Intel Haswell-E Core i7-5960X CPU & X99 Chipset @ [H]

I'm really disappointed there was no 4790K tested here. I realize a lot of people don't claim there's much more OC potential, but mine is currently running at 4.9 GHz at 75c stable with just a H100i.
 
As tempting as it is to want to own all that 8 core goodness, I can't help but feel a bit disappointed as to where CPU's are these days. Simply put, it comes down to a lack of good old competition to drive innovation. There's just been a series of baby steps for years now and it's getting boring, I really wish AMD could get back in the game in a serious way...
 
AMD's best kind of got destroyed by everything.

Reading comprehension fail. The AMD chip actually sat between Haswell and Ivy Bridge on several benchmarks.

There is also the cost difference.

It says something when despite considering several generation old Intel chips that are cheaper than the rest... it's still cheaper to buy a motherboard and newer amd cpu than it is to buy an older intel cpu alone of equivalent performance.
 
why disappointed?.. [email protected] tested in the review = [email protected] its the same chip..

In theory same chip, however their thermal signatures are quite different according to Intel. Now whether their the different thermal temps affect clock speeds is a different story.


At the moment 5820K offers serious competition to the 4790K. The ivy X versions will drop in price (massive) as they seem to offer the same performance as the Haswell E Ks. AMD...enuff said.

This is going to be an amazing chip and socket as they can capitalize on the following:
broadwell is coming out Q4 14
only a "tock"
at best you get a smaller 14nm package
theyre staying on Z97
so if your in the market for a new system X99 is the way to go.
If your upgrading to Broadwell, not really a benefit from DC.
 
Last edited:
I'm one of those X58 holdouts that has been waiting anxiously for Haswell-E. I was waiting also for SB-E/X79 back in late 2011 (since I built my 920 rig in Jan 2009), but reading all the reviews on the launch of X79 left me discouraged enough to wait for the next generation of Intel's Enthusiast-class platform/chipset.

Now it has finally arrived and I have come down with a major case of Déjà vu. Kyle could barely contain his disappointment with Sandy Bridge-E in his review. As is often the case, [H]'s excellent analysis had a big impact on my purchasing decisions, even though at the time I was rather surprised at Kyle's harsh criticism of SB-E.

...we have a lot of past AMD fans and current Intel fans today looking for the next great thing, and hopefully that is Intel's new Core i7-3960X processor. I hate to tell you, while it is impressive, it is hardly a product many desktop users will be positioned to purchase. And honestly, it just left me kind of meh.
[Zing!]

But let me say this, while Intel has been beating the drum about this being the "Ultimate Desktop Processor for Gamers," I think that is a lot of horse shit. This Sandy Bridge E is not going to do much of anything for gamers if I am making the right guess based on what I have seen, possibly with one exception, and that is multi-GPU, multi-display gaming. And certainly we will investigate that. I have no reasons to believe Sandy Bridge E is going to do anything for a gamer that is already using a 2500K or 2600K that has been clocked up. Surely those two will save you some money and power as well. But I guess if you are looking for a new way to heat your computer room this winter, Sandy Bridge E should be on your short list. Maybe they will start selling these at Home Depot?
[Ouch!]

I am not sure who is supposed to buy a 3960X. I really do not see it benefiting gamers. I do not see it being a boon to overclocking enthusiasts due to price, power usage, and subsequently heat output. I guess if I sat around all day ripping Blu-ray disks and encoding those for torrent sites, it would be awesome. Maybe that could be Intel's new 3960X motto, "Sandy Bridge E, maximizing BitTorrent ratios, one desktop at a time." Meh. Let's see what the K series brings before we totally turn our noses up at this beast of a processor...that none of us really need, or I think even want. I think we have enough cores for now. Get your noses back on the grindstone and give us stellar IPC gains or even better, 5GHz stock clocks.
[Pow! Ooof!]

Reading this initial Haswell-E review, I cannot see what the big difference is from the launch of SB-E to warrant the Gold Award and such an about-face from Kyle. IPS has improved slightly, yes, but nothing remotely like the moves from Netburst --> Merom --> Nehalem (and to a lesser extent) --> LGA1155/1150. With a 140W TDP, this is another "Home Depot space heater" when OC'd. Anandtech provides some troubling numbers for Haswell-E overclocking:

"At 4.1 GHz, peak power is +104W over the system power draw at stock, with another 40W at 4.3 GHz. This shows that Haswell-E can be a power hog from even small overclocks, and thus users must have cooling to match. If we add the 140W TDP and the +140W more from the overclock (it would most likely be more than this due to the change of efficiency in the PSU curve), then a mildly overclocked CPU is fast approaching 300W. One can imagine that a highly clocked 4.7 GHz sample would be nearer 400W, and thus users should purchase power supplies to match."
...
"Another issue with Haswell-E is the current draw of the CPU. ASUS is stating that the standard current draw for the CPU can reach 25 amps, meaning that the power supply must be capable of supplying at least 30 amps on the EPS12V cable."
[Zap! Whammo!]

