Intel Core i7-5960X vs i7-5930K vs i7-5820K

[Scrizz]

they run in circles arround your I7 920 when crunching numbers and demanding tasks. In games not so much especially if you got a D0 CPU, then turn it up, and the old X58 platform has new life

My 970 at 4.7GHz was beast ...
It's actually faster than what I'm on now. :/

 

[xorbe]

Strange that in few places 4770K was ahead of 4790K

Page 5, the 4790K lost to 4770K every single time. I noticed too.

 

[Schmuckley]

I think 3dmark11 score is not calculated in KB/s lol

I know that 3dm11 has a separate CPU and GPU score

 

[farjam]

I know that 3dm11 has a separate CPU and GPU score

Yes, Sure mate , but i think i wrote my comment a little bad
by that comment i meant, when we wanna write a 3d11 score , no need to write "KB/s" after the "score number".
because the calculated score by 3dmark11 [and all 3d benchmarks] is not in "KB/s unit" and 3dmark's final score calculated from the "FPS/s" not "KB/s"[like some of 2Ds] and writing KB/s next to 3dmark11 score[that calculated by FPS] is wrong and maybe it's was a typing mistake

 

[eliberate]

my first post here and i somehow feel the need to spread shit around!!!
A lot of people give thanks to the reviewer for the article but i say that's plain bullshit.

You test a highend platform in 3dmark and games. That's oke for the LGA1150 platform cuz that's where it's aim is at, but to test a LGA2011-3 platform that way is stupid.
8 core system is not aimed at games!!! that;s meant for productive people not gamers.
Make a decent review and test SBE VS IBE VS HE in photoshop/cinebench and other core hungry/bandwidth damanding aps.
Ure testing a Lambo on a offroad course in ure review.

 

[Ikaruga]

my first post here and i somehow feel the need to spread shit around!!!
A lot of people give thanks to the reviewer for the article but i say that's plain bullshit.

You test a highend platform in 3dmark and games. That's oke for the LGA1150 platform cuz that's where it's aim is at, but to test a LGA2011-3 platform that way is stupid.
8 core system is not aimed at games!!! that;s meant for productive people not gamers.
Make a decent review and test SBE VS IBE VS HE in photoshop/cinebench and other core hungry/bandwidth damanding aps.
Ure testing a Lambo on a offroad course in ure review.

Welcome to TPU!

Firs of all; dropping "shits" like "bullshit" and "stupid" in your first post might not be the best possible approach imho, but maybe I'm wrong.

It is true that running those tests today in resolutions where the games are all GPU limited tells nothing, but things will probably change in the future with games written for dx12, and then 4, 6 or 8 cores might make a difference after all.

Just my two cents.

 

[eliberate]

yes, but that's not the case now, and lets please look at the markets the two paltforms at aimed at.
For gamers u need max 4 cores and higher speeds, for working u need more cores and as much speed but more cores is better then less with more mhz.
Sry for droping sht on my first post but i hate it when wannabees endup writing missleading reviews. Based on this review some poor kid may spend 5-8k $ on a rig with 8 cores to play AC Unity at 4k, but a sli of 970's on a 1150 with a 4.5 clocked K is best solution and wont cost 5-8k.

PS. people improve based on the criticism they receive, i hope my "shit" will have some effect cuz that the only reason i"ve spread it

 

[Caring1]

my first post here
........
Ure testing a Lambo on a offroad course in ure review.

Welcome to the forums.
While I understand the gist of what you are saying, your grammar leaves a lot to be desired.
I do agree however that testing the 2011-3 platform for gamers should be done on a high clocked CPU(3.5GHz or higher), four cores would suffice, and a range of GPU's to show affects of bandwidth on FPS.
A 3.0GHz octocore CPU isn't going to beat a 3.5GHz quadcore CPU in games.

