Tuesday, September 3rd 2013

Intel Core i7 "Ivy Bridge-E" HEDT Processors Start Selling

Intel made its newest Core i7 high-end desktop (HEDT) platform official with the launch of three new socket LGA2011 processors based on the swanky new 22 nm "Ivy Bridge-E" silicon. The launch includes the top-end Core i7-4960X Extreme Edition, priced at $990, followed by the Core i7-4930K at $555, and Core i7-4820K at $310. Expect a 10 percent markup across the board for these prices. Of these, the i7-4960X and i7-4930K are six-core parts, while the i7-4820K is quad-core.

The Core i7-4960X features a CPU clock speed of 3.60 GHz, with up to 4.00 GHz Turbo Boost frequency, 15 MB of L3 cache, and HyperThreading, which enables 12 logical CPUs for the OS to deal with. The i7-4930K clocks in at 3.40 GHz, with up to 3.90 GHz Turbo Boost, 12 MB of L3 cache, and HyperThreading. The i7-4820K, at its price point, can be extremely inviting for people with their minds set on a Core i7-4770K. It features 3.70 GHz clocks with up to 3.90 GHz Turbo Boost, 10 MB of L3 cache, and HyperThreading, enabling 8 logical CPUs. All three parts feature quad-channel DDR3 integrated memory controller with native support for DDR3-1866, 48-lane PCI-Express gen 3.0 root complexes, and 130W rated TDPs.
Add your own comment

36 Comments on Intel Core i7 "Ivy Bridge-E" HEDT Processors Start Selling

#1
lilhasselhoffer
...and a resounding meh...



I'm looking forward to seeing actual sales prices, actual performance numbers, and a comparison to SB-e. I can see not having any new PCH, but it's a shame. Hopefully IB-e will provide a more significant performance boost than its mainstream brethren. That TIM really held things back.

I'm really surprised that IB-e is actually cheaper than SB-e was (the x930k variants). Intel never seems to do that. Maybe how bad SB-e sold indicated something....


I guess the message is fingers crossed, but hopes set on low.
Posted on Reply
#2
buggalugs
lilhasselhoffer...and a resounding meh...



I'm looking forward to seeing actual sales prices, actual performance numbers, and a comparison to SB-e. .
Reviews are up and its another massive failure. Even Intels OWN slides says "18% lower performance than 4770K for everyday computing"

LOL, what a joke. Anyway its meant to be 5% faster than SB-e which is still slower than haswell for single threads. Boooo

I guess they have lost interest in the highend and they want us to do the same. Haswell here I come. You get faster performance 6x sata 6GB/s ports, native USB 3.0, much lower power and all the new add-on features like ac wifi, thunderbolt, better audio etc.
Posted on Reply
#3
Shinshin
We need AMD back in business!
Seriously!
Posted on Reply
#4
Kaynar
Wow the first reviews are so disappointing for the overclocking...

No review managed more than 4.5Ghz stable on the 4960X and the ones with 4820K can't even OC without a crash yet... what a sloppy job, hopefully some fixes will come for the BIOS of the motherboards to allow for something better...
Posted on Reply
#5
xvi
Not to invoke flaming, but the now old Phenom II X6, despite its IPC inefficiency compared to Intel offerings, still seems like a noteworthy competitor. The 4930K, for example, costs $555 ($610.50 if the 10% markup is to believed). IPC and HyperThreading should net a good gain in performance (50%?), but compared to a high end X6 (~$100), it just seems like highway robbery.

Coupled with the reports of unimpressive overclocking, I don't see what Intel is trying to pitch here. Performance per watt and out-of-box performance, I suppose.
Posted on Reply
#6
nickbaldwin86
Interested in the 4820k... the rest are WAY over budget... see what reviews have to say... I am not into building a system of these magnitude anyways
Posted on Reply
#7
Random Murderer
The Anti-Midas
/sigh. Time to straighten some things out.
lilhasselhoffer...and a resounding meh...