At this point, I am sorry I didn't take Kyle's advice and jump on the 2600K when I was waiting instead for X79. What are the odds that Intel will really shake things up (Nehalem-style) with it's next enthusiast platform (Skylake-E)? I mean, more than just a gimped, rebadged Xeon with a microarchitecture at least a year behind Intel's latest mainstream K part? (tick-tock indeed) I think my days of insisting on Intel's "Enthusiast" platform to replace my LGA1366 are over. Broadwell is most likely in my future. Gaming is the most computer-intensive thing I do for any length of time. My transcoding, editing, file compressing, and other CPU-limited tasks do not take up enough of my day to bother with the *-E chips' exorbitant prices (considering whole system) and power-draw/heat. And I've yet to hear of dual-channel memory bandwidth being a bottleneck IRL on non-server systems (synthetic tests love them, though).

Obviously, this is only one usage scenario, and others will certainly have a different basis for their cost-benefit analysis.

TL;DR: I don't see where the money/power/heat delta between LGA2011-3 vs 1150 has changed all that much from LGA2011-0 vs 1155. SB-E gets slammed on this basis while Haswell-E gets the Gold. Have IPC or stock clocks improved so much between them?
 
I can see the gold award. It has two more cores - 4 more threads of processing power while powerwise it is not much more. The multi-threaded benchmarks are much better in general. The platform is more up to date as well. Then there is DDR4 support, more PCI-E lanes etc. For gaming maybe having more 16x PCIe 3 is probably not going to give much but if you are using the GPU for other stuff or processing it should make a good difference. This is a more high end professional solution beyond just gaming.

Thanks to the staff for including AMD 9590, it does do better with faster memory and other tweaks but at least it was included for comparisons sake. I think for most of the graphs it would not have offset the bars significantly but the review was great.
 
Good review! Interesting choice about the AMD system. It can make it quite hard to tell similar hardware apart when there's one lagging behind, stretching the scale...especially those "lower is better" timed results.

For those who are interested though, I've taken the liberty of hacking up the ones that do fit OK:







So as not to spam, here are links to the rest:

http://picpaste.com/14092645759hVccBAGr0_4_2_AMD-9YE4pXSQ.png

http://picpaste.com/14092645759hVccBAGr0_4_3_AMD-BnPaJyiZ.png

http://picpaste.com/14092645759hVccBAGr0_4_4_AMD-yonIg4HN.png

http://picpaste.com/14092645759hVccBAGr0_5_1_AMD-NNuKedRw.png

http://picpaste.com/pics/14092645759hVccBAGr0_5_2_AMD-GjxXEgeL.1409454279.png

http://picpaste.com/pics/14092645759hVccBAGr0_5_3_AMD-KVdvZ5C0.1409454317.png

http://picpaste.com/pics/14092645759hVccBAGr0_5_4_AMD-JsWWDNmQ.1409454340.png
 
Anandtech provides some troubling numbers for Haswell-E overclocking

The Anandtech staff has a very basic understanding of overclocking, or at least that's what they show in their reviews. They are focused on providing OC results that even the most mongoloid of readers can achieve. I love the quality of Anandtech reviews overall, but they do not take OCing seriously. Look elsewhere for OCing information. 1.3v requirement for 4.4-4.6ghz, depending on the quality of the chip seems pretty good to me (those numbers are straight from Asus).


I don't see where the money/power/heat delta between LGA2011-3 vs 1150 has changed all that much from LGA2011-0 vs 1155. SB-E gets slammed on this basis while Haswell-E gets the Gold. Have IPC or stock clocks improved so much between them?

SB-E offered no core count increase over the previous generation. It offered no reason for both the enthusiast crowd or the mainstream gamer crowd to upgrade. That's why it got slammed.

Haswell-E, specifically the 5960x, is a perfect fit for the content creation/enthusiast crowd. It's a breath of fresh air compared to the last 3 years of relative crap we've gotten from Intel.
 
Going from X58 to the 5960k would be a huge step up and a no-brainer in my opinion if you can justify and afford it.