 

[1Kurgan1]

Just got my 5820k, MSI x99 SLI Plus, and 16GB DDR4 2400. Love it, just an awesome setup. I keep seeing that OCing is held back by the DDR4, but too be honest it doesn't seem that bad. I got mine running stable at 4.6 on 1.285v and I even bumped up the 2400 RAM to 2666, and it's awesome. I got to get my waterblock on my 290x, sad to say thats my bottleneck now.

 

[StartBeforeYouStop]

Nice review. Can't wait to see the performance from the upcoming skylakes!

 

[Ebo]

i really like my I7-5820K more and more for each day. It runs my tasks with out any foults and does it tasks with a blink of an eye.
I use my machine for a lot more than gaming and where the high bandwith on ram and the CPU power comes into play which have made my daily life quicker and a lot easier.

Gaming dosent really show any difference, or the difference is so small it dosent really matter. It just runs everything I through at it.

 

[Vlada011]

I still read about people who upgrade on X79 and Rampage 4 Black Edition.
If I see good 5820K is better than 4960X? Only 40 PCI-E lanes compare to 28 is advantage of 1000$ processor compare to 400$. Even L3 Cache and number of cores are same.

 

[EarthDog]

@MSnyder
Could you please create a cpu bottleneck when testing 3d games? For example, 800x600 and low setting.
Regards

This is good to see to really isolate the cpu performance differences in games, however that really stray from reality/typical resolutions people run when playing those same games so it can be pretty misleading at only those low resos..

What do you mean by this msnyder?

Considering the DDR4 DIMMs do not actually run at 3000 MHz, overclocking these processors was challenging. Also, luckily for me, I had the very capable and, moreover, reliable EVGA 1600W P2 power supply on hand, which instantly rid me of any instability issues due to a lack of power because of other peripherals.

What do you mean by "don't actually run at 3k" and what does that have to do with being more challenging to overclock? Is it because of the math of 125 bclk x multi as opposed to 100mHz?

Also, you have a 1.6kw psu, and talk about it like you need something close to that wattage? Why are you lucky when I would EASILY run the test system, any of the cpus at any ambient coooled overclock (cpu and gpu) with a quality 500w psu that wouldn't flinch... It's an incredibly Hugh quality psu and I would recommend it, however only to someone that can use the majority of its power. What you did to it, it had no.idea there was a load on it!

 

[Ebo]

I still read about people who upgrade on X79 and Rampage 4 Black Edition.
If I see good 5820K is better than 4960X? Only 40 PCI-E lanes compare to 28 is advantage of 1000$ processor compare to 400$. Even L3 Cache and number of cores are same.

The way I made my desission, it that I dont need more than 28 lanes to GFX, since I allways go for 1 big GFX for starters.
Even if you put a 2nd GFX in my machine, the real world difference in games, would be just about 5% running 16xpci-e 3.0 and 8xpci-e. Instead where I live, the price difference between I7-5820K og I7-5930K is simply too high to justify the gain Im getting.

 

[Vlada011]

I will buy i7-5820K this months too. I order GSkill from Germany, 2800MHz 4x4GB, model F4-2800C16Q-16GRK.
I didn't want to pay 300-330e for Kingston 3000MHz in Serbia, we can buy only one brand of memory last five year.
GSkill 2800 4x4GB is best ratio price performance for DDR4. Latency are not to high as 18, voltage is 1.2V, nice kit.
CORSAIR ask 350e, that's 130e more for higher latency and lower speed than GSkill, and inside I think in both is Hynix.
I still have some hope that I will see teaser for Rampage 5 Black or some special Rampage 5 some for 20th Years of Anniversary.
But because Anniversary is whole year ASUS should think and on GM200 6GB.
i7-5820K is best possible processor, last summer same performance cost 1000$.
Same number of cores, same L3 Cache as 4960X, only support newer 2133MHz DDR4 instead older 1866MHz DDR3 and motherboards have advantage of M.2.
I will not use SATA Express at all, Disable Immediately, first OS drive will be Samsung PCI-E Gen 3 256bit or Intel if present something good for OK price.
Until than I will use old Force GT 120 SATA III.