I'm looking forward to seeing actual sales prices, actual performance numbers, and a comparison to SB-e. I can see not having any new PCH, but it's a shame. Hopefully IB-e will provide a more significant performance boost than its mainstream brethren. That TIM really held things back.
The IHS is soldered. Temps are looking great on OC'd chips.
buggalugsReviews are up and its another massive failure. Even Intels OWN slides says "18% lower performance than 4770K for everyday computing"

LOL, what a joke. Anyway its meant to be 5% faster than SB-e which is still slower than haswell for single threads. Boooo

I guess they have lost interest in the highend and they want us to do the same. Haswell here I come. You get faster performance 6x sata 6GB/s ports, native USB 3.0, much lower power and all the new add-on features like ac wifi, thunderbolt, better audio etc.
While the 4820K is slower than the 4770K at stock, with both overclocked to the same speeds the 4820k should be a bit quicker thanks to the quad-channel RAM. Also, the -E variants of SB and IB get about the same boosts as the mainstream parts, a 4.5GHz IB-E is roughly the same as or slightly faster than an equivalent SB-E at 4.8GHz.
ShinshinWe need AMD back in business!
Seriously!
Back in business? They're still around and doing well, they're just not focused on super-high-end $1000 CPUs. Frankly, AMD's Price:Performance ratio is far better than Intel's. $200 for an 8-core FX8350 4.0GHz proc vs. the direct Intel competitor, the 3770K at $320.
KaynarWow the first reviews are so disappointing for the overclocking...

No review managed more than 4.5Ghz stable on the 4960X and the ones with 4820K can't even OC without a crash yet... what a sloppy job, hopefully some fixes will come for the BIOS of the motherboards to allow for something better...
Again, a 4.5GHz IB-E is roughly the same as a 4.8 SB-E. Not to mention reviewers generally don't have enough time to push the max stable clock of a proc before they have to send it back. Case and point: check out MetalRacer's clocks in this thread. 5.0-5.3 with RAM at 2400. Granted he's on SS, but there's definitely more to be had out of these processors than reviews are showing, especially considering he was running 5GHz at a measly 1.256 VCore.
That being said, BIOS updates and patience during overclocking(my 3820 was capped at 4630 until I updated BIOS a few months back, now it's running 4.9 24/7 stable) should get these chips higher than 4.5 for 24/7 use, not to mention some guys over on XS have had IB-E chips running 3000+ on memory which one cannot deny is a massive improvement over SB-E. 3000MHz quad-channel RAM...:twitch:


Guys, I guess what I'm trying to say here is that these are part of Intel's "tock" cycle, with the corresponding "tick" being SB and SB-E, and the following "tick" being Haswell. The "tock" cycle has always been a simple refresh with some slight improvements. If you were expecting this to be a Haswell killer, you expected too much.
Personally, I'm still torn between keeping my 24/7 4.9GHz 3820 or upgrading to a 4930K. It seems reviewers are focusing on the 4960X and 4820K right now, so experiences regarding the 4930K are still few and far between.
Posted on Reply
#8
EpicShweetness
It is hard to justify these processors for "work" especially when their own processors (4770k in particular) is 70-80% as fast with a 50% reduction in power consumption with a lower price tag. Never mind power consumption and AMD's processors with an even LOWER price are within 70-80% as powerful.
You can't help but just look and sigh :ohwell:
Posted on Reply
#9
fortiori
Random Murderer...While the 4820K is slower than the 4770K at stock, with both overclocked to the same speeds the 4820k should be a bit quicker thanks to the quad-channel RAM....
You used the word "should" and put it in italics which is spot on. Xbitlabs did some great work on evaluating the actual benefits of Quad Channel Ram. The relevant page is here:

www.xbitlabs.com/articles/memory/display/lga2011-ddr3_3.html#sect0

But the whole article is worth a read. The takeaway: Quad Channel RAM is only beneficial in a few rare scenarios. Quite disappointing.
Posted on Reply
#10
Random Murderer
The Anti-Midas
fortioriYou used the word "should" and put it in italics which is spot on. Xbitlabs did some great work on evaluating the actual benefits of Quad Channel Ram. The relevant page is here:

