FRONTPAGE Intel Skylake i7 6700K CPU Review

[trents]

Ah! So there's what we call the "power phase" components that are under the big heat sink west of the socket and now the FIVR is being moved under there too with Skylake and z1xx. It was confusing to me because on the AMD side we often used the terms "power phase" and "VRM" interchangeably, or so it seemed to me.

 

[EarthDog]

Well, its not called a FIVR when its in the VRM. There isn't a FIVR (rember what a FIVR is Fully INTEGRATED Voltage Regulator - its not integrated). The terms 'power phases' and 'VRM' are still interchangeable. Nothing changes there.

 

[trents]

What I meant was the function of what was called FIVR is being moved under that big heat sink with the rest of the power phase components.

 

[Brando]

i was looking forward to skylake. i guess 10% a year is the new moores law. bummer. i guess if nothing else the better igpu may help more with dx12 and free anti aliasing if that whole thing pans out. still a decent upgrade for me and my 3570k but it's hard to see upgrading from a haswell. so skylake should be about 30% better clock/clock over an ivy bridge or am i off on this? if so a 4.8ghz 6700k would be pretty sweet over my 3570k at 4.3ghz.

 

[wingman99]

Good review, skylake is a good upgrade if you have a 3 year old PC.

 

[Woomack]

i was looking forward to skylake. i guess 10% a year is the new moores law. bummer. i guess if nothing else the better igpu may help more with dx12 and free anti aliasing if that whole thing pans out. still a decent upgrade for me and my 3570k but it's hard to see upgrading from a haswell. so skylake should be about 30% better clock/clock over an ivy bridge or am i off on this? if so a 4.8ghz 6700k would be pretty sweet over my 3570k at 4.3ghz.

I don't think you will be able to overclock average 6700K to 4.8GHz fully stable at reasonable voltages. We will see soon but most chips I've seen so far could make something like 4.7 1.35-1.45V and 4.8GHz+ for quick benchmarks. My 6600K is not passing any more demanding tests @4.9GHz regardless of voltage and I'm testing on pretty good water cooling. 4.8GHz is max for tests like HyperPi or Cinebench and it's sometimes crashing.
Anyway it seems like a good upgrade from IB , especially that DDR4 price is not much higher than DDR3 and 8GB RAM is still more than enough for most games.
After some tests I can say that Skylake reacts better on high memory frequency than low timings. It's also overclocking more like Haswell-E than Haswell ( long story, maybe for some other thread ).

 

[Brando]

now i'm reading that skylake is inferior to the previous generation for gaming. can this really be true? there must be a problem with software or something.

 

[ATMINSIDE]

now i'm reading that skylake is inferior to the previous generation for gaming. can this really be true? there must be a problem with software or something.

Did you read this review at all?

 

[Lvcoyote]

Everyone's system is going to react different, but our testing had Skylake a little ahead of Haswell.

 

[Janus67]

now i'm reading that skylake is inferior to the previous generation for gaming. can this really be true? there must be a problem with software or something.

It would help if you find information counter to what most other reviews are saying to post a link to said review (hopefully it isn't some random 16 yr old youtuber).

 

[ChanceCoats123]

I don't think you will be able to overclock average 6700K to 4.8GHz fully stable at reasonable voltages. We will see soon but most chips I've seen so far could make something like 4.7 1.35-1.45V and 4.8GHz+ for quick benchmarks. My 6600K is not passing any more demanding tests @4.9GHz regardless of voltage and I'm testing on pretty good water cooling. 4.8GHz is max for tests like HyperPi or Cinebench and it's sometimes crashing.
Anyway it seems like a good upgrade from IB , especially that DDR4 price is not much higher than DDR3 and 8GB RAM is still more than enough for most games.
After some tests I can say that Skylake reacts better on high memory frequency than low timings. It's also overclocking more like Haswell-E than Haswell ( long story, maybe for some other thread ).

Have you put your chip under cold yet? My 3770k died this past week and I'll need something new to bench with. If Skylake is clocking decently well on the cold stuff, it could be a good contender to replace what I lost.