From what I have seen the 5960k is nearly twice as fast as the i7 4770k in some workloads: http://images.anandtech.com/graphs/graph8426/66933.png That is worthy of a gold rating IMO..

However if your primary use case is gaming then moving to X99 makes less sense unless you just want to "future proof" your setup or something, which usually is not a good idea in my opinion. With gaming IPC and raw clock speed is still king over core count, so it's likely Broadwell will be a better option than X99 in that case.

It's also an exciting platform for new system builders, since you can get a 6 core Intel CPU now for not that much more money over high end Z97. I only think the choice to upgrade is less clear if you're replacing a rig that already has an i7 chipset in it. Unless you NEED more cores at this point in tme, I'd wait it out... but that doesn't make this a disappointing release.
 
No he made it clear that you need to stop whining.

Any comment contrary to what the majority are saying can be termed "whining", especially when it concerns a product just given a Gold award.

As tempting as it is to want to own all that 8 core goodness, I can't help but feel a bit disappointed as to where CPU's are these days. Simply put, it comes down to a lack of good old competition to drive innovation. There's just been a series of baby steps for years now and it's getting boring, I really wish AMD could get back in the game in a serious way...

I agree. Intel only releases its best when forced to do so by competition...which hasn't occured in a good long while. An example of this is when AMD launched their FX...Intel immediately responded by taking an internal Gallatin 2MB test chip (part of the Xeon MP line) that had been made into Socket 478 and paper-launched it as the first Extreme Edition (it was months before it was available in retail). When AMD launched the dual FX (2x4 platform), Intel immediately responded with Skulltrail, effortlessly converting the Penryn Extreme Edition into a Socket 771 variant. This shows how Intel can respond very quickly and release top shelf products if it wants to. What I'm advocating is for more of this side of Intel, an Intel whose top shelf products are configured to broaden their market reach, if possible, especially when there are people interested in them and can afford them. These products aren't cheap, nor should they be, but when the low end product isn't cutting it, Intel should be trying to make their most profitable products as appealing as possible to as much of the market as possible.

Now it has finally arrived and I have come down with a major case of Déjà vu. Kyle could barely contain his disappointment with Sandy Bridge-E in his review. As is often the case, [H]'s excellent analysis had a big impact on my purchasing decisions, even though at the time I was rather surprised at Kyle's harsh criticism of SB-E.

Reading this initial Haswell-E review, I cannot see what the big difference is from the launch of SB-E to warrant the Gold Award and such an about-face from Kyle. IPS has improved slightly, yes, but nothing remotely like the moves from Netburst --> Merom --> Nehalem (and to a lesser extent) --> LGA1155/1150. With a 140W TDP, this is another "Home Depot space heater" when OC'd. Anandtech provides some troubling numbers for Haswell-E overclocking:

Indeed. Many of the EXACT same arguments that were made against SB-E can easily be levelled against Haswell-E. The last 2-3 years were a long interlude with little overall improvement, making enthusiasts very hungry for any improvement to the HEDT platform. When that modest improvement arrives (as it has with Haswell-E), all that matters is the fact that this hunger has been sated for the time being, with few ill words or ire expressed toward those who caused and exacerbated this hunger in the first place. All that matters is the fact that the hunger has been alleviated. We were spoiled by the good progress made with Nehalem and Westmere and had the i7-3960X actually been an 8-core back in 2011 it would likely have been met with the same (or worse) scorn as the actual 6-core was met with. But at the time enthisiasts hunger for improvement had been addressed by the additional cores present on Westmere and the advancements made in that generation of CPU. Ironically, at the end of the day, the more poorly that Intel treats enthusiasts by extending the interval between advancements, the more they'll be worshipped as silicon gods when they actual roll out even a modest improvement. I fear that the more that this phenomenon occurs, the greater the period of time between advancements will be (and quite likely the advancements will become more and more incremental as well).

My $0.02...;)
 
I fear that the more that this phenomenon occurs, the greater the period of time between advancements will be (and quite likely the advancements will become more and more incremental as well).

My $0.02...;)

Or maybe it's more simple. Maybe SB-E and IB-E sold like crap compared to x58. Intel likes money.

Also, the rate of advancement with processor tech is slowing down. This isn't some conspiracy theory - we are hitting the limits of silicon. Each new advancement outside of core count requires more effort than the last - for those advancements to not cost consumers an arm and a leg, we need to wait until Intel finds a cheap way of implementing those changes in their mainstream products.

Those thinking that intel could release an unlocked 18 core today that would be stable at high clock speeds really have no idea whats going on.