www.xbitlabs.com/articles/memory/display/lga2011-ddr3_3.html#sect0

But the whole article is worth a read. The takeaway: Quad Channel RAM is only beneficial in a few rare scenarios. Quite disappointing.
On the exact page you linked, there was only one(two?) test where quad-channel didn't come out on top, and quad-channel 1600 consistently beat out dual-channel 1866 except for in a few tests.
This is not a new topic of discussion, it seems to resurface every time more channels are added. It originally started years ago when dual-channel was first introduced, then again when the LGA1366 platform introduced triple-channel, and here we are discussing quad-channel. Also remember that the more threads that are used, the more potential bandwidth saturation, or, simply put, an application with multiple(4 or more) threads will run better on a system with quad-channel RAM than a dual- or triple-channel system with identical RAM modules at identical speeds and latencies. Other variables include motherboard, interleaving, banks, etc. There is no definitive test to prove what's best.
The fact of the matter is that they would have had substantially greater margins between each setup if they tested at, say, 2133 or higher. The higher the speed, the more channels, the greater the bandwidth. Simple as that. The fact is, until now ~2400 has been the speed cap of LGA2011(on most chips. mine will run 2520 and I've seen some pro-clockers get 2600ish). Quad-channel 2400 simply cannot put out the bandwidth of dual-channel 2800+ that IB and Haswell have been turning out. However, I expect to see the tables turn when people start clocking the hell out of these IB-E IMCs. Quad-channel 2800+ should give X79 the bandwidth crown back. Notice I used the italicized "should" again? ;)

Overall, I'm optimistic about the capabilities of IB-E, but still hesitant to plunk down $550+ for a proc until I see some solid numbers, benchies, and more reviews of the particular chip I'm looking at, the 4930K.
Posted on Reply
#11
MxPhenom 216
ASIC Engineer
I really like the power consumption improvements of these chips. Kind of the reason, aside from cost, that I didn't go SB-E when planning to go Haswell.
Posted on Reply
#12
hardcore_gamer
4820K looks like a better deal than 4770K; if you are willing to spend $200+ for the motherboard.
Posted on Reply
#13
Shinshin
Random MurdererQuote: Originally Posted by Shinshin
We need AMD back in business!
Seriously!
Back in business? They're still around and doing well, they're just not focused on super-high-end $1000 CPUs. Frankly, AMD's Price:Performance ratio is far better than Intel's. $200 for an 8-core FX8350 4.0GHz proc vs. the direct Intel competitor, the 3770K at $320.
Well, what i wanted to say is that we need AMD back in business not because AMD is doing bad (it certainly has a good lineup of APUs and CPUs) but because if we want to see Intel improving much more with each tick and tock then AMD should design a great CPU that will challenge Intel in the high end market. Like what happened with the first Athlons/Opterons where Intel was caught with their pants off...
Posted on Reply
#14
Wile E
Power User
Still holding out for dat octocore. LGA2011 is a total pass for me.
Posted on Reply
#15
buggalugs
Random Murderer/sigh. Time to straighten some things out.

.
Maybe you should straighten yourself out. INTELS OWN SLIDES says its comparing the 4960X to the 4770K



Uploaded with ImageShack.us

Its on the anandtech review.

And we knew ivy-e was going to suck like sb-e, we've been talking about it for months. Up until now its rumour. With the official release and official numbers Intel deserves an ass whoopin again.

There is lots of things they could have done to improve the platform, even if CPU performance wasn't great, like updating the old X79 chipset with better features.

In anyone's book "18% lower everyday performance than 4770K" or single threaded performance is a joke. This is the first time in my memory where this has been the case. SB-E was at least faster the SB, and SB-e was released before Ivy mainstream.

