Core i7-5960X 5930K 5820K Overclocking & Performance @ [H]

this CPUs doesn't clock well, I have seen a lot of them and if you push that hard they need a lot of voltage.
many kids says 4.5GHz with 1.25V, don't trust them :)

I think the lowest voltage I've seen on a 5960X that was stable is 1.28v. Most seem to require 1.3v or higher.
 
I think I found a 5820K that can actually overclock. My previous one that I got back in December couldn't even do 4.2GHz 1.3v. If I ran Prime95 or y-cruncher it would error or BSOD within an hour. I thought it was the board as it has issues booting at times but now that I got another CPU to test I can now confirm that the previous one sucked ass.

Here's it @ 4.3GHz 1.2v, stable with y-cruncher.



Currently testing 4.4GHz, 1.2v.
 
I can do 4.5 Ghz with 1.3 volts. Good result?
Temperatures can go as high as 80 C with my H80. I have a brand new Corsair H110i that I will install over the weekend so it may drop temperatures.
 
I think the lowest voltage I've seen on a 5960X that was stable is 1.28v. Most seem to require 1.3v or higher.

I've been able to OC my 5960x @ 4.5ghz with voltage of 1.22 under load using Aida64 for 8 hours straight overnight stable between 55c - 70c. I have a Corsair H110i GT and replaced the fans that came with it with 2 Noctua NF-A14 @ 2000 RPMs (maybe too noisy for some but it doesn't bother me). I am working on OCing the uncore now.
 
So I installed the Corsair H110i GT on my processor and now I can do 4.6 GHz. It idles at about 35-38 C during browsing, watching movie etc. It has temperatures of around 65 C during gaming. I did a Real Bench stress test and it topped out at 82 C on Core 2 (which seems to run hotter than other cores) and 70-75 C on rest of the cores.

I am giving it 1.32 volts right now. Do you guys think pushing 1.35 volts through the chip to get 4.8 GHz is a good idea?
 
If you can do that without it throttling I think it is fine. You might look at bumping VCCIO on CPU Input voltage a tiny bit before pushing the core voltage higher. I have instances where variations on those two other values have actually allowed me to drop core voltage giving some better overall temperatures.

You got a great piece of silicon there too.
 
KickAssCop

How does it do with AID64 and OCCT? My 5930K has two hot cores. My cooling keeps them in the low to mid 70c range. With the other 4 cores staying in the low to mid 60c range. I'm lucky that I don't have to volt the cpu up much. And after 4.8 ghz no amount of voltage will make it run faster.

Kyle's suggestions are very good ones. Wish they worked with my cpu. I think my two hot cores just throw the cpu into a panic at some point. :)
 
KickAssCop

How does it do with AID64 and OCCT? My 5930K has two hot cores. My cooling keeps them in the low to mid 70c range. With the other 4 cores staying in the low to mid 60c range. I'm lucky that I don't have to volt the cpu up much. And after 4.8 ghz no amount of voltage will make it run faster.

Kyle's suggestions are very good ones. Wish they worked with my cpu. I think my two hot cores just throw the cpu into a panic at some point. :)

it's is very incredible. 4.8GHz on HEDT are very very very rare, I really doubt you can get that frequency if not for a screenshot.
 
it's is very incredible. 4.8GHz on HEDT are very very very rare, I really doubt you can get that frequency if not for a screenshot.

Sorry for the quality. The info is there. Download and blow up if needed. I stopped the run at a little over 17 min. because my office ambient temp hit 85F. Anymore doubter questions?
Screenshot_20_.png


The Rig.
Define_S_Start.jpg


:p
 
Last edited:
Tried 1.32 and 1.92 input and real bench failed at 4.7. Does it matter to do 125 bclk and 38 multi instead of the 100 and 47 that I used?
 
Sorry for the quality. The info is there. Download and blow up if needed. I stopped the run at a little over 17 min. because my office ambient temp hit 85F. Anymore doubter questions?
Screenshot_20_.png


The Rig.
Define_S_Start.jpg


:p

ok you are lucky, congrats. very very very rare cpu.
 
I have an X99s MSI Gaming 7 motherboard w/5930K CPU, and Corsair DDR4 2800 memory. I only really tried out my first O/C last night because I found the process somewhat confusing.

Months ago since got the system, I've always enabled XMP in BIOS (not the OC Genie - just XMP)

Thist set my ram to it's rated spec of 2800Mhz. (bios repots 2801mhz actually but who's counting)

Doing that also sets my CPU to 3820Mhz, and sets BCLK 127.03. My multiplier getsset to 30.

What I didn't realize is that I can modify that multiplier. I thought it was locked when XMP was enabled -- clicking on it does nothing.....but page up page down lets you change it.
Stupid me.

So I set Multiplier to 34. BCKL is still 127.03 and the chip is now @4327Mhz.