Indeed. Many of the EXACT same arguments that were made against SB-E can easily be levelled against Haswell-E.

Not really. Haswell-E makes sense for content creaters and enthusiasts. SB-E made sense for no one, as gamers already had an x58 or SB system, and content creators/enthusiasts had little reason to transition from x58 systems.
 
And here goes the 11th time with the same whining.

Hardly. I'm contributing to this civil conversation, which is a simple discussion of the facts. If you have nothing to add to it or anything intelligent to say, then STFU...
 
Or maybe it's more simple. Maybe SB-E and IB-E sold like crap compared to x58. Intel likes money.

Also, the rate of advancement with processor tech is slowing down. This isn't some conspiracy theory - we are hitting the limits of silicon. Each new advancement outside of core count requires more effort than the last - for those advancements to not cost consumers an arm and a leg, we need to wait until Intel finds a cheap way of implementing those changes in their mainstream products.

Those thinking that intel could release an unlocked 18 core today that would be stable at high clock speeds really have no idea whats going on.

Sure, but the fact of the matter is that an unlocked 18-core at high clock speeds exceeds the conventional TDP envelope. Enthusiast systems and high-end workstations typically have the headroom in their power supplies, VRMs and other components to permit and accomodate a processor operating outside of this conventional envelope. It's done all the time with your typical overclocked enthusiast system. An unlocked 18-core permits use of the processor inside that conventional TDP envelope, while permitting those with the hardware and expertise to operate outside of that envelope the ability to do so.

Not really. Haswell-E makes sense for content creaters and enthusiasts. SB-E made sense for no one, as gamers already had an x58 or SB system, and content creators/enthusiasts had little reason to transition from x58 systems.

SB-E was being directly compared to SB 2600K in the review, with SB-E having two more cores than a 2600K and being lambasted for it as hot and unnecessary. HW-E has a further two cores (now four more than the mainstream 4770K) and is praised for it.:confused: Given that the mainstream chips for Haswell-E are still 6-core, most people have no reason to transition from either a SB-E or Ivy-E system unless they need the 8-core, which many people have mentioned, is more than the average enthusiast needs.
 
Sure, but the fact of the matter is that an unlocked 18-core at high clock speeds exceeds the conventional TDP envelope. Enthusiast systems and high-end workstations typically have the headroom in their power supplies, VRMs and other components to permit and accomodate a processor operating outside of this conventional envelope. It's done all the time with your typical overclocked enthusiast system. An unlocked 18-core permits use of the processor inside that conventional TDP envelope, while permitting those with the hardware and expertise to operate outside of that envelope the ability to do so.

Correct, let me clarify high clockspeed - anything over ~3.8ghz. The heat output and voltage requirements at that clockspeed would be immense, so immense that it puts it out of reach of even 99% of the enthusiast community based on technical issues if the price itself doesnt do the job. A product that specific makes no sense for Intel to create or support.

Matter of fact, when have we had unlocked server processors with higher core-count than the HEDT option? I don't recall any. So assuming that higher core counts will clock well, even with appropriate cooling, is a very very bold assumption with nothing to back it up.

Another reason we arn't seeing unlocked server processors any longer is because it can delay server upgrade cycles unexpectedly. Why upgrade the server CPU to one 20% faster if we can OC our current processors by 20% for free, assuming cooling etc can handle it? I know if I was a server admin, I'd recommend the latter and ask for a raise for saving my company money. All depends on electricity costs i suppose, but the short term savings would be immense, potentially large enough to warrant skipping a processor generation.

SB-E was being directly compared to SB 2600K in the review, with SB-E having two more cores than a 2600K and being lambasted for it as hot and unnecessary. HW-E has a further two cores (now four more than the mainstream 4770K) and is praised for it.:confused: Given that the mainstream chips for Haswell-E are still 6-core, most people have no reason to transition from either a SB-E or Ivy-E system unless they need the 8-core, which many people have mentioned, is more than the average enthusiast needs.

Anyone who had a x58 6-core system for content creation had little reason to upgrade to SB-E.

Anyone who had a SB gaming system had no reason to upgrade to SB-E.

SB-E was a processor line designed for literally no one and it was sold as an Enthusiast product. This is why HardOCP justifiably slammed SB-E - it was not exciting as an enthusiast product and no one would see a large advantage by upgrading to the platform.

Additionally, SB-E was releasing right when IB was releasing ,giving even less of a reason for people to go with the SB-E platform. That generational overlap problem is not happening here.

