anandtechIntel Core i7 4960X (Ivy Bridge E) Review

[Smartazz]

What a let down. Nearly two years after Sandy Bridge-E and this is what we get.

 

[crashtech]

Depending on what you have now, the best move may be to wait for Haswell-E. Even if the mainstream desktop gets 14nm Broadwell, rather than just Haswell Refresh, there will probably be no significant improvement over Haswell. In fact overclocking may be worse, seeing as 22nm overclocks worse than 32nm. Meanwhile, the new high-end platform will catch up to and surpass the current Z87 chipset thanks to SATA-E etc.

Yeah, by embracing I suppose I mean my doubts have been eliminated about quad-core Haswell's dominance over Sandy and Ivy quads. Haswell-E is far enough away to where I can be pretty sure the upgrade bug will bite before then. If so, 4770K it will likely be, unless there are game titles released this holiday season convince me otherwise. If there is a lot of multi-core support in the next round of games, the choice will become even more difficult.

 

[birthdaymonkey]

I really thought Ivy-E would release the full potential for OCing the Ivy Bridge architecture with a soldered IHS. If 4.4GHz is the best it can do, it's certainly a disappointment.

 

[crashtech]

I really thought Ivy-E would release the full potential for OCing the Ivy Bridge architecture with a soldered IHS. If 4.4GHz is the best it can do, it's certainly a disappointment.

That's still a bit of a question mark. Perhaps the heat generated by those extra cores negate the advantage of a soldered IHS, but then again production samples might do better.

 

[beginner99]

dead on arrival

lol. yeah you found the most compact way of the same thing I wanted to say.

This chip is pretty much useless. Outdated platform and only advantage it has is relevant for niche use cases. Anyone needing those PCIe lanes probably already has a SB-E and 6-cores, well for heavy encodes it's great but else? And if you don't need best quality QuickSync in haswell will utterly demolish this.

IMHO this is mainly why intel sales are declining. if enthusiasts see no point in upgrading, what about regular consumers? if you don't play games or do heavy video encoding, your basically fine with the first core 2 duos from 2006...

 

[JimmiG]

That's still a bit of a question mark. Perhaps the heat generated by those extra cores negate the advantage of a soldered IHS, but then again production samples might do better.

Reviewers didn't seem to have any problems with heat, though. It looks like the CPU simply refused to stabilize above a certain frequency no matter what voltage they used.

 

[Fjodor2001]

At $990 they've at least got some nice margins on the 4960X. But I don't think they'll sell many of them at that price.

If they'd have a 6C Haswell on an updated chipset selling for $450-500 or so instead the margins would be less, but they'd also sell more of them. So maybe they would have made more profit in the end that way?

 

[Xpage]

disappointing, i'll be keeping my 2500k for a long time i think

 

[toyota]

disappointing, i'll be keeping my 2500k for a long time i think

lol like you were even considering this in the first place?

anyway, what a waste of money if gaming is the main goal as the 4770k makes much more sense. and whats up with the poor overclocking? what a joke for a flagship cpu and almost no progress has been made in 2 years as for as gaming goes.

 

[kimmel]

If they'd have a 6C Haswell on an updated chipset selling for $450-500 or so instead the margins would be less, but they'd also sell more of them. So maybe they would have made more profit in the end that way?

If they thought people would buy a 6C Haswell they would have released one. They must have different demand numbers and decided that releasing a 6C Haswell would have been a waste of time and effort and money.

The market has spoken: cheap, low power, slow cpus sell. High performance, high core count expensive server chips sell. Everything else doesn't really matter at the moment. That includes the vast majority of the enthusiasts in this forum. The mass market doesn't want to subsidize your chips anymore. Sorry.

 

[oceanside]

Pretty tame.

Smaller lithography is yielding diminishing results for the power system builder. Seems Intel is still suffering from leaky 3D finfets and the maturation of the 22nm process is a stubborn beast.

Hope they have better success when Has-E rolls around.

 

[Arachnotronic]

Pretty tame.

Smaller lithography is yielding diminishing results for the power system builder. Seems Intel is still suffering from leaky 3D finfets and the maturation of the 22nm process is a stubborn beast.

Hope they have better success when Has-E rolls around.

"leaky 3D finfets" - are you serious? Maybe at clock speeds far beyond what the designs/process tech were aimed at, but at stock (and in low power situations), the FinFETs are anything but "leaky".

 

[blackened23]

Pretty tame.

Smaller lithography is yielding diminishing results for the power system builder. Seems Intel is still suffering from leaky 3D finfets and the maturation of the 22nm process is a stubborn beast.

Hope they have better success when Has-E rolls around.

This is a joke, isn't it? Intel's 22nm process is not "leaky", it is the exact opposite. It is optimized for mobility and efficiency and excels in both of those areas.

You should do some research into the market. The desktop market is dipping by 20% worldwide by quarter (higher in the US) while mobile sales are skyrocketing - therefore with that being the case, intel designs their uarch around efficiency and mobility. Efficiency, by definition, is not "leaky". Silicon with high leakage is not efficient and has power consumption in the stratosphere at stock clockspeeds, so that is not the case with intel's uarch. The efficiency also has the added benefit of being good for enterprise and data centers - just FYI, efficiency is super important for the corporate sector these days. Aside from that fact - witness the fact that Haswell has a mobile battery life of 12-13 hours on mobile SKUs. That is where intel is headed - efficiency, super long battery life; broadwell will improve even further on this.

Intel is doing smart business and catering their products exactly where they need to - I'm sorry I love desktop, but you'd have to wear blinders to not see where the market is headed. Again, with the market conditions - intel is designing every uarch from the ground up with efficiency (this, again, by definition requires low leakage) for the mobile and enterprise markets.

