Intel Skylake / Kaby Lake

[LTC8K6]

http://akiba-pc.watch.impress.co.jp/docs/news/news/20150813_716197.html

Official LGA1151 Factory cooler.

Looks nice. Why didn't they offer this for the 4790K?

https://www.ple.com.au/ViewItem.aspx?InventoryItemId=620292

http://www.mwave.com.au/product/intel-ts15a-cpu-cooler-ab65544

 

[LTC8K6]

http://www.benchmark.pl/aktualnosci...nie-dla-odblokowanych-procesorow-skylake.html

More on that cooler.

 

[Sweepr]

So, if the 'author' hasn't been paid off by Intel to make the Skylake seem as if it has high IPC, like the author you quoted, then it doesn't count, huh? I'll let you know when the things you say in this thread start to surprise me.

Right. Eurogamer, Hardware.fr, Hardware Canucks, PurePC, PCLab, Guru3D, etc. All payed by Intel. Crippling Skylake with DDR4 2133 CL15 is the ultimate way to test it, if you want to feel good about your outdated Intel chip.

 

[ShintaiDK]

Very nice with the TS15A. A cheap 130W cooler. 24€ without VAT it seems.

But I would like to see a review of the noise.

 

[AtenRa]

http://www.eurogamer.net/articles/digitalfoundry-2015-intel-skylake-core-i7-6700k-review

Except for GTAV and FC4, at 4.4GHz OC for all the CPUs with an OC Titan X the performance differences are negligible. If you have a 4.4GHz Sandy or later CPU better upgrade the GPU than the CPU/Platform.

 

[LTC8K6]

http://www.eurogamer.net/articles/digitalfoundry-2015-intel-skylake-core-i7-6700k-review

Except for GTAV and FC4, at 4.4GHz OC for all the CPUs with an OC Titan X the performance differences are negligible. If you have a 4.4GHz Sandy or later CPU better upgrade the GPU than the CPU/Platform.

I see a few impressive improvements in there...

http://i.imgur.com/FK3Ypex.png

 

[Anubis]

do any of the sites out there have a good comparisons of how SL preforms at bone stock settings, vs all other CPUs at bone stock settings?

I couldn't are less about OCing anymore

 

[LTC8K6]

do any of the sites out there have a good comparisons of how SL preforms at bone stock settings, vs all other CPUs at bone stock settings?

I couldn't are less about OCing anymore

Well, they are unlocked chips meant for overclockers.

No regular ones have been released.

Plus you have the problem of turbo speeds, which are different.

Also, it's beginning to look like Skylake could move up a notch with DDR4 with better timings, if it ever becomes available.

One more thing, Skylake with edram will almost certainly be a step up, given Broadwell's numbers.

Anandtech Bench is a good place for such comparisons, though.

http://www.anandtech.com/bench/product/1543?vs=1260

 

[myocardia]

Crippling Skylake with DDR4 2133 CL15 is the ultimate way to test it, if you want to feel good about your outdated Intel chip.

1) My Intel Devils Canyon chip is newer than your old Haswell-E. 2) Yeah, it's totally fair to penalize only one competitor, while allowing the other unlimited overclocking, as well as unlimited memory controller overclocking. How about this, let's try it the other way around. A 4.8-5.0 Ghz 4970k, running DDR3-2133-2400, while the Skylake runs bone stock, including DDR4-2133. Then it would be fair comparison, wouldn't it?

 

[myocardia]

Also, it's beginning to look like Skylake could move up a notch with DDR4 with better timings, if it ever becomes available.

Even if they never get the timings extremely low, the speeds will continue to rise. I'd much rather have a Skylake with 4,000-4,400 Mhz RAM, with timings of 20-20-20, than 2133/2400/2666 and timings of 10-10-10. When the RAM speeds are the same or close, then timings matter. When the speeds are on the order of twice as fast in Mhz, timings don't matter much, higher Mhz will always win out.

edit: As a matter of fact, when you double the RAM's speed, and also double timings, the two have exactly the same latency. I.e., 2133 Mhz @ 10-10-10 has the identical latency as 4166 @ 20-20-20, since the timings are only related to the speed in Mhz that the RAM is operating, and 20 cycles @ 4,166 Mhz takes the identical amount of time as 10 cycles @ 2,133.

One more thing, Skylake with edram will almost certainly be a step up, given Broadwell's numbers.

Undoubtedly. Have we heard yet officially how long it will be until the release of the edRAM Skylake?

edit #2: I meant the desktop Skylake with the edRAM cache, not the laptop/tablet Y or U versions.

 

[StrangerGuy]

1) My Intel Devils Canyon chip is newer than your old Haswell-E. 2) Yeah, it's totally fair to penalize only one competitor, while allowing the other unlimited overclocking, as well as unlimited memory controller overclocking. How about this, let's try it the other way around. A 4.8-5.0 Ghz 4970k, running DDR3-2133-2400, while the Skylake runs bone stock, including DDR4-2133. Then it would be fair comparison, wouldn't it?

