Skylake-K SKUs leaked

[CHADBOGA]

4.0 GHz base 4.2 GHz Turbo? Only other CPU that does is FX-8350.

Intel copying AMD again. D:

 

[Fjodor2001]

Intel Core i5-6600K (Skylake, 14 nm): 3.5/3.9 GHz, 95 W TDP
Intel Core i5-3570K (Ivy Bridge, 22 nm): 3.4/3.8 GHz, 77 W TDP

Whoever said Intel is prioritizing lower power consumption over performance/frequency increase on desktop these days?

I'm seeing nearly no frequency increase, so that part is true. But I'm also seeing 95/77=1.23 => 23% TDP increase. Soooo.... lower power consumption?? Not so much, in fact the opposite... 😱

Is the additional TDP wasted on the iGPU?

 

[jj109]

All we have is rated TDP, clock speed, and a name. There's not enough information to begin to guess at CPU/iGPU performance.

Is the additional TDP wasted on the iGPU?

Or the CPU cores are beefier and have significantly higher IPC, requiring more joules per cycle.

See? Pointless speculation.

 

[Nothingness]

Or the CPU cores are beefier and have significantly higher IPC, requiring more joules per cycle.

Preliminary Geekbench results seem to indicate there's no significant IPC increase.

 

[ClockHound]

See? Pointless speculation.

So...could you point us in the direction of pointed speculation?

If we can't speculate in this forum about skylake-k where can we? Home theatre forum?

 

[ClockHound]

Preliminary Geekbench results seem to indicate there's no significant IPC increase.

Stop it! You're hurting my eyes! :biggrin:

 

[NTMBK]

Intel Core i5-6600K (Skylake, 14 nm): 3.5/3.9 GHz, 95 W TDP
Intel Core i5-3570K (Ivy Bridge, 22 nm): 3.4/3.8 GHz, 77 W TDP

Whoever said Intel is prioritizing lower power consumption over performance/frequency increase on desktop these days?

I'm seeing nearly no frequency increase, so that part is true. But I'm also seeing 95/77=1.23 => 23% TDP increase. Soooo.... lower power consumption?? Not so much, in fact the opposite... 😱

Is the additional TDP wasted on the iGPU?

My guess is that the voltage/frequency curve is steeper for 14nm than it is for 22nm- giving significant power reductions at lower frequencies, but at top end frequencies it is actually detrimental. (Similar to what AMD did with Carrizo: http://cdn.wccftech.com/wp-content/uploads/2015/02/AMD-Carrizo-APU_Excavator-Core-Architecture.jpg )

 

[witeken]

My guess

Skylake 14nm goes from 84W to 65W.

 

[Tsavo]

Preliminary Geekbench results seem to indicate there's no significant IPC increase.

Lol. More of the same.

Maybe Intel need to revive the P4 architecture. What they have now is exciting as raw toast.

 

[NTMBK]

Skylake 14nm goes from 84W to 65W.

Within a certain frequency range, yes, but the top frequency bin goes from 88W to 95W. Which is exactly what I was saying 🙂 For most of the frequency range power usage is lower, but top end frequency uses more power.

 

[ShintaiDK]

We dont have the TDP, hence the note at the TDP. Its platform TDP.

 

[Nothingness]

We dont have the TDP, hence the note at the TDP. Its platform TDP.

I'm not sure where you got that from, and what "platform" means.

 

[ShintaiDK]

I'm not sure where you got that from, and what "platform" means.

IB and HW is also 95W platform TDP. Platform TDP is 35, 65 and 95W on the LGA115x platform. Whatever the CPUs actually end up with may be an entirely different case. It could be anything between 66 and 95W for these parts. SB was the last one to actually hit the platform limit.

Example:

95W turned out to be 77W.

 

[Fjodor2001]

I think that slide was from an early IB estimate, and no TDPs were specified for individual SKUs either.

We're now just months away from Skylake launch, so the TDP numbers are more likely to be known and accurate, especially when individual SKUs are mentioned.