This is the first time where the already released mainstream platform is "18% faster" than the high end:" Intel deserve to be called out.
Posted on Reply
#16
Aquinus
Resident Wat-man
You know, a little perk with the 4820k, much like the 3820 is the extra L3 cache. In some instances, it does make a difference. Performance aside, LGA2011 is a nice platform for those of us who want the extra PCI-E lanes and would like to be able to expand memory beyond 32GB.

Haters are going to hate, but I still love my 3820 and performance aside, it still does everything I throw at it fairly well. :confused:

Don't you love it when people who don't own a platform complain about it?
buggalugsIntel deserve to be called out.
Or maybe people like you need to be called out for being a tool and talking about things that you don't have any first hand experience with. My 3820 can do anything almost as fast as a 4770k can and with 40 PCI-E lanes I couldn't give a rats ass if the PCH isn't spectacular. If I wanted more I would buy a RAID card since I have the PCI-E lanes for it. Don't talk about something if you don't even know the purpose of the platform in the first place.

Computers do not exist for gaming, so you should stop acting like they do.
Posted on Reply
#18
NeoXF
Wow... nearly abysmal... 2 years for this?

Need I go on about how outdated it's one (and only) supporting chipset also is (considering what a high-end platform this is supposed to be).

Only two reasons I see for getting a IVB-E is...

1. You absolutely need a ton of PCI-E lanes.
2. You don't have a SB-E system and you really really need the extra cores.
Posted on Reply
#19
Aquinus
Resident Wat-man
NeoXFNeed I go on about how outdated it's one (and only) supporting chipset also is (considering what a high-end platform this is supposed to be).
Anyone who actually knows what they're doing would rather have dedicated hardware than rely on the PCH to do so much.

...and no, you don't need to go on about it because the only people who complain about it are mostly gamers who won't use a PCI-E slot for anything other than a graphics card. :slap:

Once again, you're another person who is missing the point of a cut down xeon rig.
NeoXF1. You absolutely need a ton of PCI-E lanes.
2. You don't have a SB-E system and you really really need the extra cores.
You forgot the people who benefit from having >32Gb of RAM, which is only a handful of people, but the group of people who actually need a skt2011 platform is a very small number and it's worth remembering that and everyone seems to be forgetting that.

Most people only need a skt1155/1150 rig anyways and all of this complaining is like saying "OMG why does a 2Ghz 8-core Xeon CPU run Crysis like crap!?" and that's because you're completely missing the point of the platform. Remember that skt2011 is a server platform. :shadedshu

I do more development work and testing on my skt2011 rig than I game, that's for sure.
Posted on Reply
#20
ensabrenoir
I have two rigs one 1156 and one 2011. Seriously looking at that 6core ivy-e. My 3820 is nice but this is the reason i really built my 2011 rig. Really starting to wonder what the performance nay sayers are talking about..... wonder what they're using their rig for and what requires more performance than what intel is providing....
Posted on Reply
#21
buggalugs
ensabrenoirI have two rigs one 1156 and one 2011. Seriously looking at that 6core ivy-e. My 3820 is nice but this is the reason i really built my 2011 rig. Really starting to wonder what the performance nay sayers are talking about..... wonder what they're using their rig for and what requires more performance than what intel is providing....
I guess the problem is, why would you spend more money for less performance and less features?. Call me weird, but I would expect a $1000 CPU to perform better than a $380 one in all tasks, and I would expect the platform features (# of 6GB/s sata ports, native USB 3.0 thunderbolt etc) to be at least as good as the mainstream platform, instead of having worse/old/less features.
Posted on Reply
#22
Aquinus
Resident Wat-man
buggalugsI guess the problem is, why would you spend more money for less performance and less features?
First of all, it's not less performance and that is your first problem. There isn't a whole lot of variation between the 3770k/4770k/3820. So the first problem is that your missing the point of the platform (my third time saying it?). The PCH is not a final solution. The PCH offers everything that a regular user will need off the platform and the extra PCI-E lanes are to let you make your own decision about what you want attached to your computer.

