Intel Kaby Lake Core i7-7700K IPC Review @ [H]

AMD, do we matter? Lisa Su? AMD has a hugely influential and substantial fanbase waiting to wave your flag again. We all still have that Blue Core Thunderbird and9700 Pro love in our hearts. We are older now and have lots of money to spend on tech and its toys. We are established, influential, and well informed, and all our family members and all their friends ask our advice on computer purchasing and then it trickles down. That is the HardOCP reader profile. Wouldn't you love to have us once again direct all those purchasing dollars with a comment like, "Just look for the AMD Zen (and beyond) badge and you will be getting a quality product."
This is just gold! Great review Kyle!
 
Kaby Lake was always said to be on Skylake architecture, Intel has made no secret of it. The 14nm+ is designed to deliver better clocks or lower power at the same clocks. IPC was never going to go up.
 
Kaby Lake was always said to be on Skylake architecture, Intel has made no secret of it. The 14nm+ is designed to deliver better clocks or lower power at the same clocks. IPC was never going to go up.
I don't think it is unreasonable to hope for better performance over its predecessor and right now performance is indistinguishable which makes this chips desktop existence befuddling.
 
I don't think it is unreasonable to hope for better performance over its predecessor and right now performance is indistinguishable which makes this chips desktop existence befuddling.

Clocks go higher. Kyle even said OC headroom goes up 200-300MHz. That's higher performance.
 
Wow i read your conclusion and it's spot on. Door is wide open for Zen to be disruptive, not even in terms of outright superiority in IPC or clocks, as I suspect GloFo 14nm is gimped and won't clock much above 4Ghz, but Zen 8Core 16 Thread SMT design with good IPC for a good price is going to be a great competitor.

Reading your link to the 9700 Pro review..

http://www.hardocp.com/article/2002/08/19/ati_radeon_9700_pro_benchmarks/6#.WEsxcPl97AQ

Man, WHAT A DIFFERENCE!

While the Comanche4 and Code Creatures benchmarks are supposedly the most "advanced" I don't really put a tremendous amount of stock in the numbers as both companies have worked hand in hand with NVIDIA during the development of the code. This is not to say they are intentionally tipping the scales their ways but I think that there may be some optimizations for the ATi hardware that have been overlooked. This is just my theory with little or no proof to really base it one....just a hunch.

It's actually quite amazing that NV has been actively gimping ATI for so long. lol

The amazing thing however, is back then, you editors at [H] were still more like regular gamers who had BALLS to call out these sort of anti-competitive behavior. These days, you seem to defend it.

IHVs should NOT be sponsoring game development in a competitive market... though somewhere along the lines, tech journalists no longer mention this conflict of interest.

Nowadays, it's expected that an AMD or NV sponsored title to run worse on the competitor and nobody bats an eye. Quite disappointing that nobody in the tech field is willing to call out PC-hardware-fragmentation and anti-competitive practices like they used to.
 
So it is ever so slightly more efficient... and thats about it. Looks like I will be hanging on to my 6700K another year or two.
Or 5 realistically. Unless you are in a business setting and using the PC for mass amounts of encoding work.
 
The amazing thing however, is back then, you editors at [H] were still more like regular gamers who had BALLS to call out these sort of anti-competitive behavior. These days, you seem to defend it.
I don't defend it. It is simply the way it is. We griped for years and made no difference. At some point you have to move on. I do not see any foreseeable changes in the future either.

Let's get back on topic however and save this discussion for the video card forum.
 
Or 5 realistically. Unless you are in a business setting and using the PC for mass amounts of encoding work.
This i had my 3930k rig for 4 or 5 years and only upgraded when the mb started to die. Its sad how litle cpus have come in the last few years higher ipc but lower clook room and less power thats about it.
At least gpus still need upgrading
 
I appreciate these IPC reviews. They've been some real eye-openers, and this one is no different.

All these years, [H] is still only one of the few sites out there that actually does things like this.

The case reviews are another example e.g., the dedicated coverage to showing potential AIO configurations.