Nah, 4790K has to be stock with the lousiest DDR3 imaginable versus a cherry-picked, delidded Skylake running DDR4 way above official Intel spec to tell the world how good an overclocker it is.

 

[Sweepr]

1) My Intel Devils Canyon chip is newer than your old Haswell-E.

Haswell-E came after Devil's Canyon. Haswell-E is Intel's current enthusiast chip, Devil's Canyon is not.

2) Yeah, it's totally fair to penalize only one competitor, while allowing the other unlimited overclocking, as well as unlimited memory controller overclocking. How about this, let's try it the other way around. A 4.8-5.0 Ghz 4970k, running DDR3-2133-2400, while the Skylake runs bone stock, including DDR4-2133. Then it would be fair comparison, wouldn't it?

No it wouldn't because of much higher DDR4 latencies. You didn't provide any kind of reasonable argumentation as to why other review results aren't valid while AnandTech is other than calling them payed by Intel, which shows that your only purpose here is to troll and defend your current system.

DDR3-1600 CL8/CL9 DDR3 (used in most Haswell tests) is superior to DDR4-2133 CL15 DDR4 in true latency (used in AnandTech's Skylake review).

Haswell DDR3 1600 CL9 = 1.25 ns x 9 CL = 11.25 ns true latency
Skylake
DDR4 2133 CL15 = 0.94 ns x 15 CL = 14.06 ns true latency
DDR4-2400 CL15 = 12.5 ns latency
DDR4-2800 CL15 = 10.7 ns latency
DDR4-3000 CL15 = 10 ns latency

PurePC was one of the few reviews where a clock per clock comparison at (nearly) the same true latency was done and Skylake was 15.8% faster than Haswell per clock in games (overall).
Skylake DDR4 2666 CL13 (9.75 ns)
Haswell DDR3 1600 CL8 (10 ns)

Then there is PCLab which favoured Haswell in terms of true latency and clock/CL ratio and Skylake still came out on top.

AnandTech's method:

Normally in our DRAM reviews I refer to the performance index, which has a similar effect in gauging general performance:

DDR3-1600 C11: 1600/11 = 145.5
DDR4-2133 C15: 2133/15 = 142.2

As you have faster memory, you get a bigger number, and if you reduce the CL, we get a bigger number also. Thus for comparing memory kits, if the difference > 10, then the kit with the biggest performance index tends to win out, though for similar kits the one with the highest frequency is preferred.

PCLab
DDR4-2666/16 = 166 (Skylake-S)
DDR3-1866/9 = 207 (Haswell)

...and Skylake still manages 12.8% better performance per clock overall in 14 games.

Should we stop? No, here's Hardware.fr.

Here's Eurogamer:

Core i5 6600K vs Core i5 4690K (same 3.5-3.9GHz base/turbo)
- 17% faster @ The Witcher 3
- 11% faster @ GTA V
- 10% faster @ Battlefield 4

Can you explain us why even in applications AnandTech showed different results than some other reviews that also used DDR4-2133 (or at most DDR4-2666)?

Even in the same applications the results differ:

7-Zip 9.2
- AnandTech: Skylake 1% slower than Haswell per clock
- Hardware.fr: Skylake 7% faster than Haswell per clock
- Hardware Canucks: Core i7 6700K 11.8% faster than Core i7 4790K with lower turbo clocks

POV-Ray
- AnandTech: Skylake 9.7% faster than Haswell per clock
- HardOCP: ''POV-Ray is a ray tracing for creating high quality graphics. We are using the benchmark included in the software and using its multicore ability. Again we see Skylake stretch its legs. Skylake rewards us with a 16% decrease in render time compared to Haswell, a 21% decrease compared to Ivy Bridge, and a 27% decrease compared to Sandy Bridge.'' (per clock, fixed 4.5GHz)

WinRAR
- AnandTech: Skylake 0.1% slower than Haswell per clock
- Hardware.fr: Skylake 10.85% faster than Haswell per clock
- Lab501: Core i7 6700K 9.4% faster than Core i7 4790K with lower turbo clocks

Now should I ask you, are most reviews being unfavourable to which chip? Hint: it's not Haswell.
Core i7 6700K is the better chip, deal with it.

 

[Burpo]

Can you explain us why even in applications AnandTech showed different results than some other reviews that also used DDR4-2133 (or at most DDR4-2666)?

The media company that acquired Anandtech is about moving products. We can't have all that Haswell stock piling up..

"Purch integrates content and commerce into a powerful data-driven platform that makes complex buying decisions easy for customers"

 

[Edrick]

Core i7 6700K is the better chip, deal with it.

That should be the title of this thread.

 

[Edrick]

Undoubtedly. Have we heard yet officially how long it will be until the release of the edRAM Skylake?

I would be happy if I could just get a 6700K.

 

[Pandamonia]

Haswell-E came after Devil's Canyon. Haswell-E is Intel's current enthusiast chip, Devil's Canyon is not.