CORSAIR is really not fair. I have Dominator Platinum DDR3 and I would go again on Dominator but price is not normal and they not give warranty for spectacular OC...They use same chips as GSkill... Difference is not small

GSkill 2800MHz 4x4 (16-16-16-36-2N) 220e
CORSAIR 2666MHz 4x4 (16-18-18-35-2N) 310e
CORSAIR 2800MHz 4x4 (16-18-18-36 2N) 350e

That mean if you choose GSkill over Dominator Platinum you will have same speed tighter latency and free EKWB Rampage 5 Extreme Monoblock, or EK Supremacy + 240mm PE...
It's not fair so high price only for better look. Even you need to gamble for Version because ASUS test 5.29 Version and you can get completely something else maybe.

 

[Exceededgoku]

Would be nice to see this compared to the flagship AMD CPU's as well...

Or is this the way CPU reviews are going now with AMD's recent performance :S.

 

[CTMtech.net]

I upgraded from a quad core Intel Core i7 2600K purchased back in 2011 to the Core i7 5820k and the Asus X99-A motherboard. i was able to sell my old build and cover most of the cost of the new build, so it worked out pretty well. Definitely notice a difference when editing 4k footage (but I also upgraded my GPU as well to the Windforce G1 GTX970, which made a huge difference).

Here's an overview of my current Haswell build:

 

[Sunset1]

while I really like this type of review I was surprised that the 4930 was not included.. I bought mine right before the 5930 came out and regretted it at first but was glad I stayed with ddr3 for a while and bought up a rive black a lot cheaper..

it would be interesting to see the results
thanks again for quality work.
Sunset1

 

[radrok]

while I really like this type of review I was surprised that the 4930 was not included.. I bought mine right before the 5930 came out and regretted it at first but was glad I stayed with ddr3 for a while and bought up a rive black a lot cheaper..

it would be interesting to see the results
thanks again for quality work.
Sunset1

Your CPU is basically the same as a 5930, minor IPC gains aside that's it. DDR4 isn't worth the upgrade alone and everything else is just the same.

No differences to be seen except for small percentages on syntethics.

X99 is basically good only for people who have a quad core and want to step up to a six core or for people who want the 8 core, the 5930 has no point in existing.

 

[Sunset1]

thats what i was thinking, i basicly got in to benching and went buying way too much hardware..
then my health cought up with me. So now im finally getting time to test the hardware.. while i ignore those newegg ads.
btw thank god for the download section on techpowerup!
i have always found it to be a reliable place for info.
sunset1

 

[Ebo]

Your CPU is basically the same as a 5930, minor IPC gains aside that's it. DDR4 isn't worth the upgrade alone and everything else is just the same.

No differences to be seen except for small percentages on syntethics.

X99 is basically good only for people who have a quad core and want to step up to a six core or for people who want the 8 core, the 5930 has no point in existing.

I wont exactly give you right.

If you have a relatively highend system with 2 GFX in either SLI/Xfire, and a Nvme PCI-e SSD, 1 M2 xpress SSD then you need all the lanes that I7-5930 has.

 

[Moofachuka]

"The ACX 2.0 is a 6-slot, 3-phase motor"

That's a big ass cooler from EVGA

 

[radrok]

I wont exactly give you right.

If you have a relatively highend system with 2 GFX in either SLI/Xfire, and a Nvme PCI-e SSD, 1 M2 xpress SSD then you need all the lanes that I7-5930 has.

No way you will saturate the bus even with that configuration, assuming you aren't doing sequential file transfers while gaming just to bottleneck the thing for the sake of it.

Two GPUs can easily be fed by two 8x links with no performance hit, so 16 lanes on gpus and 4x + 4x on the storage.

Some of you guys really give too much importance to PCIe lanes, imo.

Besides if you have two GPUs and those storage solutions the 5960x is a much more plausible CPU for that system.

 

[EarthDog]

Performance hit is 1-2% with 2 cards (see tpu article). Not much, but more than negligible.

 

[radrok]

I'm pretty sure you'd never notice 1% unless benching, we're nitpicking here :O