It's not a parallel comparison between the SB-E release and Haswell-E release. We have a core-count increase over the previous 3 generations. This is raw power for content creators, not a minor IPC bump like we've gotten over the past few years.

Sure, people who only play games still have no reason to get excited over this chip. Who cares, it's not meant for your average gamer. We've known for years now that this is how intel plans to do things - mainstream, IPC focused chips for mainstream gamers, then 6mo-1year later we get the bigger die option of the same architecture for content creators and enthusiasts.

An enthusiast does not talk in terms of "need", either. It's more about "want". And we want more power, noticably more power than the last generation. Haswell-E delivers that for me as an Enthusiast and content creator.
 
Last edited:
I can't see all the exciting breakthrough about H-E platform and X99 chipset.
X99 bring USB 3.0, 10 ports SATA 6G. 8 core (@$1000 btw) yes it's a step forward, maybe 2, from X79 and IB-E on the paper.
But if you think about the DMI interface CPU-PCH limited at only 2GB/s, the fact that x99 will be here for at least 3 years (Broadwell-E will run on the same platform and it'll not be out maybe until 1Q 2016), don't look so much better that a X79 + USB 3.0 card (standard on every board) and a SAS/SATA 8 port controller card on the pci-e.

There's rumors about skylake desktop available next summers, if broadwell-K is cancelled or short lived and we jump at Skylake-K at the end of 2015, we're talking about 2 generation ahead.
Intel want to push for DDR4 and lower the price so if skylake is ready make sense skip broadwell and push out as soon as it's ready.
4 SSD will saturate the DMI interface. So you have 10 for ? and if you add the USB 3.0 ports and 8 lane of PCI 2.0 traffic (Nic, audio, extra controler) , maybe not even 3 SSD will be running at full speed at best. (Forget 4x RAID0).
This is pretty ridiculous.

On X79/C600 there's a dedicated extra 4 PCI-E lane to connect the PCH to CPU for the SCU of extra SATA ports (disable to X79), on X99 I have not see it in any schematics
You can maybe get better performance from C600 board Vs X99. No good.
There's DDR4 and sky-high prices.
Clock-for-clock is 8% over IV-E and the Power rating is up from IV-E, even more under OC conditions.

On X79 and IV-E you'll sure get pretty good deals in the next few months.
If you don't want only the latest and greatest maybe a X-79/4930K a bargain-based prices can still be relevant awaiting the skylake-E and new chipset.

IMHO the X99 is the let down here. A Big let down in 2 years. Haswell and iVR 'failure' was expected, too much heat, and I'm expecting that Broadwell and 14nm will set a new records about it.
 
Last edited:
Correct, let me clarify high clockspeed - anything over ~3.8ghz. The heat output and voltage requirements at that clockspeed would be immense, so immense that it puts it out of reach of even 99% of the enthusiast community based on technical issues if the price itself doesnt do the job. A product that specific makes no sense for Intel to create or support.

Matter of fact, when have we had unlocked server processors with higher core-count than the HEDT option? I don't recall any. So assuming that higher core counts will clock well, even with appropriate cooling, is a very very bold assumption with nothing to back it up.

The E5-2699 V3 is slated as an OEM only processor, with no retail version. Intel doesn't support OEM chips anyway...that task is left to the OEMs. The ES version of this chip that coolaler tested overclocked to 2940MHz on a mere 0.895V. While the power consumption of this chip is much higher than an HEDT chip, this still shows clearly that there is a great deal of headroom left in this chip, headroom that is inaccessible. Most people who would buy this chip to overclock, would simply use the headroom and call it a day. Pushing a $4K CPU hard wouldn't be too smart...;)

Another reason we arn't seeing unlocked server processors any longer is because it can delay server upgrade cycles unexpectedly. Why upgrade the server CPU to one 20% faster if we can OC our current processors by 20% for free, assuming cooling etc can handle it? I know if I was a server admin, I'd recommend the latter and ask for a raise for saving my company money. All depends on electricity costs i suppose, but the short term savings would be immense, potentially large enough to warrant skipping a processor generation.

Server boards completely lack the ability to overclock, so the server admin you mentioned has no way to exploit the unlocked nature of the chip. Only a couple dual boards are capable of it, the last ones being the SR-2 and the Z9PE-D8, two boards that are incredibly unlikely to end up in a corporate server. Only a very few users would overclock this chip if it were unlocked, and those users are typically always looking for performance increases, so they're unlikely not to upgrade to the next generation of CPU. In any event, if being unlocked would encourage these users to buy a few of them, then Intel directly benefits, as these chips are their highest margin E5 SKUs.


