5820K vs 6700K CPU Showdown

[nicklmg]

Intel Core i7 5820K

Amazon: http://geni.us/25jW

NCIX: http://bit.ly/1E6sBq5

 

Intel Core i7 6700K

Amazon: http://geni.us/3eNk

NCIX: http://bit.ly/1NR2c44

 

How much will having more cores on your CPU help you? Should you be investing in an X-series Chipset or will the more mainstream Z-series be more than enough?

 

[KemoKa]

Wow, look at the power stages between X99 and Z170.
 

That's what happens when you move the FIVR off the board

[Duskie]

Wow, look at the power stages between X99 and Z170.

 

That's what happens when you move the FIVR off the board

Yup, my LAPTOP with a 6820HK is stable at 4Ghz without any voltage changing  :lol:

[jurian123]

X xtream money loss. 

[Qyoon]

I have no  much use of so many core... but X99 is damn sexy 

[MaxBunny]

Or maybe waiting on a Zen is the best option?

[Jaggsta]

i7 5820k $379 at Superbizz

ASRock X99 Extreme4 $150.99 AR @ Newegg,

G.Skill Ripjaws V Series 16GB (2 x 8GB) DDR4 $96.99 @ newegg

 

Total $626.98 beats $745 price anyday.

[Streetguru]

I usually go with

"If it's within your budget, you might as well go X99"

[Kirky2k15]

I usually go with

"If it's within your budget, you might as well go X99"

Especially when z170 is ALMOST the same price as x99 (in UK anyway)

[Sharif]

Hyped to watch this! 

[Egad]

This was a pretty disappointing test to be honest. I would have loved to see both chips dialed in with the same clock speed and then run head to head so we could see exactly how much Skylake's IPC increase adds. Especially on benchmarks like Attila Total War, where we know the game will make use of extra cores. So then we'd get to see exactly where we are in terms of displacement vs IPC with the current processors.

Also running K chips at stock clocks is somewhat pointless. Intel sells them all with with clocks on the more conservative side so that almost all of them overclock and Intel can avoid any customers feeling angry over losing the Silicon Lottery. Almost everyone gets a moderate overclock and thus very few folks get mad at Intel for getting a bum chip. The X series chips are especially prone to underrating themselves and then putting out massive overclocks so their owners can walk around feeling like OCing demigods. Linus himself touched on it 11 months ago

discussing the 5820K as the best bang for the buck and one of better OCing options. So basically you end up with the battle of chips whose clocks were partly set by Intel's marketing teams to avoid any pissing and moaning to customer support.

[Deli]

This was a pretty disappointing test to be honest.  I would have loved to see both chips dialed in with the same clock speed and then run head to head so we could see exactly how much Skylake's IPC increase adds.  Especially on benchmarks like Attila Total War, where we know the game will make use of extra cores.  So then we'd get to see exactly where we are in terms of displacement vs IPC with the current processors.   

 

Also running K chips at stock clocks is somewhat pointless.  Intel sells them all with with clocks on the more conservative side so that almost all of them overclock and Intel can avoid any customers feeling angry over losing the Silicon Lottery.  Almost everyone gets a moderate overclock and thus very few folks get made at Intel for getting a bum chip.  The X series chips are especially prone to underrating themselves and then putting out massive overclocks so their owners can walk around feeling like OCing demigods.  Linus himself touched on it 11 months ago in his overclocking video discussing the 5820K as the best bang for the buck and one of better OCing options.  So basically you end up with the battle of chips whose clocks were partly set by Intel's marking teams to avoid any pissing and moaning to customer suppor.  

Agree. There are idiots out there trying to prove their 6700K is so much superior than 5820K by testing them at stock clock. Come on, man.

[Dr_Curry_MONSTER]

Maybe redo this vid with overclocking!!

[PrincessLirael]

Ok I am a little confused and I'm wondering if some one can help me out please?

Watching the video Luke says at about the 1 minute 30 mark that "170 chipset you get twenty lanes with the 6700k". And looking at it on the Intel web site it says in the specifications for the Intel® Core i7-6700K Processor the max number of PCI Express lanes is 16.

My maths here might be wrong but 20 is grater that 16?

So were is he getting the extra 4 lanes from, or does the chipset its self contains 20 lanes?

Thanks. =^ ,^=

[NumLock21]

Ok I am a little confused and I'm wondering if some one can help me out please?