 

[Brando]

It would help if you find information counter to what most other reviews are saying to post a link to said review (hopefully it isn't some random 16 yr old youtuber).

"In our discrete gaming benchmarks, at 3GHz Skylake actually performs worse than Haswell at an equivalent clockspeed, giving up an average of 1.3% performance. We don’t have much from Intel as to analyze the architecture to see why this happens, and it is pretty arguable that it is noticeable, but it is there."
http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/23

EDIT: a couple of places are showing better ram making a significant difference. that was unexpected. i wonder where the performance will level out with ram speed and how much you need to spend to get all of what skylake can give? hopefully ddr4 gets cheap soon.
http://www.hardocp.com/article/2015...76700k_ipc_overclocking_review/6#.VcgwnvlezRI

 

[Brando]

not too bad i guess
http://www.newegg.com/Product/Product.aspx?Item=N82E16820231902

 

[Woomack]

Have you put your chip under cold yet? My 3770k died this past week and I'll need something new to bench with. If Skylake is clocking decently well on the cold stuff, it could be a good contender to replace what I lost.

Not yet. I'm trying to find what memory will be best for these chips and maybe I will get something new soon. Right now I have only Hynix and Micron based kits as I decided to return Samsung based G.Skill 3200 ( it was barely working at stock ).
I can only say that Hynix is pretty good. I'm running 3000 C15 kit at 3333 14-16-16 and test results are fine. Could be better but I'm not complaining. I was checking it at 3466 too but memory had stability issues and more relaxed timings were giving worse performance.

My 6600K has 1.344V stock voltage and hard to say how it will overclock on cold. I will probably check it next weekend on ss or maybe I get dice. Right now we have 30*C+ ambient temp so it's not good for benching.

Since CPU and cache voltage is tied then I'm wondering which one will die first because of too high voltage Really I have enough RMA already

 

[Robert17]

Saw this regarding the question of TIM vs solder, a delid of a 6700k changing the stock TIM for better. And the results.

 

[ATMINSIDE]

"In our discrete gaming benchmarks, at 3GHz Skylake actually performs worse than Haswell at an equivalent clockspeed, giving up an average of 1.3% performance. We don’t have much from Intel as to analyze the architecture to see why this happens, and it is pretty arguable that it is noticeable, but it is there."
http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/23

EDIT: a couple of places are showing better ram making a significant difference. that was unexpected. i wonder where the performance will level out with ram speed and how much you need to spend to get all of what skylake can give? hopefully ddr4 gets cheap soon.
http://www.hardocp.com/article/2015...76700k_ipc_overclocking_review/6#.VcgwnvlezRI

You're talking about @3GHz... On a 4GHz chip...

 

[Janus67]

You're talking about @3GHz... On a 4GHz chip...

That's true, but in the test all CPUs were downclocked to 3GHz, I assume for a common baseline that doesn't require overclocking to reach other speeds.


My problem is they were using some garbage DDR4-2133 memory in their tests with it. If they had used DDR4-3000(ish) I think we would have seen at least a tie or a few % better. But I can understand that there are issues with coming up with a DDR4 constant kit at this early point in its life cycle. Otherwise you have people complain that they used RAM that is far and beyond what most people would buy (as people complained with Dino's 6700k review with the DDR4-3200 [IIRC] memory).

 

[ATMINSIDE]

That's true, but in the test all CPUs were downclocked to 3GHz, I assume for a common baseline that doesn't require overclocking to reach other speeds.

No mention of cache or RAM speeds on any of the older gens though, unless I'm missing something.

 

[Janus67]

http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/6

says they are using 1866 ddr3

 

[ATMINSIDE]

http://www.anandtech.com/show/9483/intel-skylake-review-6700k-6600k-ddr4-ddr3-ipc-6th-generation/6

says they are using 1866 ddr3

Also goes on to say:
*Memory Timings used were the supported frequencies of each architecture, except DDR3L vs DDR4 testing, which used DDR3-1866 C9.

This to me reads that the 1866 C9 was used for DDR3L testing, but only saying they used what was supported on each generation.

Still no mention of cache speeds (since each of the four chips was underclocked, this makes a difference)