Report: Specifications of Ivy Bridge-E CPUs

[Valentin_N]

What a waste of time and money! They should have skipped Iyv-B and gone with Haswell! For that money, if anyone truly buys a i7-4970X, because you can get a good Xeon E5-26xx or v2, such as E5-2650 that has okey GHz and is a octacore for the same price, roughly. Or you can get two E5-2620 that are hexacores cpu with a too bit too low GHz.

 

[lp231]

[citation][nom]amuffin[/nom]Where dem 8 core's at?[/citation]
There with the Xeons. Some Xeon E5 are 8 cores.
My guess is when Haswell-E arrives, the Xeon based may have all 12 cores, while the desktop version will just have 8 cores (# of cores is just a example and does not represent the actual amount when Haswell-E arrives). It seems Intel want to keep it that way, their desktop CPU cores will always be less than a Xeon.

 

[georgekn3mp]

Good news for me...for later in 2014. I have the i7-3930k on LGA2011...and glad they are keeping that chipset for one more round.

I bought 3930k in mid-2012, and it's great, but 4930K just looks like a memory controller boost more than anything, since WHO would run a "K" at the stock speed anyway?

Since 3930K runs very stably at 4.6ghz on O/C, 4930K will hopefully do that at 4.7Ghz, UNLESS it's temp-limited like the other Ivy bridge chips.

For a $600 dollar enthusiast chip though, I really doubt they will cheap out on the thermals.

At least I have one more round "tick" to upgrade this time, not like my old X79 series i7-920 which never did have an architecture upgrade to just "drop in"

Sandy-Bridge-E to Ivy Bridge-E upgrade...worth the money? It is if the 3930k dies, don't have to upgrade the $350 motherboard at least to replace it, since the CPU can just be swapped up for one more generation.

 

[ojas]

Hmm. Well, on the bright side, overclocking for all, more bandwidth, and fewer models. When and if the 4970X comes out, it'll probably turbo beyond 4 GHz.

I remember reading about DDR4 memory support on some of the Xenons.

I do hope that Haswell-E is forced via AMD to arrive in the form of a 6C/12T and 8C/16T at minimum.

Do the X-series chips come with IGPs? If they don't then i think Haswell should enable 8 cores to stay within the 130w limit they seem to be so obsessed with staying within.

 

[dgingeri]

Again, pretty much no progress, same as the IB release and the next Haswell release. We're going nowhere because Intel has no competition from AMD. We're going to end up with a stall for another decade like the P4 situation. Is it even going to end this time? fifteen years ago, I had a PII-300Mhz overclocked to 450MHz, and I currently have a six core 3.2GHz system. I don't want to end up having a 3.8GHz six core system 15 years from now.

 

[DrShocker com]

Does Ivy Bridge-E Have more PCI lanes ? I was shocked when i used up all my 3960X lanes and needed more

 

[dextermat]

The problem is a lot of ignorant will buy them as they will fade out sandy and ivy...

just milking out the poor people who doesn't know more than to power up a computer

 

[f-14]

[citation][nom]Non-Euclidean[/nom]Right... Because of that overwhelming demand for 8 cores on the desktop. I know that Intel's stock price has been melting due to their lack of response to the marketplace.[/citation]
+1

also, if you clowns want to know where your 8 core chips are at, they are in the same place as the programmers that are coding for 8 cores = non-existent!

programmers coding for 4 cores are so few and far between it's not really worth making a 4 core chip for.

i remember the switch from 16 bit to 32 bit being pretty fast, but the switch to 64 bit hasn't even been fully integrated into programmers world. programmers have barely been scratching into 2 core coding.

seems programmers are demanding 6GHZ speed while all of you are running around like chickens with their heads cut off in pursuit of more cores that don't even matter!

What we don't see is a successor to the i7-3970X;

i see it, it's now called the i7-4960 it comes with a 100mhz boost on the low end with an increased FSB of 266mhz. my guess the reason for the change was an old ATI patent on the model number or Intel didn't want to associate with an ATI model number in a belief that it would create confusion and problems (ATI 4970 vs Intel 4970)

 

[goodguy713]

well my only thoughts on this is when they do release this hopefully you can get the 3930k a little cheaper then before there's not a whole lot of improvement here nothing you cant gain with a mild overclock. currently 530 dollars for a 3930k is a bit steep been waiting for them to go on sale or to find a real discount on one .. but as of yet I have not seen anything of the sort.. hopefully my luck changes .. I have been wanting to upgrade from my 1090t but am unwilling to buy a processor at that price and really want the full lanes and platform of the lga 2011 boards.