No it wouldn't because of much higher DDR4 latencies. You didn't provide any kind of reasonable argumentation as to why other review results aren't valid while AnandTech is other than calling them payed by Intel, which shows that your only purpose here is to troll and defend your current system.

DDR3-1600 CL8/CL9 DDR3 (used in most Haswell tests) is superior to DDR4-2133 CL15 DDR4 in true latency (used in AnandTech's Skylake review).

Haswell DDR3 1600 CL9 = 1.25 ns x 9 CL = 11.25 ns true latency
Skylake
DDR4 2133 CL15 = 0.94 ns x 15 CL = 14.06 ns true latency
DDR4-2400 CL15 = 12.5 ns latency
DDR4-2800 CL15 = 10.7 ns latency
DDR4-3000 CL15 = 10 ns latency

PurePC was one of the few reviews where a clock per clock comparison at (nearly) the same true latency was done and Skylake was 15.8% faster than Haswell per clock in games (overall).
Skylake DDR4 2666 CL13 (9.75 ns)
Haswell DDR3 1600 CL8 (10 ns)

Then there is PCLab which favoured Haswell in terms of true latency and clock/CL ratio and Skylake still came out on top.

AnandTech's method:



PCLab
DDR4-2666/16 = 166 (Skylake-S)
DDR3-1866/9 = 207 (Haswell)

...and Skylake still manages 12.8% better performance per clock overall in 14 games.

Should we stop? No, here's Hardware.fr.

Here's Eurogamer:



Can you explain us why even in applications AnandTech showed different results than some other reviews that also used DDR4-2133 (or at most DDR4-2666)?



Now should I ask you, are most reviews being unfavourable to which chip? Hint: it's not Haswell.
Core i7 6700K is the better chip, deal with it.

Skylake isnt any faster in games since none of them are CPU limited. Plus your slides are not all at the same clock

 

[guskline]

Devil's Canyon (4790k) release date June 2, 2014. Haswell-E (5820k,5830k, 5960x) release date August 29,2014. August 5, 2015 release date for the Skylake 6700k.

I just bought a 4790k so I guess one could say I had one of those Homer Simpson DOH moments. I'm pleased as my chip OC's well to 4.8Ghz and I'm running it 24/7 at 4.7Ghz.

That being said, Skylake 6700k is now top dog in it's category.

 

[LTC8K6]

I'm starting to think those who buy a 6700K now, might regret it when/if Skylake DT with edram comes out.

Starting to think of backing out of buying a 6700K.

 

[ShintaiDK]

Remember its a GT4e model with eDRAM at 65W. So clocks will be just as different as with 5775C vs 4790K.

I wouldnt wait if I needed a CPU upgrade.

 

[cmdrdredd]

I'm starting to think those who buy a 6700K now, might regret it when/if Skylake DT with edram comes out.

Starting to think of backing out of buying a 6700K.

Help me out. Would that matter to anyone using discreet GPU configurations and not using the iGPU at all?

 

[Sweepr]

That should be the title of this thread.

Lol, certainly would make things easier.

Skylake isnt any faster in games since none of them are CPU limited.

Agreed, important point too.

Plus your slides are not all at the same clock

PurePC and PCLab tested at fixed 4.5GHz. You can find the per clock Hardware.fr comparison here (12.1% faster than Haswell at CPU-limited games, overall). Eurogamer's comparison above is pretty close to clock-per-clock (Core i5 6600K and Core i5 4690K both operate at 3.5-3.9GHz and can sustain high MT turbos).

 

[Absolute0]

I'm starting to think those who buy a 6700K now, might regret it when/if Skylake DT with edram comes out.

Starting to think of backing out of buying a 6700K.

There's always something better coming down the pipeline, so the appropriate adage here is: "If you always play the waiting game, you'll be waiting forever."

On the other hand if you upgrade constantly, this one applies:
"It hurts to be on the cutting edge."

I like both of these

 

[alecmg]

Help me out. Would that matter to anyone using discreet GPU configurations and not using the iGPU at all?

if it is same as broadwell setup then yes. Edram latency is roughly half of ddr4 and it helps in many cases including games.

 

[LTC8K6]

Help me out. Would that matter to anyone using discreet GPU configurations and not using the iGPU at all?

Yes, Broadwell beats or ties Skylake in DGPU gaming, despite the much lower clock speed. Broadwell is often impressive with minimum fps over Skylake.

http://www.anandtech.com/bench/product/1500?vs=1544

http://www.anandtech.com/bench/product/1501?vs=1542

This is a big hint that Skylake with edram will be impressive.

 

[mohit9206]

Intel Skylake is like that super pretty girl who everyone hyped up to be the girl of their dreams but when you actually spent some time with her getting ready to dump my current girlfriend , she turns out to be barely any better than my girlfriend. So instead of dumping my current girlfriend for her, i choose to stick with my current girlfriend. I will dump my current girlfriend for a new boyfriend next year (AMD ZEN)