First "FULL" review of Skylake

[Flake87]

http://wccftech.com/intel-skylake-core-i7-6700k-cpu-review-processor-gaming-performance/

Published at around 9:00pm BST

 

[opethdisciple]

Don't get it. Are those screenshots at stock?

Core voltage is very high on Skylake?

 

[therealjustin]

The gaming results are very strange indeed and I don't see how the slightly higher turbo boost on the 4790K would allow it to beat the 6700K like it is.


The two Skylake-K chips are for gamers and enthusiasts but it seems like the new architecture is a failure(in terms of any advancement in performance) in those areas.

 

[Martini1991]

I'll wait till we get proper sites.
But on the "gaming benchmarks" if you look at the 3.6GHZ 4790K, it's beating Skylake at the same clock.

Since that's almost certainly some type of mistake. Or Intel have released a fail (Unexciting perhaps, but there's little chance it's a performance failure)

Although, I have no hypes anyway for Skylake lol.

 

[muon]

There are other differences. The chipset and memory. Either of those could be causing it, especially if the DDR4 memory has poor latency.

 

[Joe!]

I'd like to compare this to 5820k, but I'm sure we'll see that tomorrow.

I'm edging slightly towards the i7-5820k, seems to be a lot better value for what it is.

 

[Flake87]

At the bottom there's a discussion about that in the comments section, they suggest a couple of factors.

"But games wont see a difference since they dont use newer instruction sets plus they are mostly gpu limited."

and

"it looks like there is the need for some driver refinement. In benchmarks where the CPU is the main contributing factor we clearly see Skylake in the lead over DC even with lower clockspeeds. However, in the games the final results obviously are not merely based on raw CPU performance but on the interaction between CPU and GPU where there's still the need to get everything tuned properly."

There appears to be some concensus that the games can't use the latest instruction sets, and are not so much CPU dependant as GPU. Also that the driver set up between CPU Motherboard, and GPU are not yet fully optimised.

 

[muon]

Well it shouldn't necessarily require new instruction sets for games to perform better. At the very least it should be at least as good as Haswell.

Also what new instruction sets does Skylake have?

 

[Dave2150]

Seems we should take these results with a pinch of salt, since the DDR4 used for the 6700k was using very crappy timings, while the DDR3 for the 4790k was using very tight timings.

Waiting for the Anandtech review tomorrow to paint the true picture.

 

[Boomstick777]

4790K still best for gaming then, as a few of us thought, little performance increase. Behind in some things.

The Skylake refresh (Kaby Lake) next year should bring the 6790K, that would be a better 4790K replacement, although imho X99 is still the best way to go if your spending that kind of money, unless your specifically building a low power small form factor PC. In that scenario Skylake will be ideal.

Skylake i5 + cheap mobo will be new gaming sweet spot, X99 for high end stuff.

 

[muon]

Kaby Lake would be the 7xxx chips?

edit:

or is that cannonlake?

 

[Boomstick777]

Kaby Lake would be the 7xxx chips?

edit:

or is that cannonlake?

Yeah Kaby Lake is the refresh of Skylake before Cannonlake arrives (Cannonlake is releasing later than first planned).

Like Devil's Canyon was to Haswell 4770K > 4790K..

Expect the refresh chip will be called 6790K and come with much higher stock clocks, mid way through next year.

 

[Orangey]

I think a reality check is needed here, how many red flags can you brush off before you admit Skylake is tot. CPUs don't need drivers and timings don't mean jack.

Higher voltage, performing worse than 4790K even at same clocks, higher TDP, time to give up.

 

[Datass]

Damn, no overclocking results or even temps.

Those voltages though, the 5775c reviews picked up on the 1.2v on broadwell, seems 1.3 is actually quite likely especially given the 95w power draw. OC on this isn;t going to be fun.

Welp, guess i'll stick with my 4790k

 

[Will Lucky]

1.3v at stock. Christ if this is true this has flopped on its face.

 

[Orangey]

1.3v at stock. Christ if this is true this has flopped on its face.

Pretty clear Intel is being pigheaded about Moore's Law for PR purposes and pushing out new nodes before they're ready. Not that they would've offered any uplift anyway, but they'd sip less power at least.

 

[Joe!]

So, as an overclocking noob... I hear anything above 1.3v on the current set of processors will degrade the life of the chip if it's not properly cooled? Or will it degrade regardless?

Does this apply to Skylake, even though it's a new chip?

I'm entirely new so my apologies for the blind ignorance.

 

[muon]

So, as an overclocking noob... I hear anything above 1.3v on the current set of processors will degrade the life of the chip if it's not properly cooled? Or will it degrade regardless?

Does this apply to Skylake, even though it's a new chip?

I'm entirely new so my apologies for the blind ignorance.

In theory it should be even worse for 14nm.

http://www.anandtech.com/show/2468/6

When 45nm chips came out, people were claiming it was happening.

I don't know if Intel have subsequently done something to take care of the issue. They must have if 1.3v is acceptable for 14nm chips.

 

[Joe!]

I don't know if Intel have subsequently done something to take care of the issue. They must have if 1.3v is acceptable for 14nm chips.

This is the line of thought I had. I can't imagine them selling Skylake if 1.3v was detrimental to the processor's life.

 

[Orangey]

1.4 will degrade a Sandy 32nm after only a year, so they must have done something. Very stringent QA on CPUs, not like GPUs.