Intel Core i7-6850K Pictured, Tested

[btarunr]

Intel's upcoming Core i7-6850K six-core processor made its way to the hands of an enthusiast on OCN, who wasted no time in picturing the chip, and putting it through a few handy tests. Built for the LGA2011v3 socket, the i7-6850K is based on the 14 nm "Broadwell-E" silicon, and features six cores, HyperThreading enabling 12 logical CPUs, 15 MB of L3 cache, and a nominal clock speed of 3.60 GHz. To begin with, while the i7-6850K is pin-compatible with existing socket LGA2011v3 motherboards (and logically features an identical pin-map to "Haswell-E,") the package is slightly different. Its fiberglass substrate is slightly thinner (1.12 mm vs. 1.87 mm of "Haswell-E,"). Its thickness is made up for by a chunkier IHS (integrated heatspreader).

The i7-6850K sample was installed on a machine with ASRock X99 Extreme3 motherboard (BIOS: P3.30), 16 GB of quad-channel DDR4-2133 memory, and a single GeForce GTX 980 Ti graphics card. It was compared to a Core i7-5820K processor on the same setup. The i7-6850K based setup was barely (~1%) faster at 3DMark FireStrike Extreme in its final score, however, its CPU-intensive Physics score was 14.9% higher. Moving on to the community favorite Cinebench R15, the i7-6850K yielded a 10% higher score. To test its single-core performance, the chip was put through SuperPi 32M, where the i7-6850K crunched through the test in 8m 27.854s, compared to 8m 38.866s by the i7-5820K.



View at TechPowerUp Main Site

 

[proxuser]

flexible 0.03mm thin intel cpu's coming soon.. looks like no need to upgrade next 5 years.

 

[R-T-B]

flexible 0.03mm thin intel cpu's coming soon.. looks like no need to upgrade next 5 years.

The whole thickness of the substrate thing isn't really worth worrying about if coolers are just made properly and you know not to literally crush your CPU by over tightening.

 

[GhostRyder]

Not bad, however if I am reading correctly it has 300mhz more stock clocks which kinda knocks down the performance improvements a notch. I wonder though how many lanes it has?

 

[Rannick1982]

In comparison to the 5820K, which I recently upgraded to from an FX-8350, I don't think there is much reason to come up from the 5820K. Just my opinion though, since I don't need a large number of PCIe lanes.

 

[xkm1948]

Not bad, however if I am reading correctly it has 300mhz more stock clocks which kinda knocks down the performance improvements a notch. I wonder though how many lanes it has?

You are reading it wrong. The testing was done at 4.2GHz for both 5820K and 6850K. So a solid 10% IPC improvement.

 

[Chaitanya]

Not bad, however if I am reading correctly it has 300mhz more stock clocks which kinda knocks down the performance improvements a notch. I wonder though how many lanes it has?

Most probaly 40, since there is a rumoured 6800K which will sit at the bottom of Hedt stack and will have a 28 pci-e lanes.

 

[badtaylorx]

I wonder if it'll make the intel retail edge summer sale???

 

[techy1]

You are reading it wrong. The testing was done at 4.2GHz for both 5820K and 6850K. So a solid 10% IPC improvement.

where did you read that? in case if your claim is false .... there is precise 0.00 IPC improvments

 

[bogami]

So thin. There is no good to remove the cover as it will certainly twists as we sow on the 6700K .and thermal behavior will certainly worse as the disperser is thicker .81 ° but we do not know with what there are cooling.

 

[xkm1948]

where did you read that? in case if your claim is false .... there is precise 0.00 IPC improvments

Original post is here. Read on

http://www.overclock.net/t/1599068/6850k-vs-5820k

 

[silentbogo]

where did you read that? in case if your claim is false .... there is precise 0.00 IPC improvments

HWInfo window on the screenshot. Maximum core clock column.

 

[Keullo-e]

Why they had to make the PCB thinner even on the enthusiast platform? What next, crappy TIM instead of solder on these chips?

 

[Ferrum Master]

It looks a bit more cool too, looking at the Max temps. I was expecting more.

 

[jaggerwild]

He's testing it on a low End AsRock

Cooler is Intel 120mm BXTS13X

,

 

[Caring1]

He's testing it on a low End AsRock ,

How did you deduce that from using an Intel aftermarket AIO cooler?
They could be using any brand motherboard.
I see they used the same Asrock Extreme 3, not quite low end.
It seems they have overcome the TSX bug and implemented it again in this processor.

 

[Ferrum Master]

How did you deduce that from using an Intel aftermarket AIO cooler?.

Read the OCN thread.

 

[cdawall]

How did you deduce that from using an Intel aftermarket AIO cooler?
They could be using any brand motherboard.
I see they used the same Asrock Extreme 3, not quite low end.
It seems they have overcome the TSX bug and implemented it again in this processor.

I don't know what you consider low end but the asrock extreme 3 is about as bottom of the barrel as x99 goes.

 

[GhostRyder]

You are reading it wrong. The testing was done at 4.2GHz for both 5820K and 6850K. So a solid 10% IPC improvement.

Oops, just noticed the maximum core clocks and forgot that Cinebench does not report that. Then its about what I expected in performance jumps.

Most probaly 40, since there is a rumoured 6800K which will sit at the bottom of Hedt stack and will have a 28 pci-e lanes.

Yea, sounds about right but still it would not surprise me if they made two that had the lower PCIE amounts.

 

[cdawall]

Yea, sounds about right but still it would not surprise me if they made two that had the lower PCIE amounts.

I'm hoping they bump it up some to allow more nvme. 36 lanes would be rather nice in the entry level chips

 

[newtekie1]

I don't know what you consider low end but the asrock extreme 3 is about as bottom of the barrel as x99 goes.

Probably why he can only get 4.2GHz on the 5820k, what he said is why he limited the 6850K to 4.2GHz, but still respectable enough for testing performance improvement between the two chips.

Though, I'd be interested in what is able to get the 6850K to in the same board.

 

[cdawall]

Probably why he can only get 4.2GHz on the 5820k, what he said is why he limited the 6850K to 4.2GHz, but still respectable enough for testing performance improvement between the two chips.

Though, I'd be interested in what is able to get the 6850K to in the same board.

I would be as well, would be interested in power consumption between the two is.

 

[EarthDog]

where did you read that? in case if your claim is false .... there is precise 0.00 IPC improvments

Look at the screenshots and the peak clocks... or the thread.. .or...........

If you look at the screenshots, he made all the settings the same. 4.2GHz core, 3.3GHz Cache, 2666MHz RAM (4x4GB 13-14-14-39 2T 347 TFRC). Only thing I see different is 6850k is 0.02v lower VID (1.2v 5820k vs 1.18v 6850k). All the performance increase you see is from better IPC.

As far as the mobo... not sure it will make too much difference... remember, Broadwell still has the FIVR on the CPU so the motherboard plays less of a role on power delivery as compared to CPUs without it. Unless, I have my processors mixed up?

 

[truth teller]

10% improvement in cinebench
2% improvement in superpi
substrate easier to bend with cooler weight
definitely priced higher than 10% over previous competitor

what am i looking at here? is this supposed to make me excited? cause its not...

 

[cdawall]

As far as the mobo... not sure it will make too much difference... remember, Broadwell still has the FIVR on the CPU so the motherboard plays less of a role on power delivery as compared to CPUs without it. Unless, I have my processors mixed up?

Unless he picked 4.2 out of his ass for a number the board still plays some roll. My chip does 4.5 on one of the two gb x99's and 4.6 on the other all other settings equal and those are two identical boards. I would imagine a cheaper board would do worse.