Anyone who had a x58 6-core system for content creation had little reason to upgrade to SB-E.

True enough, except those who could benefit from the SATAIII ports. There were some modest performance increases with SB-E...more so than going from SB-E to Ivy-E.

Anyone who had a SB gaming system had no reason to upgrade to SB-E.

True...but anyone who owns a mainstream Haswell-K system has no reason to upgrade either.

SB-E was a processor line designed for literally no one and it was sold as an Enthusiast product. This is why HardOCP justifiably slammed SB-E - it was not exciting as an enthusiast product and no one would see a large advantage by upgrading to the platform.

Additionally, SB-E was releasing right when IB was releasing ,giving even less of a reason for people to go with the SB-E platform. That generational overlap problem is not happening here.

True enough. Intel is slowly bringing their different processor generations into closer alignment. Broadwell-K should be released now toward the middle of next year from what I was reading. This is a very welcome and excellent thing, as it gives HEDT purchasers more value and keeps the processors on it better aligned with the mainstream processors. However, Haswell has been out in the mainstream for well over a year and those people who have a K SKU-based system based on it who have no need of the extra cores have little reason to move to Haswell-E.

It's not a parallel comparison between the SB-E release and Haswell-E release. We have a core-count increase over the previous 3 generations. This is raw power for content creators, not a minor IPC bump like we've gotten over the past few years.

Sure, people who only play games still have no reason to get excited over this chip. Who cares, it's not meant for your average gamer. We've known for years now that this is how intel plans to do things - mainstream, IPC focused chips for mainstream gamers, then 6mo-1year later we get the bigger die option of the same architecture for content creators and enthusiasts.

For most users (6-core purchasers) of Haswell-E, the benefits are mainly in the platform, but those are tempered somewhat by the lousy DMI interface and high DDR4 prices. For 8-core buyers who can exploit the additional cores/threads, there is a bigger benefit. SB-E brought a fair number of decent (although not earth shattering) increases to the table. The i7-5960X does the same except that the gains are primarily from threaded workloads that can use all of the threads (not all multi-threaded workloads can take advantage of 16 threads) so in reality, only those applications that can will show the full benefit of the core count increase.

An enthusiast does not talk in terms of "need", either. It's more about "want". And we want more power, noticably more power than the last generation. Haswell-E delivers that for me as an Enthusiast and content creator.

I agree, although many people who argue against unlocking the top SKU Xeon consistently argue that there's no need, or demand to know what workloads you're running on the processor would justify having the chip unlocked. Part of my desire for an unlocked, top SKU Xeon is indeed want. I want to have a CPU capable of enormous power at my disposal when I need it that's capable of tackling any workload, single-threaded, mixed or multi-threaded, and no matter how heavy. Personally, when I buy a product, I want a company's best. And I'm willing to pay a very high price to obtain that in this case, indeed whatever Intel wants to charge, and many others are in the same boat. Intel just chooses not to offer their best in an unlocked configuration.
 
And I'm still having a hard time trying to find a 1650 or 1660. :(

Broadwell will probably be the only justifiable upgrade option for current high spec'd X79 owners.

Why won't Intel just release an overclockable 8C/16T non Xeon, Haswell for X79?
 
And I'm still having a hard time trying to find a 1650 or 1660. :(

Broadwell will probably be the only justifiable upgrade option for current high spec'd X79 owners.

Why won't Intel just release an overclockable 8C/16T non Xeon, Haswell for X79?

There were two on Ebay recently, sold for about $500 USD each. Intel is done with X79...they won't release anything else for that chipset because it's going to be discontinued and besides, they want you to buy the i7-5960X instead.;)
 
From what I see (after going back and reading the reviews on all the SB-E, IB-E, and Haswell-E again) Kyle gave the gold award to Haswell-E because it shows improvements which haven't been seen since the x58 days. Look at the SB-E review of the 3960X, even during content creation testing where the cores/threads were fully leveraged, there was only about a 20% difference between the 3960X and the 2600K - not very impressive considering the extra horsepower under the hood.

Now look at the content/multimedia tests of the 5960X review, you're seeing on average a 30% difference between the 5960X and the 4960X (same situation as the 3960X to 2600K - two extra cores under the hood) and on average a doubling of performance compared to the mainstream Haswell 4C/8T chip. Is a lot of that core count, fuck yea, but to go from a 20% difference between mainstream and top-end HEDT processors to a 100% difference between mainstream and top-end HEDT processors is simply fucking nuts considering the price for the X branded processors has stayed at the $1k mark.

