6700K or 5820K?

[StrangerGuy]

Actually, the Sandy Bridge Core i5 2500K appears to hold more original value than the i7 2600K. (ie, i5 2500K and i7 2600K used prices are not that much different from each other considering the price difference when new was much greater).

I am not sure though if the price gap will close in the same way for the i5 6600K and the i7 6700K (re: unlike the Sandy Bridge processors, the Skylake i7 has much higher stock clocks).

P.S. Your comments about the new processors having TIM are interesting in light of future resale value.

The 2600K was only clocked 100MHz higher than 2500K. If it was been 4+ GHz at stock like the 4790K/6700K the resale prices would be a different story.

Looks like the X99 won by a landslide in the poll; seems like most agree marginal increase in ST performance doesn't compensate the potential of 50% more cores.

 

[MrTeal]

I'd probably go with the 6700k for a work computer, though it's close. For productivity, higher IPC and frequency scaling trumps more cores unfortunately. In my case Cinebench and encoding are not productivity apps; compiling and running CAD software is. Those are often terribly threaded, so single core performance is king. Altium for example doesn't scale past a single core.

If I was a graphic designer or similar, it'd probably be the 5960X.

 

[alcoholbob]

Considering a 5820K + Mobo is like $500 at retail stores right now with all the sales and 6700Ks are selling for $500-600 on marketplace sellers, I'd say bang for the buck goes to the X99 platform...

 

[CakeMonster]

The 5820K will need quite an OC from stock speed to even match the performance of the 6700K in ST performance though. Not sure what the average expected OC is.

 

[Blue_Max]

For YOUR needs, change my original vote to the 5820k instead. Voted before reading your needs.

 

[Lyfer]

If OCin 5820k no brainer. If not, 6700k IMO.

 

[Techhog]

Yeah but thats showing a overclocked 6700k vs a stock 5820k.. when both are overclocked the 5820K would rip the 6700K a new hole..

Both overclocked (4.7GHz) and stock (4.2GHz boost) are in every image...

 

[jim1976]

I bought 5820k and I'm running @4.2 for 24/7. I am stable at 4.5 for 24/7 with reasonable vcore but why push so hard?
At gaming it's completely irrelevant at 4k/1440p for me so I think it would be a wiser choice if someones wants/needs to upgrade now.
With DX12 and more cores in mind for future multithreaded apps, why would I care for such an insignificant increase in performance even for cpu limited resolutions up to 1080p?
The only real disadvantage compared to Skylake for me is the power requirements. Otherwise I think X99 is a more complete (albeit more pricey) platform atm for someone to upgrade.

 

[cmdrdredd]

Pfft ROG and 6700K. An Extreme 4 and 5820K would knock it down once you start hammering threads. What USB issues? Only X99 issues I saw/see are the wonky OC socket on Asus X99 murdering those poor hexa cores and wonky BIOS's on Gigabyte's X99 boards.

Some USB controllers share PCIe lanes with the slots on some boards so if you have multiple GPUs a soundcard and a PCIe SSD and want to use the USB 3.1 you're screwed since some PCIe slots will be shut off. It depends on the board.

 

[cbn]

I remember once I overclocked a bunch of AMD processors that I was using for computational chemistry and the results came out very unexpected. My professor was all sorts of skeptical, I was not...but my professor had no idea I was OC'ing the CPUs at the time.

Then I set the CPUs back to stock clocks and re-ran the calculations. Lo and behold the results came back very very different.

That day I learned that there is a reason processors are tested, validated, and binned for their operating parameters. Just because the damned thing doesn't crash when OC'ing doesn't mean it won't compute 1+1=3.

Aside from the overclocking you mention, I do have to wonder how the TIM (i7-6700K) vs. solder (i7-5820K) issue could affect binning degradation over time? This if the TIM under the heatspreader begins to dry out over many years.

P.S. It was mentioned in post #10 and #13 of this recent thread, that some AM2 Processors (which used TIM) in this day suffer from being too hot. One person in that thread claimed this is from the TIM under the heatspreader drying out.

 

[cmdrdredd]

If we dive deeper into various motherboards, we can see this isn't the case, yet what you said keeps being repeated on forums like gospel.