 

[Dave3000]

lol like you were even considering this in the first place?

anyway, what a waste of money if gaming is the main goal as the 4770k makes much more sense. and whats up with the poor overclocking? what a joke for a flagship cpu and almost no progress has been made in 2 years as for as gaming goes.

I agree with you. I right now own an i7-3930k and was thinking about replacing my 3930k with a 4820k as all I care about is gaming performance but now I don't think it's worth it as the 3930k is a little better at the games that were being tested on one of the Ivybridge-E reviews than the 4820k and the 4930k was only a little bit faster than the 3930k in everything. I don't even think it's worth it for me to upgrade to the 4930k.

I also wonder if Turbo boost was disabled in the 4820k in one of the reviews because CPU-Z only showed 3.7GHz in the review and not the the turbo speed and it was also slightly slower than the 3770k. This makes me wonder if the 4820k had it's turbo disabled in the review. Is turbo disabled on the 4820k when all 4 cores are loaded?

 

[blackened23]

Indeed. Enthusiast CPUs are not about gaming. Once you get into the realm of video encoding and editing (among many other things), they make a lot of sense.

That is, until games start taking advantage of heavily threaded CPUs. Next gen consoles will definitely do this, but PC ports? Who knows. Maybe , maybe not. I'm guessing garbage PC ports will still be garbage, minimal effort PC ports while only using 2 cores at a time. Time and time again has shown that most developers put minimal effort into multiplatform PC ports, with some exceptions. The majority rule, however, is bad PC porting with terrible CPU utilization. I do hope i'm wrong at this time next year, though. I hope it changes.

**I should note :. Once you're into high surround resolutions, E CPUs make a lot of sense. When you have Tri SLI with 3 screens, EVERY SINGLE GAME will be CPU limited. So that is an argument for E/X CPUs. As well, a few games already are heavily MT/MC optimized. BF3 multiplayer benefits very well, crysis 3 does too. Maybe that trend will continue?

 

[SPBHM]

it's not a bad improvement in terms of power usage but... after what, 2 years? the overall performance gain, and using the same X79 is way to unimpressive... in a way skipping IB-E and keeping SB-E longer wouldn't make much of a difference... but perhaps they need to move production to 22nm or something!?

 

[blackened23]

it's not a bad improvement in terms of power usage but... after what, 2 years? the overall performance gain, and using the same X79 is way to unimpressive... in a way skipping IB-E and keeping SB-E longer wouldn't make much of a difference... but perhaps they need to move production to 22nm or something!?

The performance is in line with what we've known for a year now, and is not surprising. The uarch is designed for efficiency (ie MOBILE, ENTERPRISE), that's the bottom line. Mobile and enterprise sales = going up, desktop sales = going way down. It's the smart business move.

Intel should, IMO, release E and mainstream CPUs at the same time and that would make the E/X CPUs more compelling. That being said, IB-E is replacing SB-E. It wasn't designed as an outright upgrade for those with SB-E systems, I don't know why anyone has the mindset. If you're buying X79 now *new*, you're buying IVB-E. SB-E will be a legacy and end of life product soon, i'm sure, as IB-E (as mentioned) a mere replacement.

Haswell-E should be a compelling product regardless - it will have a ton of new technologies such as DDR-4 and SATA Express. Even though it's "late" in comparison to Haswell, it should be interesting.

 

[Idontcare]

In a very disappointing move, Intel confirmed to me that none of its own X79 boards will support Ivy Bridge E. I confirmed this myself by trying to boot a Core i7-4960X on my Intel DX79SI - the system wouldn’t POST. While most existing X79 motherboards will receive BIOS updates enabling IVB-E support, anyone who bought an Intel branded X79 motherboard is out of luck. Given that LGA-2011 owners are by definition some of the most profitable/influential/dedicated customers Intel has, I don’t think I need to point out how damaging this is to customer relations.

What a dick move by Intel :\

Would it have really killed them to release support for IB-E for the very customers that spent their mobo coin buying an Intel brand mobo?

All this communicates is: "dear valued customer, we will screw you over in a heart-beat if we can save two pennies in the process"

 

[Shmee]

What a dick move by Intel :\

Would it have really killed them to release support for IB-E for the very customers that spent their mobo coin buying an Intel brand mobo?

All this communicates is: "dear valued customer, we will screw you over in a heart-beat if we can save two pennies in the process"

Is there a possibility of it working on a user modded/custom bios?

 

[blackened23]

What a dick move by Intel :\

Would it have really killed them to release support for IB-E for the very customers that spent their mobo coin buying an Intel brand mobo?

All this communicates is: "dear valued customer, we will screw you over in a heart-beat if we can save two pennies in the process"

Wow. That is pretty unbelievable. Definitely a bad move by Intel.

 

[Makaveli]

If they thought people would buy a 6C Haswell they would have released one. They must have different demand numbers and decided that releasing a 6C Haswell would have been a waste of time and effort and money.

Do not agree.

 

[guskline]

Glad I have a 3770k! For the $$ ($229 at MC) gives me the ability to encode, video edit on occasion but is a great balance for the $$.

If I was building new now, unless I was doing primarily encoding, the 4770k makes more sense and cents ( or is that $$$).

 

[Kippa]

It all depends where you are coming from. If you hand a Sandy Bridge E the this doesn't look like a worth while upgrade. As for me I have an 6 core i7 970 pre Sandy so this does look quite interesting. As for me personally even though I have an i7 970 I am not going to upgrade to the Ivy E and wait for the Haswell E for the extra 2 more cores. I do 3d rendering and would see a love increase with 2 extra cores.

 

[exar333]

When does Haswell-E arrive?

 

[crashtech]

When does Haswell-E arrive?

About a year.