I can only say thank you and please keep up the hard work.
 
Last edited:
Thankyou for the article. Just when I thought to spend some money upgrading because I'm bored and want the latest. I think I'll put my money on some other toys. Since my son left home I don't play games the way I use too. Sure miss those days though.

For the love of god don't get into tools, especially hand tools and planes, saws and chisels. do not do it.
 
Thanks Kyle! - I am still rocking a 2600k, and at this rate I will continue to do so for a while longer. Only issue I'm having is with one of my on-board controllers...nothing a $20. pci card can't fix...I do really want usb 3.1 - but I'm sure it's a luxury I can wait on. Maybe 4.0 will be out by the time cpu's start cooking again ; )
 
Its really simple actually....

Necessity is the Mother of Invention.

There is absolutely no need for better hardware, The gaming industry is the only thing that keeps all this going, and games haven't taken any real steps since Glide & 3Dfx died. Keeping up with the Jones's only makes sense if the Jones's are actually going somewhere.

VR? Please... more horseshit to distract folks from the stagnancy of the Market. They try to bring that bullshit back into the mainstream every couple of years, just like goddamn 3D....its a useless, pointless gimic used to convince you to re-buy everything yet again.

Necessity is the mother of invention that's true. A lot of people seem to gloss over that Intel switched to performance per watt instead of raw performance a long time ago. The desktop market is small enough that it can be served by modified mobile and server processors. Developing something that gives better performance and does so with less energy usage is more difficult than building for raw performance. Even keeping performance in the same place while lowering power consumption isn't easy. By their own admission Intel is getting near the limits of what can be done with silicon. Building for raw performance is easy. If Intel wanted to, it could double the size of Kaby Lake and jack up the clock speeds to bring up performance. Intel could do that all day with relative ease. The market just isn't there for it.

Does it clock any higer atleast?

It does. As Kyle said, its essentially what we were promised two years ago with Devil's Canyon but this time Kaby Lake makes good on that promise. Devil's Canyon barely clocked better than Haswell. Kaby Lake does what its supposed to.


So we keep waiting for games to go full multi core, but one has to really ask would that actually give any tangible benefits. We've had multi core CPU for a long time now and games still rarely use more than two. If the benefits where obvious you would think it would have happened allready.

This is an important point. Games like Lost Planet and engines like Frost Bite are already multi-threaded. Even so, our own testing on Lost Planet back in our 2007 Skulltrail review showed that you only saw the benefits of multi-core CPUs at lower resolutions. At higher resolutions you became GPU bound very quickly. The problem is that certain tasks simply do not benefit from parallelism the way others do. This is why performance doesn't scale in a linear fashion as you add more cores. Some tasks really aren't all that CPU intensive anyway. Its why you can't get linear performance scaling from simply assigning one aspect of the game to an individual core and get somewhere with it. The concept of that is AI one one or two cores, audio processing on another, physics on another etc. Those tasks aren't created equal in terms of their demands.


Kaby Lake was always said to be on Skylake architecture, Intel has made no secret of it. The 14nm+ is designed to deliver better clocks or lower power at the same clocks. IPC was never going to go up.

Yes it is. However, lots of CPUs were built on one architecture or another. Subsequent iterations often had minor improvements which gave us a little better performance over the previous generation. The Pentium vs. Pentium MMX, Pentium II over the Pentium Pro, Core 2 Duo vs. Pentium M, etc. Die shrinks were not always part of the equation and even when they are, performance gains aren't necessarily substantial. Devil's Canyon gave us nothing over Haswell. Intel promised higher clock speeds and stated that the CPU was designed for the enthusiast. Intel never deliverd on this promise. Kaby Lake at least clocks higher and uses less power while doing it. Still, I think many people were hoping that there would be some architectural tweaks as Intel has done those in the past and remained secretive about that until release or close to it.

I don't think it is unreasonable to hope for better performance over its predecessor and right now performance is indistinguishable which makes this chips desktop existence befuddling.

