Intel Core i7-7700K Retest: More Oveclocking, Less Heat

[Crashman]

Our saga to find the best Z170 motherboard for Kaby Lake continues, as we grow impatient for Z270 replacements.

Intel Core i7-7700K Retest: More Oveclocking, Less Heat : Read more

 

[ssdpro]

I guess someone didn't like the initial findings. Was it the motherboard mfg or Intel themselves? Or both? Blanket statements like "That makes this the perfect CPU to buy..." smell of marketing influence.

 

[anbello262]

I guess someone didn't like the initial findings. Was it the motherboard mfg or Intel themselves? Or both? Blanket statements like "That makes this the perfect CPU to buy..." smell of marketing influence.

It was the community. Have you read the comments on the previous article? A lot of people were saying that this MOBO had known firmware issues, and many asked for a retest.

And that final statement emphasises "instead". Meaning "if you were going to buy Skylake, you should buy KabyLake instead".

 

[rantoc]

Thx for the preview, will be more interested to see what the retail ones will be averaging. If the OC potential without exotic cooling averages in retail is good the product may have some added benefits beside the pretty much non-existent other ones.

I hope AMD's Zen will deliver because intel sure as hell isn't going to without a kick in the nuts.

 

[xFeaRDom]

I'm ready for people to complain about something this time. :loL:

If it's something that's affecting you so much, go and train, become an engineer/scientist and find something that will allow Moore's law to speed back up instead of decrease again.

 

[Brian_R170]

If a person were in a position to choose either a 6700K or a 7700K, then it's unlikely that they already have a Z170 board, and given that the 6700K and 7700K are priced identically, buyers should really wait and see the performance and price of 200-series boards before deciding which is "the perfect CPU to buy".

 

[knowom]

Wow entirely unimpressive congrats Intel you've managed to produce virtually the same CPU how tempting no thanks...I'm feeling a little Zen right now.

 

[xFeaRDom]

Wow entirely unimpressive congrats Intel you've managed to produce virtually the same CPU how tempting no thanks...I'm feeling a little Zen right now.

It's not like Intel can improve their CPUs much, only thing is to bring the 6 cores down to the current 4 cores price range, which wont happen.

Just because AMD has a lot of headroom to improve their CPUs, such as much stronger cores etc. Thats why Zen seems to be hyped up so much.

The last few releases of CPUs haven't had much improvement generally, probably 5-10%, maybe a little extra overclocking room, more efficient, it's not going to go from what it currently is to being able to create a supercomputer.

 

[ComputerSecurityGuy]

It is looking like the i7-7700K is basically a better binned slightly tweaked i7-6700K.

No shock there. I would like to see a comparison of the i7-6700K@4.2gHz versus the i7-7700K at stock. Same configuration. I want to see, exactly, whether there is any IPC change.

 

[Nintendork]

Skylake and KL are the same damn chip (not even a new stepping, remember the C2D or Phenom II with new stepping improvements?), better yields for intel. even the Haswell Refresh decrease a bit of the voltages.

 

[JamesSneed]

Wow entirely unimpressive congrats Intel you've managed to produce virtually the same CPU how tempting no thanks...I'm feeling a little Zen right now.

It's not like Intel can improve their CPUs much, only thing is to bring the 6 cores down to the current 4 cores price range, which wont happen.

Just because AMD has a lot of headroom to improve their CPUs, such as much stronger cores etc. Thats why Zen seems to be hyped up so much.

The last few releases of CPUs haven't had much improvement generally, probably 5-10%, maybe a little extra overclocking room, more efficient, it's not going to go from what it currently is to being able to create a supercomputer.

Its pretty obvious Intel shifted to focus on built in GPU performance instead of CPU performance over the last few iterations. They can do this since AMD is so far behind for the last decade. From a consumer perspective this has sucked but I don't blame Intel as they make the same amount of money and don't get deemed a monopoly. I think many are excited for Zen because its the first chance in a long time AMD will be competitive once again and with that we all hope the pace of change moves faster once again.

 

[xFeaRDom]

Wow entirely unimpressive congrats Intel you've managed to produce virtually the same CPU how tempting no thanks...I'm feeling a little Zen right now.

It's not like Intel can improve their CPUs much, only thing is to bring the 6 cores down to the current 4 cores price range, which wont happen.

Just because AMD has a lot of headroom to improve their CPUs, such as much stronger cores etc. Thats why Zen seems to be hyped up so much.

The last few releases of CPUs haven't had much improvement generally, probably 5-10%, maybe a little extra overclocking room, more efficient, it's not going to go from what it currently is to being able to create a supercomputer.

Its pretty obvious Intel shifted to focus on built in GPU performance instead of CPU performance over the last few iterations. They can do this since AMD is so far behind for the last decade. From a consumer perspective this has sucked but I don't blame Intel as they make the same amount of money and don't get deemed a monopoly. I think many are excited for Zen because its the first chance in a long time AMD will be competitive once again and with that we all hope the pace of change moves faster once again.

Agreeing with the AMD part, although I don't like AMD at the moment, I think they do have a chance with Zen if they live up to their expectations. Although their CPUs are based on lower end of the market, and creeping up across the whole board to the high-end. If they actually make competition for Intel on this release, they will actually force Intel to create something that will put them much ahead on the market. I've held off of upgrading my CPU due to this, if Zen brings the core power, and efficiency to match with Intel, then hopefully it'll be more of a battle if you know what I mean.

At the current standards, Intel can just stand there and let AMD swing at them, because they're much ahead in the CPU game at the moment. So i'm hoping AMD can take that huge uppercut and match with Intel but at a lower, or competitive price range.