 

[frozentundra123456]

My guess is that the voltage/frequency curve is steeper for 14nm than it is for 22nm- giving significant power reductions at lower frequencies, but at top end frequencies it is actually detrimental. (Similar to what AMD did with Carrizo: http://cdn.wccftech.com/wp-content/uploads/2015/02/AMD-Carrizo-APU_Excavator-Core-Architecture.jpg )

Well the i7 does have a frequency increase of over 10% compared to ivy bridge at the same 95 watt tdp as the skylake i5, as well as an unknown ipc increase on top of the 10 to 15% we have already seen with haswell. So actual performance per watt is probably at least as good as IB. I also think as Shintai said, the 95 watt may just be a general category limit, and may or may not be the final tdp.

 

[Magic Carpet]

20% IPC increase over HW likely?

1 or 0.

 

[Arachnotronic]

20% IPC increase over HW likely?

1 or 0.

0.

I think we'll see something on the order of 10-15%, with the upper end of that range really a stretch.

 

[ShintaiDK]

I think we should be happy if its even 10%.

 

[Arachnotronic]

I think we should be happy if its even 10%.

BDW brought a 5.5% improvement over HSW, so if SKL is just a 5% improvement over BDW, then you get ~10% over HSW.

 

[ShintaiDK]

Arh ye sorry. I was thinking from Broadwell. You are right in 10-15% then.

 

[Arachnotronic]

Arh ye sorry. I was thinking from Broadwell. You are right in 10-15% then.

:thumbsup:

So, assuming our thinking is right, if SKL brings >= 10% perf/clock, then a 4GHz 6700K should deliver at least the same performance as a 4.4GHz stock 4790K.

That's some serious computing power.

 

[Magic Carpet]

Even 10-15% range is fine by me. My 4770K can't do more than 4.2 on stock volts. Two other things still unknown are power consumption and temps. At the very minimum I expect it to be not worse than my current chip.

Any word on MT performance? I heard it will be further improved. And the last, with the FIVR removed, is it reasonable to expect the idle watts to go up slightly? i.e. less power efficient at lower loads.

Thanks guys.

 

[Nothingness]

:thumbsup:

So, assuming our thinking is right, if SKL brings >= 10% perf/clock, then a 4GHz 6700K should deliver at least the same performance as a 4.4GHz stock 4790K.

That's some serious computing power.

Serious computing power, yes, enough to justify a switch from Haswell 4770 certainly not, at least not for me 🙁

 

[IntelUser2000]

Intel Core i5-6600K (Skylake, 14 nm): 3.5/3.9 GHz, 95 W TDP
Intel Core i5-3570K (Ivy Bridge, 22 nm): 3.4/3.8 GHz, 77 W TDP

Or that 6700K is a better quality/binned chip than 6600K. Otherwise how does it clock 10%+ better?

Personally I am hoping they can pull 15%+ gains Sandy Bridge had since its the same team with possibly same goals that's designing Skylake.

yes, enough to justify a switch from Haswell 4770 certainly not, at least not for me 🙁

How much would it take you?

For me it would take 50% ST gain and 2x MT gain at the minimum over my 2600K to justify it, and it would still be doubtful. Really to be convincing it should be 3x ST and 5x MT. Personally, I am slightly disappointed at going from Core i5 661 to Core i7 2600K.

 

[AtenRa]

http://www.kitguru.net/components/c...ns-of-unlocked-intel-skylake-processors-leak/

Core i7-6700K – 4 cores with Hyper-Threading, 4.0GHz frequency, 4.20GHz maximum Turbo Boost frequency, 8MB last-level cache, dual-channel DDR3/DDR4 memory controller with 1600MHz or 2133MHz support, 95W TDP, Intel HD Graphics 5000-series integrated graphics core, LGA1151 packaging;

Core i5-6600K – 4 cores, 3.50GHz frequency, 3.90GHz maximum Turbo Boost frequency, 6MB last-level cache, dual-channel DDR3/DDR4 memory controller with 1600MHz or 2133MHz support, 95W TDP, Intel HD Graphics 5000-series integrated graphics core, LGA1151 packaging;

Ehm, HD5000 series with Skylake ???

Now HD5000 series will be with Haswell, Broadwell and Skylake ?? The big mess with the Intel iGPU naming continues.