No it isn't unreasonable. Kaby Lake just isn't all that exciting on the desktop. In Laptops its shaping up to be better than anyone expected. You get higher clocks without sacrificing power consumption. In the mobile and server markets, that is potentially huge. On the desktop, we don't give two squirts of piss about that.

Clocks go higher. Kyle even said OC headroom goes up 200-300MHz. That's higher performance.

Indeed.

The real world performance seems negligible though.

When you are primarily GPU bound and IPC is the same, yes. The improvement is negligible. I would advocate going to Kaby Lake if and only if you are wanting to finally ditch an aging Nehalem or Sandy Bridge setup. We are enough performance increments ahead and now have clock speeds we haven't seen since the Sandy and Ivy Bridge days. It won't be a massive upgrade on the CPU side in games, but it will be noticably faster than Sandy Bridge in a number of CPU intensive applications while providing a modern platform with M.2, DDR4 and all the other I/O options that modern motherboards have to offer. I don't think that's insignificant if you care about such things. While the benchmarks aren't earth shattering, I don't think anyone will feel disappointed in going from a Sandy Bridge based system with older drives, slower RAM and an older video card to a completely modern PC.

Thanks Kyle! - I am still rocking a 2600k, and at this rate I will continue to do so for a while longer. Only issue I'm having is with one of my on-board controllers...nothing a $20. pci card can't fix...I do really want usb 3.1 - but I'm sure it's a luxury I can wait on. Maybe 4.0 will be out by the time cpu's start cooking again ; )

And again, the platform has become more of a selling point for upgrading than the CPUs are. It sucks but there it is.
 
While the power draw numbers show an improvement is it possible that once Z170 boards get BIOS updates that Kabylake and Skylake will draw the same total wattage because the small power savings here may actually be due to a more efficient chipset? I'd love to see a quick one paragraph review dropping both chips into a weaker Z170 board not known for good OC or cooling to see how much the platform changes this already pointless equation.
 
While the power draw numbers show an improvement is it possible that once Z170 boards get BIOS updates that Kabylake and Skylake will draw the same total wattage because the small power savings here may actually be due to a more efficient chipset? I'd love to see a quick one paragraph review dropping both chips into a weaker Z170 board not known for good OC or cooling to see how much the platform changes this already pointless equation.

A Z170 motherboard with a BIOS update will not significantly alter its power draw. If it draws less power using Kaby Lake than it does Skylake, that will not change.
 
Still rocking my good `ole x58 LGA1366 platform with a hexacore Gulftown + GTX680. Yes, not very efficient but stable as hell and still decent for 1920x1200 gaming.
Me too but with a Pascal Titan X. Only reason I want to upgade is for M.2 slots.
 
Thanks Kyle! - I am still rocking a 2600k, and at this rate I will continue to do so for a while longer. Only issue I'm having is with one of my on-board controllers...nothing a $20. pci card can't fix...I do really want usb 3.1 - but I'm sure it's a luxury I can wait on. Maybe 4.0 will be out by the time cpu's start cooking again ; )
I liked Kyle's articles when he tested where a cpu stopped and the gpu took over.
 
A Z170 motherboard with a BIOS update will not significantly alter its power draw. If it draws less power using Kaby Lake than it does Skylake, that will not change.
I think I may have stated that backwards then. What I meant was what if kaby and sky draw the same on Z170 and kaby less on z270 because it's not actually drawing less but the motherboard chipset itself is using a few less watts.
 
I think I may have stated that backwards then. What I meant was what if kaby and sky draw the same on Z170 and kaby less on z270 because it's not actually drawing less but the motherboard chipset itself is using a few less watts.
I highly doubt there would be anything significant, if at all.
 
I think I may have stated that backwards then. What I meant was what if kaby and sky draw the same on Z170 and kaby less on z270 because it's not actually drawing less but the motherboard chipset itself is using a few less watts.

That would be hard to compare as the power draw of a motherboard is based on the features integrated into it. Getting an apples to apples comparison there will be difficult at best.
 