Any system admin will tell you that running 6 drives off the PCH is an exceedingly stupid idea and when you only have 16 (20?) PCI-E lanes to use, your options are rather limited. With 40 PCI-E lanes you can easily just run it in a 16/8/8/8 configuration, at least on my P9X79 Deluxe without any PLX switch or anything special going on.

I don't know why you're saying there are less features because VT-d is a huge perk for people like me who run virtualized workloads for development. There is nothing worse than waiting for I/O because your virt environment sucks (believe me, I've been down that road.)

So between giving me the option to easily upgrade my machine with just about anything I want while offering features that I actively use for development, don't see how the platform is lacking in features or performance. It's a hollow argument.

The only thing I will complain about SB-E is the power consumption, which for me, isn't all that bad because my Phenom II 940 could draw almost as much and the only reason I would get an IVB-E is to get 2 more cores and/or to reduce power consumption.
buggalugsI would expect a $1000 CPU to perform better than a $380 one in all tasks
That's your own problem interpretting what it is to be used for. 6-core CPUs rip through multi-threaded workloads and the quad-channel memory helps feed that beast. It's noth worth while for people like you to get it because you'll never fully tax it. More cores means more heat and if you think that a quad-core will make just as much heat as a 6-core at the same speed, you're sadly mistaken.

Skt2011 isn't for the "LOL, I want a fast computer!!" user. It's for people who know what they need and anyone who truly knows what they need and believes skt2011 is a good match is thinking about a lot of things that gamers do not.
buggalugs(# of 6GB/s sata ports, native USB 3.0 thunderbolt etc) to be at least as good as the mainstream platform, instead of having worse/old/less features.
Old features... Really?

When are you planning on having more than 2 SSDs? I know that if I got more than two, I sure as hell wouldn't expect the PCH to keep up with it. I would want a real RAID controller that can handle that kind of bandwidth. It sounds like someone is forgetting that DMI 2.0 is still the limiting factor between the PCH and your drives.

I guess the point that I'm trying to make is that your average consumer wants what they'll need on the motherboard (ie. the PCH,) people who need skt2011 (for most cases,) won't want to use X79 for a number of things, like storage, because of how much better a real RAID controller can handle I/O. Not to mention you remove all the I/O from the DMI bus.

I should also add the twice as many pins on the CPU will require more layers on the PCB which alone costs more money. If you look at an X79 board you can tell there is no room to add more in most cases. They're jam packed full.

So instead of bashing a platform that performs just as well as anything out there, how about you do research and find out what it's used for before saying things along the line of "Well, it isn't good for regular tasks," because that isn't what it's designed to be good at.

Also, please tell me that my motherboard isn't jam packed full. I don't know where ASUS would have put anything else if they even tried and this motherboard has done everything that I've thrown at it.

Posted on Reply
#23
MarksmanR
I fail to comprehend why the extreme series CPUs are based on a previous generation architecture? Intel should have skipped Ivy Bridge-E and simply released Haswell-E.
Posted on Reply
#24
Aquinus
Resident Wat-man
MarksmanRI fail to comprehend why the extreme series CPUs are based on a previous generation architecture? Intel should have skipped Ivy Bridge-E and simply released Haswell-E.
Tell me more about what Haswell can do that IVB-E can't? Generations aside, they're very similar processors, much like how IVB and Haswell are very similar. :confused:
Posted on Reply
#25
radrok
MarksmanRI fail to comprehend why the extreme series CPUs are based on a previous generation architecture? Intel should have skipped Ivy Bridge-E and simply released Haswell-E.
Makes more sense for Intel to try new architectures on smaller dies than going into full size dies.

Also HEDT gets what's available from the Xeon lineup.

Ivy-e is a success in its intended segment, lower consumption is a boon when building stacked 4p rigs.
Posted on Reply
Add your own comment
Dec 24th, 2024 12:58 EST change timezone

New Forum Posts

Popular Reviews

Controversial News Posts