Watching the video Luke says at about the 1 minute 30 mark that "170 chipset you get twenty lanes with the 6700k". And looking at it on the Intel web site it says in the specifications for the Intel® Core i7-6700K Processor the max number of PCI Express lanes is 16.

My maths here might be wrong but 20 is grater that 16?

So were is he getting the extra 4 lanes from, or does the chipset its self contains 20 lanes?

Thanks. =^ ,^=

Cpu has 16 lanes and chipset itself has 20 lanes.

[Cheddle]

you call this a showdown?? no overclocking... no clock for clock comparison?? no power consumption comparison?? you just plugged them in and ran them?? lame... why even bother.

[Dr_Curry_MONSTER]

yeah this was a bit of phony test imo.

[Swndlr]

you call this a showdown?? no overclocking... no clock for clock comparison?? no power consumption comparison?? you just plugged them in and ran them?? lame... why even bother.

You cant really consider overclocking a main factor. I get that everyone should get around the same results with chips nowadays, but its more of a solid comparison to compare them both as is.

I do agree that they should have given more stats rather than synthetic numbers.

[Cheddle]

You cant really consider overclocking a main factor. I get that everyone should get around the same results with chips nowadays, but its more of a solid comparison to compare them both as is.

I do agree that they should have given more stats rather than synthetic numbers.

 

overclocking is quite important with these two chips... both are unlocked... comparing them at stock speeds is pointless.

 

Most Hawsell based chips can reach 4.5ghz - be it a g3258, a 4670k, a 4790k, or a 5820k - even the 5960x generally reaches 4.5ghz.

 

given there is a 700mhz starting difference in the 5820k and a 4790k - yet they both can be clocked to roughly the same frequencies - I feel like the 5820k's huge overclockability was ignored in this "showdown" 

 

---

 

some more tests would have been great as well but comparing 3.3ghz 6 core against 4.0ghz 4 core is always going to tell the same story. 

[NumLock21]

overclocking is quite important with these two chips... both are unlocked... comparing them at stock speeds is pointless.

 

Most Hawsell based chips can reach 4.5ghz - be it a g3258, a 4670k, a 4790k, or a 5820k - even the 5960x generally reaches 4.5ghz.

 

given there is a 700mhz starting difference in the 5820k and a 4790k - yet they both can be clocked to roughly the same frequencies - I feel like the 5820k's huge overclockability was ignored in this "showdown" 

---

some more tests would have been great as well but comparing 3.3ghz 6 core against 4.0ghz 4 core is always going to tell the same story. 

Even though both chips are unlock and allows for oc, most will get these cpus because they want the best, but it does not mean they know how to oc or they're afraid of ocing as it might damage the chip. I agree they should add a oc comparison to show how well they overclock, but doing it at stock speed is better off cause it shows, the cpu's true performance.

[Egad]

Even though both chips are unlock and allows for oc, most will get these cpus because they want the best, but it does not mean they know how to oc or they're afraid of ocing as it might damage the chip. I agree they should add a oc comparison to show how well they overclock, but doing it at stock speed is better off cause it shows, the cpu's true performance.

 

No the true performance of a K chip is its expected stable frequency without having to get into voltage increasing you get via basic overclocking.  The people who do really impressively over engineered cooling loops that cost as much as the rest of their rig don't count, but neither due to the people running K chips on stock clocks.  The latter should have saved some money and bought a locked chip.  They don't need a comparison between gimped K's to help them, they need to educated on which chip to buy in the first place.  For the normal users who are looking at a K chip, a good motherboard, and a top of the line air or AIO cooler, where the K chips finish is what matters.  

 

At the very least, the floor for testing these should be the numbers using the overclocking utilities that come with lots of mobos these days (where you select cooler, etc and the mobo just dials in a conservative overclock automatically).   Most people buying K chips pay a premium for potential and those that never even try to realize the potential aren't relevant. 

[NumLock21]

No the true performance of a K chip is its expected stable frequency without having to get into voltage increasing you get via basic overclocking. The people who do really impressively over engineered cooling loops that cost as much as the rest of their rig don't count, but neither due to the people running K chips on stock clocks. The latter should have saved some money and bought a locked chip. They don't need a comparison between gimped K's to help them, they need to educated on which chip to buy in the first place. For the normal users who are looking at a K chip, a good motherboard, and a top of the line air or AIO cooler, where the K chips finish is what matters.