I'm confused by all of this vitriol towards Kaby Lake. Did Intel promise IPC improvements? Didn't Intel state explicitly that Kaby Lake was the Optimize in the new Tick, Tock, Optimize pattern? Kaby Lake shows greater promise in the low power, mobile space and I believe that is really what Intel intended all along.
 
I'm confused by all of this vitriol towards Kaby Lake. Did Intel promise IPC improvements? Didn't Intel state explicitly that Kaby Lake was the Optimize in the new Tick, Tock, Optimize pattern? Kaby Lake shows greater promise in the low power, mobile space and I believe that is really what Intel intended all along.
All that is fine, but releasing a new desktop chip that performs nearly identical to a chip released over a year ago and the fact they won't have a kaby lake replacement till at least 2018 means nearly two years and no real upgrade for desktop i7 users
 
Not entirely true. Should see Cannonlake, Skylake-X and Kaby Lake-X next year. Fact is, we still don't know what the OC potential will look like for desktop Kaby Lake.
 
Not entirely true. Should see Cannonlake, Skylake-X and Kaby Lake-X next year. Fact is, we still don't know what the OC potential will look like for desktop Kaby Lake.
We are not going to see 4 separate CPU launches like that.
 
We are not going to see 4 separate CPU launches like that.

Believe that is three. Fact is, at a minimum, we should see Cannonlake and the Basin Falls X platform which would mean Skylake X and Kaby Lake X are potentials. I figure Skylake X is the best bet for 2017 with Kaby Lake X in 2018.
 
I wasn't expecting anything more than SL since it was rumored forever now that they were using the same cores. But still I'm very disappointed they decided to do nothing about the CPU front.

M.2 and other features are bigger incentives to upgrade than the processor is.

Been saying this for a while :(
 
I'm confused by all of this vitriol towards Kaby Lake. Did Intel promise IPC improvements? Didn't Intel state explicitly that Kaby Lake was the Optimize in the new Tick, Tock, Optimize pattern? Kaby Lake shows greater promise in the low power, mobile space and I believe that is really what Intel intended all along.

Intel has abandoned the tick tock strategy.
 
Simply put, I am going to keep my comments to myself till the very end as there is nothing to really say.

That was freaking hysterical. I know I'll move to this platform only to supplant the 4770k I have now. And to that end, hold me off until the next great leap. Though I see a lot of disappointment. This will likely be a good upgrade from where I am. Though, on the 6700k, no chance. No reason. And I don't know that this is Intel throwing in the towel. I do think it is very telling of a shrinking enthusiast market though. We're becoming dinosaurs in our own time. Everybody wants the next phone. And while we get no answer from AMD; it has made Intel stagnate. It is sad. But we are a dying breed. Thanks for letting me know what I'll get. Tears still. :cry:
 
That was freaking hysterical. I know I'll move to this platform only to supplant the 4770k I have now. And to that end, hold me off until the next great leap. Though I see a lot of disappointment. This will likely be a good upgrade from where I am. Though, on the 6700k, no chance. No reason. And I don't know that this is Intel throwing in the towel. I do think it is very telling of a shrinking enthusiast market though. We're becoming dinosaurs in our own time. Everybody wants the next phone. And while we get no answer from AMD; it has made Intel stagnate. It is sad. But we are a dying breed. Thanks for letting me know what I'll get. Tears still. :cry:

No need to despair. From Intel's July conference call:

Sure. If you take a look at it, there were three-ish, maybe four-ish major segments of the PC that did better and continue to do better than the rest of the segment and just overall. Laptops, mobile PCs continue to do better. They did better in the second quarter. 2-in-1 devices specifically are doing very well and continue to grow in double digits. And then as you said, we often call the enthusiast gaming, you see our K SKUs in there, and then you saw us also announce the X SKU, which is our new 10-core system that has been selling much, much better than what even we anticipated. And so yes, gaming and enthusiast continues to grow at a double-digit rate.

http://seekingalpha.com/article/399...-results-earnings-call-transcript?part=single

Intel also said this in its restructuring announcement in April 2016:

Through this comprehensive initiative, the company plans to increase investments in its data center, IoT, memory and connectivity businesses, as well as growing client segments such as 2-in-1s, gaming and home gateways.