1) Scenario with a lower end X99 motherboard - For example Asrock X99 Extreme 4:

5820K has 28 lanes. That means on most motherboard, it can work as 16x/8x + M.2 PCIe 3.0 x4 configuration.

16x for the 1st GPU
8x for the 2nd GPU
4x for the PCIe 3.0 x4 M.2 SSD

"*If you install CPU with 28 lanes, PCIE1/PCIE3/PCIE5 will run at x16/x8/x4.

What happens when we insert the M.2 PCIe SSD? The 3rd PCIe slot gets disabled - but this is for the 3rd videocard so it's irrelevant.

**If M.2 PCI Express module is installed, PCIE5 slot will be disabled.
***If you install CPU with 28 lanes, 3-Way SLI™ is not supported.
****To support 3-Way SLI™, please install the CPU with 40 lanes."
http://www.asrock.com/mb/Intel/X99 Extreme4/?cat=Specifications

Going from PCIe 3.0 x16 to x8 is a 0-1% performance hit.

In this case, on a lower end X99 board, 5930K is only necessary when running 3 GPUs or 3GPUs + PCIe SSD or if running 2 GPUs + 2x PCIe SSDs, etc.

2) Scenario - higher end X99 motherboard such as Asrock X99 Professional:

It gets better, some higher end boards can do Tri-SLI/Tri-Fire + M.2 PCIe SSD with a 5820K.

"* If you install CPU with 28 lanes, PCIE1/PCIE2/PCIE3/PCIE4/PCIE5 will run at x16/x0/x4/x8/x0 or x8/x8/x4/x8/x0, and PCIE5 will be disabled.
** To support 3-Way CrossFireX™ and 3-Way SLI™ when using CPU with 28 lanes, please install VGA cards to PCIE1/PCIE2/PCIE4 (x8/x8/x8). * If Ultra M.2 PCI Express module is installed, PCIE3 slot will be disabled."

That means it's possible to run 5820K + GTX980 Tri-SLI in PCIe slots 1, 2, 4 and still have PCIe 3.0 x4 SSD.

http://www.asrock.com/mb/Intel/Fatal1ty X99 Professional3.1/

In conclusion, if someone is buying a 5930K over 5820K for dual GPUs and a single M.2. PCIe 3.0 x4 SSD, they are just wasting $ that could be used towards better cooling, monitor, games, etc. Since some boards can even support 8x/8x/8x + x4 on a 5820K, it requires additional research to be able to discount the 5820K as well.

So in theory on this board you can run 8x/8x/PCIe3 off/8x/PCIe5 off and have 2x GPUs (in slots 1 and 2), a PCIe 3.0 x4 M2 SSD installed(disables slot 3), and still use PCIe slot 4 for a dedicated sound card? That's the setup I'm looking for I think.

 

[Nec_V20]

With Skylake reviews out, I'm interested in seeing which setup everyone here finds to be the better buy for their purposes: an X99 board + 5820K or a Z170 board + 6700K.

Please vote in the poll and, if you're so inclined, include an explanation in a post!

I am sure and my vote would be neither.

I've had my X58 i7-990x for five years now.

I do not see much difference from the Z97X i7-4770K I am running now.

If it ain't broke don't fix it.

The only thing the newer Intel processors give me is that I can cadge a 4790K cheaper and get my temps down with regard to any OC.

Aside from that, who needs it?

 

[bigboxes]

If I had went the X99 route I would have gone 5930, not 5820. So, the cost to me would be at least $500 more. I admit that it's just me. And this is at the time when DDR4 was prohibitively high. I'm pretty happy with my 4790K. I've got all four-cores clocked at an easy 4.4Ghz. I can o/c to 4.7 and it still runs acceptably cool. I even smile when I'm gaming.

 

[Nec_V20]

If I had went the X99 route I would have gone 5930, not 5820. So, the cost to me would be at least $500 more. I admit that it's just me. And this is at the time when DDR4 was prohibitively high. I'm pretty happy with my 4790K. I've got all four-cores clocked at an easy 4.4Ghz. I can o/c to 4.7 and it still runs acceptably cool. I even smile when I'm gaming.

I agree with you, for the games I like to play my i7-4770K just seems to be more comfortable, but not really all that much difference from my i7-990x. There is a difference but it just feels like my 4770K is a bit more relaxed, if that makes sense to you.

