Broadwell-E Review

[csbin]

http://www.anandtech.com/show/10337...x-6900k-6850k-and-6800k-tested-up-to-10-cores

PCPER: http://www.pcper.com/reviews/Proces...well-E-Review/Hardware-and-Quick-Overclocking

Overclocking the Core i7-6950X

Hitting 4.3 GHz on all 10-cores (!!) produces some amazing performance increases.

• CineBench 11.5
  • Stock: 19.05
  • 4.3 GHz: 23.62
  • 24% increase
• POV-Ray
  • Stock: 3556.71
  • 4.3 GHz: 4262.92
  • 20% increase
• Handbrake
  • Stock: 2551.23
  • 4.3 GHz: 338.89 FPS
  • 14% increase

Getting up to 24% better performance with a simple overclock, though it does require better cooling than a standard air cooler can provide, is fantastic. As it turns out, applying an 800 MHz overlock to 10 cores nets you quit a bit!

I was disappointed that overclocking seems to be more limited on Broadwell-E than it was with Haswell-E; in my review of the Core i7-5960X I was able to run all 8-cores at 4.6 GHz. I’ll wait and see what the rest of the community finds with the other parts in the BDW-E lineup.

 

[ElFenix]

Any idea who will be doing some gaming centric reviews? Doom, wh:tw, that sort of thing?

Wondering if there is any point yet to upgrading from ivy

 

[IntelUser2000]

The Turbo Boost Max 3.0 is quite a clever feature. Basically stops core hopping + chooses best individual cores from factory. PC Perspective review is showing 13% improvement.

That's really nice. But the engineers are really eeking out every opportunities to get maximum stock performance. Core i7 4790K showed us that they basically gave us factory overclocked CPU. It just ate up overclock headroom. There definitely IS an absolute clock speed limit that process, TDP, and architecture can't seem to break.

-The new Turbo Boost
-Factory overclocking
-Speed Shift

It's just an indication of not just Moore's Law on its deathbed, but also architectural gains. They seem to be able to still bring 5-10% improvements, but with seemingly heroic efforts. This is really the reason why PC is dying folks. They are already at the end of advancement, while mobile still has ways to go. Also, process tech research is all heading into areas that will(for the near future) bring healthy gains in performance for Smartphone form factors.

I wonder in 2018 Intel, Apple, and AMD will all end up at similar general perf/clock?

 

[maddogmcgee]

Man with those clocks, a 5820k is the better gaming CPU. Power usage does not look that much better either. Kinda disappointing.

 

[StrangerGuy]

6800K may make sense if one needs a $200+ mobo for some reason.

Otherwise, what a snoozefest.

 

[2blzd]

aren't most good z170 boards $200+?

Im looking at getting a 6800k and a $200 mobo.

 

[SPBHM]

aren't most good z170 boards $200+?

Im looking at getting a 6800k and a $200 mobo.

define good? I think you can buy good z170 boards around $100-150

the gap in ST to the 6700K is pretty sad for a 1.7K USD CPU, even a 4770K from 2013 is going to give it a hard time, if they had the best MT and ST performance perhaps the price would be almost understandable but it still is a big tradeoff...

also, looking at previous "X" CPUs from generation to generation, 4 to 6 cores had a smaller price increase, 6 to 8 cores had no price increase, why the absurd increase now!? specially when Skylake is old news by now on the cheaper CPUs and they release Broadwell at this price... :thumbsdown:

 

[ZGR]

Too bad RAM overclocking wasn't part of the review. Would be nice to see BW-E IMC's speed limit. Good review overall. Nothing too surprising given the leaks we already saw.

The game benchmarks were largely useless because of the settings they used. Being GPU limited tells us very little. It is time we get some old games like ARMA 2 or an older RTS game just to show IPC improvements. I like the idea of having modern games for benchmarks, but keeping an old game (especially a CPU limited one) will give us a good idea of CPU performance on single-threaded scenarios.
A lot of older games have replays and saves to make the testing very consistent.

I hope in the future the 6950X will get another article about RAM scaling. If a 5960x owner @ 4.4 GHz+ is looking for an upgrade, significantly faster RAM support would be a big highlight versus the 2 additional cores.

I was hoping we would see much higher clocks compared to Broadwell-C. Without the eDRAM*, the lower overclocking headroom kinda hurts.

*The 25MB L3 may make it redundant-- but how will we know for sure?

 

[guskline]

Good review. I'm sitting here with my $999 5960x and the increase to the Broadwell E 6950x is just too big of a jump in price to justify switching out.

I suspect as the 6950x is out longer and produced more the overclocking might improve.

