a bit IMO (huge release, Mac's get stuff we had for a while now on PC's). And least we forget the dire hope that Light...err...Thunderbolt actually takes off; which I really think has been handled wrong (Not Intel's full fault...just not implimented).

So basically, Desktop power houses will be slow until 2014 IMO. It's not bad because I feel we're already over powered as it is. A 3570K is hard to find the end of it's usage and it's just 4 native cores.

LGA2011 will be here at least through the end of 2014 in support of Ivy Bridge E CPU's (for which the lithography covers 8 cores, though only 6 may be enabled), which will supposedly work with current LGA2011 boards through BIOS updates.

LGA1155 on the other hand ends with Haswell in less than a year.

So LGA2011 has a longer planned life-cycle as of right now and may very well continue into the first half of 2015. This is why a lot of people are screaming "2011".

If you want DDR4, then it's time for a new socket.

With Quad-channel DDR3 - I don't see a need for DDR4 anytime soon. But I agree that it would be nice to see native USB 3.0 support & 6 Intel SATA III ports.

DDR4 makes the idea of memory channels obsolete by using point to point connections, akin to PCI-Express. Basically, if every stick of RAM was a channel, four sticks of RAM is quad channel. But who knows? Maybe it'll be like PCI-Express where the length of current DIMM slots is 2 or more "channels".

Until and IF app developers start to create their apps to take advantage of more than two cores, there is little reason for most to bother. As the gamers here have stated 2 cores does well for that purpose. Except for scientific/architectural, graphics art and video rendering I don't see much use for 4 or more cores. Unless you can't get two cores or you are an optimist and want to future proof a long term system.

As for DDR4, Samsung has had working engineering models for almost 2 years and I am sure plenty of testing by various concerns. I suspect both AMD and Intel still see way too much life in DDR3 to be in a hurry. Could be wrong but I suspect servers will be first out with DDR4.

From the white papers I have read it should be about four times faster transfer which is huge technologically but not really all that useful since RAM is not a bottleneck, and one cool thing each stick will be a "channel" so you just add a RAM stick to add RAM. No 2, 3 or 4 channel memory to balance.

Wow, thank you I just Googled the Brickland EX platform and today it will eventually support DDR4, but note I don't know whether that will be in 2013 because I suspect the first CPUs won't support it unless AMD comes out with it built in (as it did with native SATA III and native USB3 support forcing Intel's hand, just an Intel forced AMD's with built in graphics).

The problem with multi-core processors is that to take advantage of them, you need to find some way of running your routines in such a way that they aren't dependent on each other. For games, maybe the various routines have to be run in sequence, otherwise it won't work. But even so for games, the bottleneck is actually on the GPU most of the time, not the CPU. You have to go down to a Core i3 to make Battlefield 3 start choking, and even then it's not that much.

Intel didn't force AMD into built in graphics, AMD had been working on an APU since around 2008, and had their first working protoypes in 2009. Of course they didn't go to market till after Intel, & some people think that was on purpose in order to see what Intel would be bringing to the table. Of course, AMD's integrated graphics solution is vastly superior if you ignore the CPU performance side of the APU equation & that's not even looking at the upcomming A10 APU's...

Just as AMD has not forced Intel into SATA III or USB3.0 support... AMD has supported 6 onboard SATA III ports since their 800 series chipsets (when coupled with the 850 south bridge) z68 launched in May 2011 with native SATA III support for 2 ports. Was this in response to the 800 series chipsets from AMD??? I think not, but advantage AMD due to full SATA III support on all ports... yet a year later when Intel launched z77, still only 2 SATA III ports supported by chipset... kind of makes you think that Intel doesn't give a shit about what AMD is doing...

Native USB 3.0: AMD has only had it since FM1 platform was launched last November. That's right the 900 series chipset boards with USB3.0 all use a third-party chipset for USB 3.0 support. Gigabyte & ASRock uses an Etron chipset, ASUS uses an ASMedia chipset, MSI uses a NEC chipset etc... So in actuallity, AMD still does not have a mainstream board with native support for USB 3.0 and won't till 2013... & before you say yea, but... FM1 is a niche board that was already EOL when it was launched. FM2 will launch shortly, but its still a niche platform, and destined for a quick EOL as well.

AMD is going to have to triple their current performance while reducing power requirements if anything they do is going to have any affect on Intel. Right now Trinity is the best thing going for AMD, and it's not even out yet. But I'd rather pay more and go with LGA2011 than save money & go with Trinity because I do High Definition video editing & encoding, where 12 threads are fully utilized, & I already have 4 DDR3 DIMMS for quad-chanel support. "yea, but you should be looking at the upcomming Piledriver based FX-8350, not Trinity for video encoding"... yea, but who knows when those will actually be available, and the power requirements are still outrageous... Performance per $ & Performance per Watt will still go to Trinity, not the FX chips.... AMD is on the right path with Trinity, I just hope the new CPU architect AMD hired can latch on & improve it, then we might have some real competition going again & then Intel might start taking notice...

Well, if we ever get the whole GPGPU idea fully standardized, then multi-core CPUs will be rendered moot. A GPU can run circles around a CPU for SIMD tasks. At that point, you only need a multi-core CPU for handling multiple requests and stuff while GPUs take over the heavy lifting.