At the very least, the floor for testing these should be the numbers using the overclocking utilities that come with lots of mobos these days (where you select cooler, etc and the mobo just dials in a conservative overclock automatically). Most people buying K chips pay a premium for potential and those that never even try to realize the potential aren't relevant.

I'm saying true performance as in the speed the product is advertised at. Showing the viewers the results of oc, will be false advertising cause that's not the stock speed of those cpus. Thus why, they could have also include a comparison to show how far these cpus can be pushed along with their stock results. And they have to put in fine print, that results may vary based on silicon lottery and cooling configuration.

[Egad]

I'm saying true performance as in the speed the product is advertised at. Showing the viewers the results of oc, will be false advertising cause that's not the stock speed of those cpus. Thus why, they could have also include a comparison to show how far these cpus can be pushed along with their stock results. And they have to put in fine print, that results may vary based on silicon lottery and cooling configuration.

 

LTT isn't advertising Intel within the review itself.  So false advertising in no way, shape or form applies.  As reviewers the purpose is to review the product in the manner which it will be used, which is people trying to dial in at the very least a conservative overclock and recent K chips have shown themselves to be consistently capable of at least a conservative overclock.  

[NumLock21]

LTT isn't advertising Intel within the review itself.  So false advertising in no way, shape or form applies.  As reviewers the purpose is to review the product in the manner which it will be used, which is people trying to dial in at the very least a conservative overclock and recent K chips have shown themselves to be consistently capable of at least a conservative overclock.  

I'm not saying LTT is false advertising. Just saying that when you do a review on a product, it's best to run it at stock speeds and that's what they did. They should also add overclocking benchmarks just to show how far these unlock cpus are capable of overclocking. It's like doing a car review, the car comes with 200HP from the factory, the reviewer must review the car based the factory specs, they can't just tune it for more HP and say, look how fast and well it performs, because the stock car doesn't have that extra HP. Same with cpus, If you work at a computer shop and your doing a demo on a system with a 6700K, you can't just show your customers the performance of a overclock cpu. You start with stock clocks first show them that, and then explain what's so special about these unlock cpus, then you do the demo again with overclocked results, so the customers can see the difference.

[1-0-1]

Okay so I am not completely clear on the whole PCIe lanes still. Here is my envisaged setup and usage summary:

1. I will never run SLI or Crossfire.
2. Not interested in M2 solid state drives.
3. Never really bother or will ever bother with overclocking.
4. Ordered a 980Ti as part of my upgrade from a socket 1156 (P55 Express chipset) upgrade.
5. Have no need for extra LAN ports, sound card, PCIe ports beyond for the GPU, etc.

To this end I have choosen the ASUS B150M-K paired with the Intel i5-6400. From all the Intel tech papers and wiki I cannot make out if there is a performance issue between Z179 and B150. Both have DMI 3, System Bus speed of 8 GT/s, PCI Express 3.0. So for my usage scenario I do not see the point of splashing out for a Z170 with all sorts of bells and whistle which does not impact performance directly even running DDR3 versus DDR4.

 

The only thing I am unclear of is that B150 supports a maximum of 8 PCI Express lanes and the Z170 supports a maximum of 20 PCI Express lanes. The CPU (i5-6400) supports 16 maximum PCI Express lanes.

 

How does the CPU and motherboard PCI Express lanes play of each other?

 

The motherboard manual of the ASUS B150M-K does not indicate weather the PCI Express x16 port is shared with another PCI Express port, USB, SATA, etc. So if the chipset supports a max of 8 PCI Express lanes does it mean that the advertised PCI Express x16 lane on the ASUS B150M-K board actually can only run at x8?

 

Informational links to my post:

 

Comparison of all the chipsets and CPU mentioned above: http://ark.intel.com/compare/88185,90592,90591

ASUS B150M-K motherboard information: https://www.asus.com/us/Motherboards/B150M-K-D3/specifications/

Online retailer price for the ASUS B150M-K motherboard: http://nam-webshop.com/item?item_id=14939 (displayed currency is Namibia Dollar)

B150 block diagram: http://www.intel.com.au/content/www/au/en/chipsets/business-chipsets/b150-chipset.html

Z170 block diagram: http://www.intel.com.au/content/www/au/en/chipsets/performance-chipsets/z170-chipset-diagram.html

Wikipedia chipset info: https://en.wikipedia.org/wiki/List_of_Intel_chipsets#LGA_1151

                                  https://en.wikipedia.org/wiki/LGA_1151