** Edit: Turns out EIST is disabled, as is Turbo. My mistake.

Voltage - I didn't change a thing with voltage. Whatever the bios does automatically, is what the voltage is. I will confirm it with CPUZ when I get home.

Ran Firestrike and scored almost 800 points higher with this overclock, compared to just enabling XMP alone and not upping the multiplier. Score was something like 17,500 ish. Video card is a MSI 980 Ti Gaming (also OC).
http://www.3dmark.com/fs/5557809 (after OC)
http://www.3dmark.com/fs/5557349 (pre OC)

I am very impressed with it. Just discovered this thread cause I wanted to see how other people were doing. I'm using air cooling, and idle chip temps still hover around 33 degrees or so. During games (Dragon Age, Dayz) CPU temps are maxing around 60 degrees Celsius.

Case (fractal design R90) has decent air flow,stock 144mfan rear exhaust, two 120mm noctua ceiling exhaust, stock + noctua140mm front intakes.

I will check my voltages tonight and report back, in case anyone is interested.
 
Last edited:
I have an X99s MSI Gaming 7 motherboard w/5930K CPU, and Corsair DDR4 2800 memory. I only really tried out my first O/C last night because I found the process somewhat confusing.

Months ago since got the system, I've always enabled XMP in BIOS (not the OC Genie - just XMP)

Thist set my ram to it's rated spec of 2800Mhz. (bios repots 2801mhz actually but who's counting)

Doing that also sets my CPU to 3820Mhz, and sets BCLK 127.03. My multiplier getsset to 30.

What I didn't realize is that I can modify that multiplier. I thought it was locked when XMP was enabled -- clicking on it does nothing.....but page up page down lets you change it.
Stupid me.

So I set Multiplier to 34. BCKL is still 127.03 and the chip is now @4327Mhz.

** Edit: Turns out EIST is disabled, as is Turbo. My mistake.

Voltage - I didn't change a thing with voltage. Whatever the bios does automatically, is what the voltage is. I will confirm it with CPUZ when I get home.

Ran Firestrike and scored almost 800 points higher with this overclock, compared to just enabling XMP alone and not upping the multiplier. Score was something like 17,500 ish. Video card is a MSI 980 Ti Gaming (also OC).
http://www.3dmark.com/fs/5557809 (after OC)
http://www.3dmark.com/fs/5557349 (pre OC)

I am very impressed with it. Just discovered this thread cause I wanted to see how other people were doing. I'm using air cooling, and idle chip temps still hover around 33 degrees or so. During games (Dragon Age, Dayz) CPU temps are maxing around 60 degrees Celsius.

Case (fractal design R90) has decent air flow,stock 144mfan rear exhaust, two 120mm noctua ceiling exhaust, stock + noctua140mm front intakes.

I will check my voltages tonight and report back, in case anyone is interested.

There is no magic to overclocking Intel CPUs anymore. Essentially CPU input voltage and CPU vCore are basically it for the CPU. I've only worked with 5960X CPUs so I can't comment on 5820's too much. I think I used one of those once.

A good 5960X can hit 4.5GHz around 1.32v. Some exceptional CPUs like mine can do it at 1.30v or less. That same 5960X on the other hand may not work as well with higher memory frequencies or when loaded with 8x DIMMs.

A mediocre 5960X may be able to do 4.5GHz but will require 1.35v or more to do it. Thus temperatures will run very high and the system will only attain Prime95 stability for short durations. Anything beyond 1 hour or so can create "heat soak" in which the temperatures will just slightly out pace the cooling system's ability to dissipate the heat. So your temps may start out in the low 80c range and before you know it your throttling or worse because the temps have gone well into the upper 90c range. Some of these chips could have a weak IMC that doesn't handle loading all 8x DIMM slots well and won't run DDR4 RAM at higher frequencies. My 5960X ES (Engineering Sample) is like this. There is some variance based on the motherboard, but not much. Some motherboards with better chipset cooling will allow for Prime95 stability at 4.5GHz for longer periods than others.

A bad 5960X may not be able to do 4.3GHz without a lot of voltage. I've heard of a few of these dogs and worse.

Occasionally system agent voltage matters on X99 for some memory configurations. VCCIO also has an impact in some cases as well. This is true of MSI's XPower motherboards which need at least a VCCIO of 1.1-1.2v. What is interesting is that MSI needs these values adjusted in order to run stable at stock speeds in some cases regardless of the actual RAM speeds used. I've also found that manipulation of this value has an impact on X99 systems when tuning memory, especially when going beyond 4x DIMMs. To run 8x DIMMs on a Rampage V Extreme I had to increase the SA voltage and VCCIO voltages, as well as adjust the DRAM power phase control from "standard" to "optimized."
 
Been screwing around with the overclock on my 5820k and asrock formula oc some more...

man I just don't have the patience to finetune voltages anymore using offsets and LLC... sheesh. Fortunately this board has a metric ass ton of overclocking options, allowing me to be lazy and set my vcore at 1.3, input at 1.9500, disable all c-states and just run it at full voltage and full clockspeed all the time...