Data center and gaming are getting more focus. HEDT chips are derived from server chips.

Things are going to get much better for enthusiasts over the next several years as Intel focuses more on these areas, IMO. Follow the money, and the money says that enthusiasts will get more love from Intel over time.
 
Believe that is three. Fact is, at a minimum, we should see Cannonlake and the Basin Falls X platform which would mean Skylake X and Kaby Lake X are potentials. I figure Skylake X is the best bet for 2017 with Kaby Lake X in 2018.

Kaby Lake-X comes alongside Skylake-X, both are rumoured to launch in August 2017. The 7700K is going to be fine, but the real juicy enthusiast stuff comes a little bit later.
 
Didn't really expect something
Except hoping for a part with edram

I got the lowest i5 skylake on my rig right now and OC'd it to 4.5
Through some BCLK overclocking

I was meddling with the thought of buying a pretested cpu
But kaby lake seems to be a good overclocker
So maybe not need to pay extra for that

mmm
Though I'd be curious how high a pretested could be pushed

Wonder if it's still the useless TIM Intel uses
 
As i kept reading the benchmarks i kept thinking "is this a joke article? It shouldn't be, its not april 1st. Wtf?".

Wow. Just, wow. Not even the token couple of percentages increase from last generation. What, 30w less power drain? Just a sec... (loosens one lightbulb) there, i just saved me 40w of electricity. What a shitload of...

Kaby Lake better be one helluva overclocker if Intel wants to save its face in enthusiast desktop market.
 
As i kept reading the benchmarks i kept thinking "is this a joke article? It shouldn't be, its not april 1st. Wtf?".

Wow. Just, wow. Not even the token couple of percentages increase from last generation. What, 30w less power drain? Just a sec... (loosens one lightbulb) there, i just saved me 40w of electricity. What a shitload of...

Kaby Lake better be one helluva overclocker if Intel wants to save its face in enthusiast desktop market.

IPC have been known to be the same for like a year. So if you expect 2 equally clocked CPUs to perform different you set yourself up for a disappointment long ago.

The 14nm+ process allows for higher clocks and lower power consumption. Out of the box a 7700K is 5% faster in 2-4 threads and 7.5% faster in 1 thread over 6700K. It also seems to overclock better in average with those 200-300Mhz.

the biggest change is in mobile, with higher clocks and higher sustained boost. So you end up with 20-25% performance increases. For server the same pretty much applies. First server part is Skylake-EP tho and Skylake-X for HEDT. 32 cores with a 2.1Ghz base vs Broadwell with 24 cores and 2.2Ghz base.

With Coffee Lake in a year you will see the same IPC again. However you will get 6 cores instead of 4 on LGA1151. If you are looking for IPC increases outside caches, you have to wait till Icelake.
 
IPC have been known to be the same for like a year. So if you expect 2 equally clocked CPUs to perform different you set yourself up for a disappointment long ago.

The 14nm+ process allows for higher clocks and lower power consumption. Out of the box a 7700K is 5% faster in 2-4 threads and 7.5% faster in 1 thread over 6700K. It also seems to overclock better in average with those 200-300Mhz.

the biggest change is in mobile, with higher clocks and higher sustained boost. So you end up with 20-25% performance increases. For server the same pretty much applies. First server part is Skylake-EP tho and Skylake-X for HEDT. 32 cores with a 2.1Ghz base vs Broadwell with 24 cores and 2.2Ghz base.

With Coffee Lake in a year you will see the same IPC again. However you will get 6 cores instead of 4 on LGA1151. If you are looking for IPC increases outside caches, you have to wait till Icelake.

Cannon Lake-X might come before Ice if we're lucky, but yeah, pretty much this post.
 
The CPU market over the past ~8 years had been a complete letdown. Hell if you are running 1080p you can get by with a Sandy Bridge and recent mid-range vidcard without any issue.

Come on AMD. Show me something.
 
Back
Top