Take my last sentence and couple that with the facts that the X99 chipset actually adds to and retains features already present on the mainstream Z87/Z97 platforms (go Wikipedia the different sockets for feature set comparisons) and it makes sense why the gold award was given. Intel actually got their shit together on this one and handed out a product that, if your workload demands it, is actually a very good upgrade from the previous generations X branded SKUs, and a fucking gargantuan upgrade from the 1 year old mainstream architecture it was based on - meaning its potentially worth the $$ to upgrade from a Haswell to a Haswell-E system if you really do need the processing power. You couldn't justifiably say that about going X58 (the Xeon OCers club will agree) to X79 and you certainly couldn't say that about going Z68 to X79.

TLDR: Is it to little too late from Intel to get rid of the sour taste of the past 3 years - maybe for some. But is it a step in the right direction for Intel to win back the enthusiast and give people a reason to get excited about the new releases and upgrade - absolutely. My only hope is Intel continues on this brighter path (though I do secretly hope I will be able to hold onto my future 5960X for 4-5 years before I really start to feel like an upgrade is warranted).
 
and thats why you cant compare a [H] review with Toms.. the E5-2678W its higher clocked than the 5960X so 3.2ghz vs 3.8ghz,under heavy load.. that remember me why i never liked the Toms review..

i really don't get where you took your numbers from. 5960x is 3.0/3.5GHz 8c16t. AFAIK< a 5960x can do 3.3GHz turbo on all cores. E5-2687w v2 is 3.4/4.0GHz 8c/16.It can only 3.6GHz turbo on all 8 cores.

The perceived performance gap can as you said, result of lower turbo bins, but remember that e5-2687w v2 run at stock DDR3 speeds, so at least some of the numbers on fritz benchmark are cache size related.
 
Have to agree that, while a great offering overall and perfect for *new* enthusiasts, the Haswell-E / x99 package doesn't offer much to current x79 owners

I was excited to upgrade from x79/3960x, but after double clicking on the real world facts there isn't a lot there - especially for the cost

My x79 rig is running fine with tri-SLI titan in pcie 3 mode and most OEMs (Asus in my case) have already delivered USB 3 and 6Gb SATA on x79. DDR4 is nice, but overall platform delta is negligible so the cost doesn't make sense.

This is actually good honestly. Getting more useful life out of a platform is a good thing and in the meantime folks who were still holding out on more legacy packages now have a very compelling entry point.
 
I don't know... Haswell-E does look kinda nice, and if I want to compete with the big boys on hwbot, I need to have that 8-core Haswell-E under the hood.

But as a current x79 user (3930K, R4G, 2x290X), I don't really see why I should upgrade to x99, unless I really want to compete again. I just can't fathom those DDR4 prices.
 
I have founded the 5930k and the x99 deluxe from asus but I can't find a shop that has ddr4 available here in italy
 
Thanks for your review Kyle, though I disagree that this is an 'upgrade' chip, or worthy of any awards. However, if I was looking to build a new, will last me a few years system, this would be a contender.
 
However, if I was looking to build a new, will last me a few years system, this would be a contender.

That right there pretty much sums up why it got the award. It's not like the X58-X78 transition where it just wasn't conducive at all to upgrade if you had top of the line X58 parts or top of the line Z68 parts that were great overclockers. This is the first really shiny thing being thrown to X58 guys in what, 5 years? And to guys itching to get into the HEDT platform but didn't because of the aforementioned lack of attractiveness it's the perfect answer for now.

There are a few conditions to which I believe this is a fairly good and intelligent upgrade though. What are everybody's else's thoughts/scenarios?

1) You're still on X58
2) You're on X79, Z68 or older and have been searching for the rig to build to get you through the next 4-5 years no issue.
3) You're on X79 and have been looking for a good reason to spend money on performance increases that aren't a disgrace to the definition of the enthusiast and your wallet allows you to afford the cost of entry into the platform
4) You're on Z77 or newer and have been interested in making the jump to the HEDT platform, but never could justify $500-1k on a CPU due to budget constraints, had real world usage scenarios that didn't justify the price/performance ratio over the Z series chipsets, or never liked how far behind features wise X79 was compared to the mainstream Z series.
5) You either give zero fucks or have more money than sense and want the best performing chipset and CPU on the market - but lack that Disney $$ and can't afford a $2k-4k Xeon.

