vacancies advertise contact news tip The Vault
facebook rss twitter

Intel Core i7-7700K Kaby Lake gets benchmarked and OCed

by Mark Tyson on 30 November 2016, 09:31

Tags: Intel (NASDAQ:INTC)

Quick Link:

Add to My Vault: x

The review editor over at Tom's Hardware got a nice surprise when an early sample of an Intel Core i7-7700K Kaby Lake processor landed unexpectedly on his desk. The opportunity was taken to give this chip a pretty thorough benchmark and comparison testing session against the chip it will replace, the Intel Core i7-6700K Skylake processor. It is noted that Intel declined to comment upon Tom's Hardware's findings. The Core i7-7700K sample isn't marked as an engineering sample, yet we can't be sure it is a retail sample either.

Intel is promising 300-400MHz higher clocks with its 14nm+ process, facilitated by transistor fin optimisation. The sample Intel Core i7-7700K Kaby Lake tested had base/boost frequencies of 4.2GHz/4.5GHz respectively. The predecessor i7-6700K offered a default base/boost of 4.0/4.2GHz. However the Kaby Lake chip is rated at 95W TDP compared to the 91W Skylake part. Below you will find power consumption and overclocking charts from Tom's Hardware tests.


On to the benchmarks… Editor Thomas Soderstrom went through various synthetic benchmarks, 3D gaming benchmarks and timed mixed task benchmarks to check out the bottom line for the King of Kaby Lake - if it shows worthwhile performance gains. Flicking through the various results I saw slight gains on most benchmarks. You can see the whole lot of them on page 2 of the review here. I've chosen just the Adobe Creative Cloud 2.0 timed test to reproduce below, as a typical and useful real-world reference.

The review concluded with temperature and efficiency measurements. This was a particularly disappointing area of the review for the Kaby Lake Core i7-7700K, with the Skylake chip running much cooler when put under duress. Further damning results came from efficiency comparisons where the out of the box fresh i7-7700K used up to 28 per cent more power for an average 3.6 per cent performance gain.


The test system used a Gigabyte’s GA-Z170X-Ultra motherboard with the latest Kaby Lake supporting BIOS. G.Skill DDR4-3600 modules were installed and the air cooling of the CPU was taken care of by a Noctua NH-U12S (testing a beefier NH-D14 saved a mere 3°C on the Kaby Lake temps).

At the end of the day these are early numbers, albeit from a trusted source. The Z270 platform and newer BIOS updates for 100-Series motherboards of various brands could make some telling difference to Kaby Lake benchmark figures.

HEXUS Forums :: 8 Comments

Login with Forum Account

Don't have an account? Register today!
something funny I am interested in Zen! The wow factor with intel dissipated when AMD showed off their next gen CPU
Whew, at least I know I won't face the temptation to upgrade from my 6700k ;-) Re temperatures it's usually also an IHS lottery there - my Skylake was too hot by default until I delidded. Shaved off about 15C afterwards.
Would have been nice to see it compared to some other cpu's other than the last gen… honestly though it hardly seems like a worthwhile upgrade for most people with anything recent from intel… like usual…

As lumireleon says I'm more interested how Zen performs and how much it costs because if they come in with their 8C/16T chips around the price of the 4C/8T intels that ‘could’ be a game changer…
It seems Intel really are trying to take the wind of the Zen sails, anybody would think they are worried…

Overall this should be a decent upgrade over my 3570k so games on, lets see how Zen compares.
Not impressed but it is early yet. That said I don't think I'll say much different when it hits as Intel seems to give as little as possible each gen while concentrating on gpu side which I turn off currently. I'll be waiting on Zen and vega before doing anything (1080ti also…of course). Bring on some ZEN benchmarks.

I really hope someone does some handbrake quality tests (for kaby and zen). Is it safe to turn gpu on now for ripping, or does it still sacrifice quality for speed? I'd hope a few revs later they've fixed the quality part. No point in ripping something IMHO if it looks like crap. Anandtech used to do these tests, but haven't in many generations. For now though (most stuff anyway), I'm fine waiting everything out with my 4790k. I'm curious to see how zen comes out on the ripping quality, and am hopeful if neither of these works out, handbrake gets it's act together and fully supports Cuda at some point. Nvidia, why are you not writing code to help handbrake? Can't be that time consuming to help a little for a top used ripping app available to everyone free.