 

[InvalidError]

[citation][nom]A Bad Day[/nom]But that's too much work. If many ported games still only went up to DX9, what makes you think they'll put in extra effort to "fold" back in the threads?Also, folding threads into one giant one?[/citation]
Who said "into a giant one"?

Folding back threads does not mean going that far. To make effective use of a slow 8-core CPU, the console code will need to be finely threaded and that means delegating many trivial tasks in a way that would not be necessary on a desktop CPU. In some cases, removing the threading can be as simple as replacing the thread start function call by a straight call to the function the thread would be executing.

For physics and other libraries, reducing the level of threading can require as little as pre-processor switches or pragmas.

 

[InvalidError]

[citation][nom]dgingeri[/nom]Again, pretty much no progress, same as the IB release and the next Haswell release. We're going nowhere because Intel has no competition from AMD.[/citation]
For there to be competition, there has to be some degree of reciprocity.

If you are accusing Intel of "not competing" because AMD is not giving them any reason to, you also have to blame AMD for not trying hard enough.

The reality is AMD has gone pretty much as far as their R&D budget can afford and the silicon/power/area cost per incremental performance improvement is not getting any better for Intel either, so both sides are at a relative standstill.

Sure, they could add more cores. But the amount of mainstream software out there that can make reasonably decent use of them are much too few and far between to let mainstream parts cannibalize Xeon sales to pros who really need that much processing power and have the budget to put the money where their mouths are.

 

[A Bad Day]

[citation][nom]f-14[/nom]programmers have barely been scratching into 2 core coding.seems programmers are demanding 6GHZ speed while all of you are running around like chickens with their heads cut off in pursuit of more cores that don't even matter![/citation]

Because Intel did a great demonstration of what happens when you mash the accelerator and give a finger to IPC and thermals.

It's called Netburst. The architecture that was supposed to run up to 10 GHz, the 4-5 GHz version was a mini-oven.

Good luck fitting that thing in a standard desktop, nevertheless a laptop.

 

[blazorthon]

[citation][nom]dgingeri[/nom]Again, pretty much no progress, same as the IB release and the next Haswell release. We're going nowhere because Intel has no competition from AMD. We're going to end up with a stall for another decade like the P4 situation. Is it even going to end this time? fifteen years ago, I had a PII-300Mhz overclocked to 450MHz, and I currently have a six core 3.2GHz system. I don't want to end up having a 3.8GHz six core system 15 years from now.[/citation]

If the performance per Hx say doubles from the 3.2GHz to the 3.8GHz, hen the performance difference is still staggering. Complaining about clock frequency is nonsensical, especially since it can't keep going up without consequences in other things such as power consumption and more.

 

[blazorthon]

[citation][nom]A Bad Day[/nom]Because Intel did a great demonstration of what happens when you mash the accelerator and give a finger to IPC and thermals.It's called Netburst. The architecture that was supposed to run up to 10 GHz, the 4-5 GHz version was a mini-oven. Good luck fitting that thing in a standard desktop, nevertheless a laptop.[/citation

Hitting high frequencies doesn't necessarily mean giving the finger to IPC. Look at Ivy without Intel's crappy paste, hitting 5GHz to 6GHz isn't too difficult and that's still probably capable of using less power than Netburst did despite having two to four times as many cores. If we did a more direct comparison with only one or two cores in Ivy without Intel's crap paste, then I bet around 7GHz would be reasonably possible.

Sure, they're not the supposed 10GHz of Netburst, but they're high frequencies considering that Ivy Bridge has a huge increase in performance per core.

 

[blazorthon]

[citation][nom]blazorthon[/nom][citation][nom]A Bad Day[/nom]Because Intel did a great demonstration of what happens when you mash the accelerator and give a finger to IPC and thermals.It's called Netburst. The architecture that was supposed to run up to 10 GHz, the 4-5 GHz version was a mini-oven. Good luck fitting that thing in a standard desktop, nevertheless a laptop.[/citationHitting high frequencies doesn't necessarily mean giving the finger to IPC. Look at Ivy without Intel's crappy paste, hitting 5GHz to 6GHz isn't too difficult and that's still probably capable of using less power than Netburst did despite having two to four times as many cores. If we did a more direct comparison with only one or two cores in Ivy without Intel's crap paste, then I bet around 7GHz would be reasonably possible.Sure, they're not the supposed 10GHz of Netburst, but they're high frequencies considering that Ivy Bridge has a huge increase in performance per Hz.[/citation]

 

[blazorthon]

One major fact that many of the members who commented here seem to ignore is that should these Ivy Bridge E CPUs have the fluxless solder instead of crap paste between the IHS and the CPU die, they can actually have a big improvement in overclocking headroom compared to Sandy Bridge E.