I'm a mix of #'s 2, 4, and 5 myself :D I want my next rig to last me 4-5 years whilst still kicking ass and taking names of all the heavy duty work I throw at it, I've been interested in going to the HEDT segment for a couple years but $$ and reviews steered me to the mainstream, and I have more money than I need in order to max out 401k contributions and still live comfortably and just gotta have the bragging rights.
 
to give you an idea of what performance could you expect in firestrike from a 5960X at stock clocks it score 17000 points.. and overclocked to 4.5ghz 21000 points...


Ok.. I played with my ram some and i'm at 1600Mhz now and cpu is almost 4.2Ghz without turbo and it scored ( Physics Score14434 )

I know some of the others are running there Xeons faster then mine and could get closer then me as I;m just trying to see the upgrade in gaming for a x58 user.

http://www.3dmark.com/fs/2680901
 
Ok.. I played with my ram some and i'm at 1600Mhz now and cpu is almost 4.2Ghz without turbo and it scored ( Physics Score14434 )

I know some of the others are running there Xeons faster then mine and could get closer then me as I;m just trying to see the upgrade in gaming for a x58 user.

http://www.3dmark.com/fs/2680901

for a X58 (with 6c/12t, because quad core+HT even sandy its a upgrade) the gaming upgrade will be noticeable even with a 4690K in 95% of the games, i think only crysis 3 may be comparable in gaming performance to any actual chip and even i think a 4790K at 4.5ghz can be a good upgrade over 1366 6c/12t for gaming.. so a 5960X for gaming its completely worthless.

anyway 3dmark its far away to be a good indicative of CPU performance in games....

you may want to compare your 6c/12t to a 4930K at stock speed.. and see how it perform vs a 4790K at stock speed this may give you an idea:

intel_i74790k-31.jpg


intel_i74790k-33.jpg


intel_i74790k-35.jpg


intel_i74790k-34.jpg


intel_i74790k-30.jpg


anyway check This page which its very recent to check several gaming test across a wide variety of intel and AMD chips..
 
I pulled the trigger today. I'm on Z97 right now but my upgrade was somewhat forced as my X58 motherboard died late last year. I plan on going tri-SLI with the next round of NV cards and although PCI-E x8 right now isn't a huge limitation, it might be with newer cards. Plus, I work with some large SQL databases on my PC and 32GB RAM is ideal right now, however I may need to upgrade to 64GB in the future so this will give me the room to do so. I'd like this build to last as long as X58 did.

So, ended up going with the 5930k, ASUS X99 Deluxe motherboard, and a 4x8GB pack of GSkill RAM.

I think it is worth it in my case. Newegg's $150 off $500 promo right now if you pay in Bitcoin made the decision a little easier :)
 
Just checked out the prices of the 5930k here and they are $600, the latest Z99 mobo's start from $450. This is X58 all over again.
 
for a X58 (with 6c/12t, because quad core+HT even sandy its a upgrade) the gaming upgrade will be noticeable even with a 4690K in 95% of the games, i think only crysis 3 may be comparable in gaming performance to any actual chip and even i think a 4790K at 4.5ghz can be a good upgrade over 1366 6c/12t for gaming.. so a 5960X for gaming its completely worthless.

anyway 3dmark its far away to be a good indicative of CPU performance in games....

you may want to compare your 6c/12t to a 4930K at stock speed.. and see how it perform vs a 4790K at stock speed this may give you an idea:

intel_i74790k-31.jpg


intel_i74790k-33.jpg


intel_i74790k-35.jpg


intel_i74790k-34.jpg


intel_i74790k-30.jpg


anyway check This page which its very recent to check several gaming test across a wide variety of intel and AMD chips..


It would be hard to use those benchmarks unless there built into the game as there is no way of knowing if those are made up time demos and I don't have a 780Ti..

I wouldn't judge the Xeon as being slow 5 year old tech as it is a 32nm chip and not a 45nm like my old i7-930 ..
 
I have been tightly holding onto my i7 920/x58 setup....alas my wait is finally over!
 
Any comment contrary to what the majority are saying can be termed "whining", especially when it concerns a product just given a Gold award.

It's not so much you don't agree.

It's that you essentially copy-pasta'ed your post around 10 times in different threads.

At that point it becomes "Wah wah" *Thumbsuck* :rolleyes:
 
Coming from an extremely long-in-the-tooth x58 setup, I'm really torn between Haswell-E and Devil's Canyon chips. I do some occasional work in Lightroom, but 99% of my PC time is gaming so it seems like the premium on the Haswell-E isn't worth it for me. At the same time, the feature set is much richer on the x99 boards and DDR4 is the way of the future.... decisions decisions...
 
Back
Top