Intel Kaby Lake Core i7-7700K Sample Tested

Not even 5% performance increase?? Is this legit?

 

Close to no difference...

 

Holy guacamole, dat performance increase! Looks like I won't upgrade my CPU anytime soon.

 

There is one - the price of the new units will be 100-150 EUR higher than the predecessors :banana:

 

So Intel now releases CPUs with no difference at all to the previous gen. What's next? A cpu which has lower performance from the previous gen?

 

Actually if you look again at the numbers, they show the 7700K is slower than the previous 6700K clock for clock!

They score almost identically in every test but the 7700k is clocked higher.

 

Intel have done that before when they released the first Pentium 4s'.

Despite Intel calling these 7th generation, they are absolutely NOT a new generation.

Realistically Sandy Bridge (classed as 2nd gen) and Ivy Bridge (3rd gen) are really first gen Core processors, since the original Core processors were pretty much modified Pentium 4s'. Haswell was really the 2nd gen, Broadwell (fail) was 3rd gen, Skylake/Kaby Lake/Coffee Lake are all 4th gen, Cannonlake/Ice Lake are 5th gen. Despite this, Intel will still progress through them as 'generations' even though that is marketing and not factual.

Original lineup:
Haswell --> Broadwell --> Skylake --> Cannonlake

First revision when Broadwell was DELAYED: So failure 1.
Haswell --> Haswell Refresh (/Broadwell, limited release) --> Skylake --> Cannonlake

Second revision:
Haswell --> Haswell Refresh (/Broadwell, limited release) --> Skylake --> Kaby Lake --> Cannonlake

Third Revision (current):
Haswell --> Haswell Refresh (/Broadwell, limited release) --> Skylake --> Kaby Lake --> Coffee Lake (/Cannonlake for low powered devices only due to ongoing issues) --> Ice Lake.

Kaby Lake is therefore originally an unintended release, it's existence is due to the failure of Cannonlake. It is therefore just a tweaked Skylake, it would be more true to call it a Skylake refresh than Kaby Lake. The new chipset along with it is inconsequential, Haswell Refresh got a new chipset as well. Coffee Lake will probably be much the same performance wise, although I do suspect it will have a little more gain from tweaking, and of course there's the 6-core option.

In any case, probably the first desirable upgrade for people insistent on Intel and have currently got a decentish rig will be Ice Lake, not expected until the end of 2018. Ice Lake will therefore have to go up against Zen's first successor! the Zen+.

Intel are basically going through what AMD went through, except they are in a far, far better position to begin with to weather the ongoing failures and mediocre improvements, at least for now when there is no competition.

Intel have great power, people still overlook Cannonlake's failure by the promise of a 6-core Coffee Lake. The 6-core Coffee Lake will probably be similar in performance to Kaby Lake in terms of IPC, possibly even lower as I suspect it may not overclock as well as the 4-core version due to heat. If it does, it means they have to cherry pick the cores, which will mean it will be unattractively more expensive than the 4-core. It will also mean the Extreme series will have to get a major upgrade, since why get a costly Extreme processor when a 6-core mainstream processor performs much the same as Extreme processors that most people seem to go for (the 6-core one)? I guess they will drop the 6-core from the Extreme processors, but that will just mean anyone wanting one will have to pay a lot more for the 8+ core variants.

The best thing an Intel fanboi could want, regardless of the fact they'd totally disagree, is for AMD to do really well with Zen. Fear of competition doesn't mean much, things will only change with actual competition. Even if Intel hurts a bit for the next couple of years, it will be more likely that Ice Lake will actually be better (through improved R&D) if they face stiff competition from AMD.

 

I was thinking about this earlier. I bet they use TIM and not solder which will make the 6-core unviable for overclockers without a de-lid.

 

Has it been tested with 7?

 

How come every time Intel releases something I got the feeling that just going for Haswell-E hexacore was pricey but still the right decision?
Not that much of an emotionally moving release imho.

 

Anyone really wondering how much will 7700k overclock when its already boosting to 4.5ghz?

This article said it can reach 5Ghz on 1.49v (burning!). How owesome is that?

http://www.pcgameshardware.de/Core-...hmarks-mit-4-5-und-5-GHz-aufgetaucht-1214658/

 

An indication of the performance of Ice Lake will be known from the low powered versions of Cannonlake, you could simply scale the clocks. It is only the inability for it to clock as high as current CPU's that is stopping it being released. I'm not sure what the low powered variants will run at, could even be half mainstream speed (for example, 2.1 GHz). However that said, not much is known about Ice Lake. It is supposedly on the same process as Cannonlake, but could possibly be different enough to be actually consider a proper new processor. This is more of a possibility since Zen looks to be such a good performer (remember that Ice Lake will have to go up against Zen+, which has been mentioned to have at least 10 percent IPC gain), Intel do have about two years before Ice Lake's release to bring out something special.

Intel probably couldn't 'afford' in a sense, for Ice Lake not to perform well against Zen+, particular since Zen+'s successor and it's successor supposedly have at least a 10 percent IPC gain as well. This is especially true when you consider that Ice Lake's successor will be likely a tweaked Ice Lake, and not a new architecture.

Not sure how AMD will manage a 10 percent gain per generation. Apparently Zen+'s successor will be supposedly on a new process, it will be interesting to see what that could bring to the table when it is released (at least thus far), circa end 2019 on the AM4+ socket.

 

That voltage is too high for realistic ongoing overclocking. Remember boost clock isn't 'permanent', so even manually setting it to 4.5 GHz would give some gain. Most likely 4.6-4.7 GHz would be the main overclock. Ideally you want the highest overclock without diminishing returns on CPU voltage and heat production. For instance, if 4.7 GHZ runs on a nice low voltage and cool, 5 Ghz would make little sense since at best it would only be 6 percent faster, and that's without consideration of thermal throttling etc.

This is especially true for 4.8 GHz over 4.7 GHz, this is at best 2 percent. If you require more voltage than you did from 4.6 to 4.7, and the temperatures run higher, apart from bragging rights you could argue the net gain is zero.

 

If you translate the coresponding paragraph from the PCgameshardware article, it reads like this (google translation for the sake of objectiveness):

http://translate.google.com/transla...214658/&hl=de&langpair=auto|en&tbb=1&ie=UTF-8



They are saying that actually, with little voltage you can get far, BUT the CPU reaches about 100°C. This are temperatures I haven't seen mentioned in the past 10 years... not sure even proper watercooling could cope with 100°C that easy. A bit scary tbh

 

The only improvement i can see is the Igp...who uses Igp, quess my 4790k will stay for a while longer, maybe 6700k prices will drop enough for a worthy system upgrade.

 

As I see it. Only reason to buy this is if you are building a system from scratch.

As for a upgrade, I will wait and compare if Zen can deliver a better product in the same price bracket. I use a lot multi core applications, so more cores would be awesome. Though I don't believe Intel will deliver in that department in the near future.

 

From Wiki:

source: https://en.wikipedia.org/wiki/Kaby_Lake

So yeah, nothing fancy. Just optimizations and some iGPU boost.

 

it will have support for Netflix 4K and more pwerfull iGPU. Definitely worth extra 100 eur

 

And what about the dimensions of the physical package without the HS? If the actual CPU is smaller it will be even harder to cool and 1.3v might equal 1.35v on Skylake. At those voltages you need to have decent cooling otherwise they're toast. You need to move the heat away as quickly as possible and without much surface that isn't going to happen!