 

[joshjaks]

At this point, it's a matter of waiting for Zen, seeing what makes sense in the price/performance ratio, and then deciding on upgrading from my aging 8350 system. Not enough performance boost to justify a rush purchase imho.

 

[TJ Hooker]

Skylake and KL are the same damn chip (not even a new stepping, remember the C2D or Phenom II with new stepping improvements?), better yields for intel. even the Haswell Refresh decrease a bit of the voltages.

No, there should be some tweaks in fin structure and interconnects, if I recall correctly. It's still the same micro architecture, and the same process node, but I wouldn't go as far to say they're the same chip. Too early to definitively say how much difference those tweaks made though.

 

[JackNaylorPE]

I guess someone didn't like the initial findings. Was it the motherboard mfg or Intel themselves? Or both? Blanket statements like "That makes this the perfect CPU to buy..." smell of marketing influence.

Wow entirely unimpressive congrats Intel you've managed to produce virtually the same CPU how tempting no thanks...I'm feeling a little Zen right now.

What's different here ? We haven't seen a marked improvement in quite a while. Since Sandy Bridge biggest jump was for Devil's Canyon.

http://www.anandtech.com/bench/CPU/39

Sysmark 2012
2600k - 212
3770k - 228 (+7.5%)
4770k - 252 (+10.5%)

Sysmark 2014
4770k - 1970
4790k - 2217 (12.5%)
6700k - 2352 (+6.0%)

From the gaming standpoint, which would cover most of the audience here:

GTAV w/ GTX 980
2600k - 69.21
3770k - 71.28 (+2.99%)
4770k - 73.82 (+1.04%)
4790k - 73.77 (< 0%)
6700k - 74.09 (< 0%)

That's less than a 5% increase over 5 generations

Three things to remember :

1. Other than bragging rights, synthetics bring nothing to the table; it's important to focus on things that you do every day when making purchase decisions.

2. Unless you are using workstation type apps (Handbrake is one app that typically benefits from a new CPU generation) , next generation CPUs haven't been anything to get excited about.

3. This is an early stepping CPU on a prototype MoBo. As production lines are tweaked ...

a) Max CPU overclocks won't increase but the number of CPUs that hit a given multiplier will.
b) Later steppings MoBos will result in firmware improvements, elimination of bugs, etc each of which will bring additional performance increases.

My guess is we'll have Anadtech's review up shortly and we can add the 7700k to the mix

 

[Daniel Ladishew]

Thank you for testing on another Mobo. I'm looking forward to coverage of this CPU on the new chipset.

 

[LOLROFL]

Tomshardware has sunk to a new low.

 

[envy14tpe]

So what did we learn? Don't use a jump to conclusions mat.

 

[RedJaron]

I guess someone didn't like the initial findings.

A mark of a good scientist or QA engineer ( of which a hardware reviewer is one ), is the ability to question results. A 50W increase in power consumption for a CPU that's only supposed to have a 5W higher TDP should set anyone's spidey-sense to tingling.

Was it the motherboard mfg or Intel themselves?

Are you not familiar with the basic scientific method? Run an experiment with one set of variables. Record the result. Change one variable and re-run the experiment. Record and compare those results with the first set. If the results are the same, then the variable had no effect on the result.

Since the mboard was the only thing changed out, it makes sense to assume the problem lies there. It is possible that the GB board had some kind of hardware problem, but since it behaved normally during its review with a Skylake CPU, that is unlikely. Considering the initial findings show fluctuations after some manual intervention with the UEFI settings, the most likely culprit was GB's firmware.

 

[Crashman]

Tomshardware has sunk to a new low.

LOL, methinks your comment might have been a bit different were it a Zen CPU.

Don't blame anyone but AMD for the lack of available preview samples.

 

[ammaross]

Wow entirely unimpressive congrats Intel you've managed to produce virtually the same CPU how tempting no thanks...I'm feeling a little Zen right now.

It's not like Intel can improve their CPUs much, only thing is to bring the 6 cores down to the current 4 cores price range, which wont happen.

Cannonlake is already slated to have a 6-core i7 in mainstream, so you'll see it soon.

 

[ah]

It goes without saying, if u r going to buy a new chip, u would buy this chip instead of its predecessor. Also, I don't think u can expect the new chip to outperform its predecessor by much more, because that the constraint of the technology and manufacturing process.

 

[Brian_R170]

Skylake and KL are the same damn chip (not even a new stepping

Besides the process improvements, I'm pretty sure they had to step it to make the changes in graphics to support 10-bit HEVC.

 

[JackNaylorPE]

Steppings that result in significant performance improvments are quite rare. In this millenium steppings have typically fixed bugs / errata or made efficiency improvements.

I always took it that's why, while we don't see any change in maximum overclocks as a CPU generation matures, we do see a significant jump in the number of CPUs that manage a specific OC. It's not that hard to pick up evidence of new MoBo steppings but Intel is kinda tight lipped on the subject. I would be surprised if Intel managed to go thru a complete generation cycle without a single improvement being made. Form wiki:

"Typically, when an integrated circuit manufacturer such as Intel or AMD invests money to do a stepping (i.e. a revision to the masks), they have found bugs in the logic, have made improvements to the design that allow for faster processing, have found a way to increase yield or improve the "bin splits" (i.e. create faster transistors and hence faster CPUs), improve maneuverability to more easily identify marginal circuits, or reduce test time which can reduce the cost of testing the part.

 

[bit_user]

More than anything, I think this highlights the influence a mobo can have.

I've grown pretty complacent about mobo performance and typically shop based on features. I guess I'll have to start paying more attention to the benchies, from now on.