Why move on?

I thought that with the pressure I had been putting on my system, after four years it might be coming to an end of life situation and I wanted a replacement in place (I saw a really good deal on a 4770K as well).

So I thought, "Why not".

I am still using my 990X as my main system.

 

[bigboxes]

I am still using my 990X as my main system.

Nothing wrong with that. It's a great platform. I recently retired my 920. I am planning to use it to build a new HTPC. Not that I need that much horsepower to do that task. I just like the thought that it would handle whatever I throw at it. That's my goal.

 

[Nec_V20]

Nothing wrong with that. It's a great platform. I recently retired my 920. I am planning to use it to build a new HTPC. Not that I need that much horsepower to do that task. I just like the thought that it would handle whatever I throw at it. That's my goal.

Seriously now after five year (I have had my 4770K for a year and just upgraded the X78X mobo to a Z79X mobo) I am still waiting for it to go south.

On the other hand, there are two universal laws

1) Murphy's Law (Sods Law) "If something can go wrong it will"

2) Hanlon's Razor "Never attribute to malice that which is adequately explained by stupidity."

With regard to 1), I knew that if I did not have a replacement then Murphy's Law would kick in at some point, so having a replacement meant that Murphy's Law would be negated.

 

[Sweepr]

Edit: Delete it.

 

[tential]

I agree with you, for the games I like to play my i7-4770K just seems to be more comfortable, but not really all that much difference from my i7-990x. There is a difference but it just feels like my 4770K is a bit more relaxed, if that makes sense to you.

Why move on?

I thought that with the pressure I had been putting on my system, after four years it might be coming to an end of life situation and I wanted a replacement in place (I saw a really good deal on a 4770K as well).

So I thought, "Why not".

I am still using my 990X as my main system.

I honestly can't imagine my 4770k needing to be upgraded in a long time.

 

[CakeMonster]

What is the worst case OC that can be expected on 5820K on air?

 

[froggermuted]

What is the worst case OC that can be expected on 5820K on air?

Depend on the case/air flow, heatsink and room temp... You should be able to get at least 4.0 from what I've read, on a normal setup.

 

[Arachnotronic]

What is the worst case OC that can be expected on 5820K on air?

I'd be shocked if you couldn't get 4.0GHz. The thing turbos to 3.6GHz which means all the cores should be good to run at 3.6GHz. A 400MHz bump, especially at this stage of the game (a design that has been in production for about a year on a manufacturing process that has been in production for ~3 years) seems more than doable.

 

[alcoholbob]

I'd be shocked if you couldn't get 4.0GHz. The thing turbos to 3.6GHz which means all the cores should be good to run at 3.6GHz. A 400MHz bump, especially at this stage of the game (a design that has been in production for about a year on a manufacturing process that has been in production for ~3 years) seems more than doable.

I'm gonna bet at least 90% of 5820Ks can hit 4.4GHz by 1.35v. Mine does 4.5 @ 1.25 and 4.6 @ 1.28 and according to the Anandtech review those are considered average-to-good overclocking numbers.

 

[Arachnotronic]

I'm gonna bet at least 90% of 5820Ks can hit 4.4GHz by 1.35v. Mine does 4.5 @ 1.25 and 4.6 @ 1.28 and according to the Anandtech review those are considered average-to-good overclocking numbers.

Very interesting!

 

[JM Popaleetus]

Does Haswell-E scale with memory speeds as dramatically as Skylake does?

I'm returning the kit in my sig due to errors, and because of Skylake's release, DDR4 prices have fallen again. So I'm pondering going with DDR4-2800 or 3000.

And yes, I have read Anand's article. But I just want a second opinion considering some might argue they butchered the Skylake tests.

I'm gonna bet at least 90% of 5820Ks can hit 4.4GHz by 1.35v. Mine does 4.5 @ 1.25 and 4.6 @ 1.28 and according to the Anandtech review those are considered average-to-good overclocking numbers.

Yours doing 4.6 at 1.28 is a pretty golden chip.

 

[YBS1]

No, it's not bandwidth starved being quad channel. It helps but nothing like Skylake is showing.

Edit: I should temper that statement by saying I've never ran below 2666, so maybe the jump from 2133 to 2666 I'd rather large, I don't know.