 

[24601]

That being said, speaking with our contacts at various motherboard manufacturers, we're told that 4.1 GHz is a reasonably average processor result for Broadwell-E. Some processors will hit 4.3 GHz on air at around the same voltage, whereas others need up to 1.4 volts, and thus results will depend on the cooling setup used or the thermal characteristics of the silicon. I have also been told that AVX is a different story: for any peak frequency attained normally, AVX overclock stable frequencies will be around 200-300 MHz lower.

Literally pointless in every way for gamers comparatively.

3.8 ghz for avx loads OVERCLOCKED is pure trash.

The 6950x makes literally zero sense at 1723 USD with this information in mind.

 

[Arachnotronic]

Bought me a 6800K

 

[Arachnotronic]

Any idea who will be doing some gaming centric reviews? Doom, wh:th, that sort of thing?

Wondering if there is any point yet to upgrading from ivy

What GPU are you running? Are you on Ivy E or Ivy mainstream?

 

[IEC]

6950X OC can only be described as miserable for an enthusiast part. And that price tag... yikes. If I needed more than 8 cores on my workstation I could spring for a proper Xeon and the more cores + better features of that platform.

We can only hope that AMD is at least reasonably close with its 8-core parts in Zen so we can get some more reasonable pricing in the HEDT segment. Yeesh.

 

[24601]

Time to hurry up and wait another full year for Skylake-E.

Hope i can fight off the urge to jump onto the 6700k from my 3570k before then.

 

[frozentundra123456]

Pretty much what was expected except for the price. Only hope is that the lower core models have somewhat better overclocking. I predicted earlier that there could be little to no performance improvement or even a bit of a regression due to poorer overclocking and minor IPC improvements. Looks like that could be coming true.

Really a lot of hubris to raise prices with this kind of lineup.

 

[guskline]

Bought me a 6800K

Very smart move. I have my 5960x at 4.4Ghz solid. Some have them higher but I wanted absolute stability in Asus RealBench.

I'll be interested in what scores you get.

 

[Arachnotronic]

Very smart move. I have my 5960x at 4.4Ghz solid. Some have them higher but I wanted absolute stability in Asus RealBench.

I'll be interested in what scores you get.

When I had a 5960X, I also ran it at 4.4ghz. 4.5ghz wasn't 100% stable for me either.

 

[AtenRa]

Bought me a 6800K

why not the 6850K ???? it has 40 lanes and you have said you are going to get 2x 1080.

 

[Arachnotronic]

why not the 6850K ???? it has 40 lanes and you have said you are going to get 2x 1080.

One 1080 per system for now. 6700K rig gets one and now my soon-to-be-Broadwell-E rig gets one

Also, 12 extra PCIe lanes isn't worth a couple hundred bucks to me, especially since I've run SLI on a 6700K with great scaling.

 

[csbin]

http://www.guru3d.com/articles_pages/core_i7_6950x_6900k_6850k_and_6800k_processor_review,9.html

CineBench 11.5

5960X（3G/3.5G）=15.16
6900K(3.2G/3.7G)=15.02

wtf.....

 

[24601]

http://www.guru3d.com/articles_pages/core_i7_6950x_6900k_6850k_and_6800k_processor_review,9.html

CineBench 11.5

5960X（3G/3.5G）=15.16
6900K(3.2G/3.7G)=15.02

wtf.....

Guess that straight up confirms the lower AVX clockspeed ratio.

 

[Arachnotronic]

Guess that straight up confirms the lower AVX clockspeed ratio.

I don't think Cinebench uses AVX.

The new version runs only on 64bit systems, but still provides no support for current instruction sets such as Intel's AVX.

http://www.hardwareluxx.com/index.php/news/software/benchmarks/28060-cinebench-r15-released.html

 

[24601]

I don't think Cinebench uses AVX.



http://www.hardwareluxx.com/index.php/news/software/benchmarks/28060-cinebench-r15-released.html

That is extremely odd then.......

 

[Arachnotronic]

That is extremely odd then.......

Yeah, who knows. Could be an error with the testing, could be that the test is sensitive to L3$ latency and the larger cache on the BDW-E is hurting performance in this manner (larger caches tend to be higher latency), or maybe something else.

Across the board though the Broadwell-E counterparts to Haswell-E seem to be ever-so-slightly faster, though.

 

[24601]

http://www.guru3d.com/articles-pages/core-i7-6950x-6900k-6850k-and-6800k-processor-review,12.html

Looks like Broadwell-E's DDR4 IMC hits saturation point very quickly, but derives the latency benefits from the higher speed ram.

Definitely looks like DDR4 2400 is the highest Broadwell-E can handle in terms of actual throughput.