Been running it about a month now 4.5 ghz, still 3.3 cache.... with 1.3v to the core. Older versions of Prime95 will run for hours with no problem, v28.5 bluescreens instantly...gaming and benchmarking have no issue either. Seeing max temps in prime 65-80..

Kill-a-watt says 460w during prime....hah.


Eventually i'll spend the needed time to tweak it out for efficiency and power...

The little time I have spent trying to get my ram up beyond the intel rated speed of 2100 has failed fantastically. Jumping from my old sandybridge setups to haswell-e has still got me confused on a few things.
 
I agree with this but if your PC isn't stabe with Prime 95 27.9 your PC isn't stable.

That's not always the case. I've seen some systems that were Prime95 stable for 24 hours that died immediately when you fired up Handbrake and started encoding something.
 
That's not always the case. I've seen some systems that were Prime95 stable for 24 hours that died immediately when you fired up Handbrake and started encoding something.

Prime is not what it use to be. AID and OCCT are much better IMHO.
 
I've seen Prime95 fail on systems that can run AIDA all day. I've never fully trusted it.
 
Been screwing around with the overclock on my 5820k and asrock formula oc some more...

man I just don't have the patience to finetune voltages anymore using offsets and LLC... sheesh. Fortunately this board has a metric ass ton of overclocking options, allowing me to be lazy and set my vcore at 1.3, input at 1.9500, disable all c-states and just run it at full voltage and full clockspeed all the time...

Been running it about a month now 4.5 ghz, still 3.3 cache.... with 1.3v to the core. Older versions of Prime95 will run for hours with no problem, v28.5 bluescreens instantly...gaming and benchmarking have no issue either. Seeing max temps in prime 65-80..

Kill-a-watt says 460w during prime....hah.


Eventually i'll spend the needed time to tweak it out for efficiency and power...

The little time I have spent trying to get my ram up beyond the intel rated speed of 2100 has failed fantastically. Jumping from my old sandybridge setups to haswell-e has still got me confused on a few things.

I'm with yeah. Just jumped from Sandy to a 5820k the past week. Havn't gotten past turning up the multiplier and VCore numbers. I'm at 4.5ghz on 1.275 V at the moment. 4.6 and above doesn't seem to fly no matter how many volts I give it. Although I havn't tried running the memory at default speed yet and running tests. Bought 3200 RAM and it's running as advertised with 1.4 volts. Asus X-99 Deluxe fwiw.
 
Been screwing around with the overclock on my 5820k and asrock formula oc some more...
Been running it about a month now 4.5 ghz, still 3.3 cache.... with 1.3v to the core.

I'm with yeah. Just jumped from Sandy to a 5820k the past week. Havn't gotten past turning up the multiplier and VCore numbers. I'm at 4.5ghz on 1.275 V at the moment.

Very nice Guys.
 
Last edited:
I've seen Prime95 fail on systems that can run AIDA all day. I've never fully trusted it.

and none of those can be trusted completely or fully reliable, Asus RealBench it's way better for stability test and benchmarking. specially since Prime95 isn't even recommended for X99 Chips.
 
Last edited:
I'm with yeah. Just jumped from Sandy to a 5820k the past week. Havn't gotten past turning up the multiplier and VCore numbers. I'm at 4.5ghz on 1.275 V at the moment. 4.6 and above doesn't seem to fly no matter how many volts I give it. Although I havn't tried running the memory at default speed yet and running tests. Bought 3200 RAM and it's running as advertised with 1.4 volts. Asus X-99 Deluxe fwiw.

What are you using for cooling, and what are your temps like?
 
That's not always the case. I've seen some systems that were Prime95 stable for 24 hours that died immediately when you fired up Handbrake and started encoding something.

That's always been the case. You can never "just" use one test to verify an overclock, not if you care about stability. In fact, the best stress test is to run a video benchmark loop in the background WHILE YOU RUN MULTIPLE SIMULTANEOUS tests.

We don't run stress test to feel good or brag, we run stress test to catch those cases where the processor breaks our data. If the processor breaks our data under less observant/stringent circumstances, we probably won't notice. And that's a bad thing, when so much of our data is important to us :D

If your processor can pass X test at stock, but it can't pass X test at overclock, then it's NOT STABLE.
 
Last edited:
Reading some of these clocking numbers I may just wait until Skylake-E to make the jump. Hope the node change makes enough of a difference...
 
Should have my 5820k heatsync (Thermalright True Spirit 140 BW Rev. A) this weekend. If I'm lucky I'll be able to get 4.2-4.4GHz out of it. At least that's what I hope anyways. Shame my Ivy Bridge motherboard is dying or I might have made it to Skylake-E. Oh well...
 
5820k at 4.3GHz stable OC 1.249vcore H100i cooler benchmarks testing temps where around anywhere between 60-75c
 
Those are very high temperatures. My highest core hits around 58-60C and all other cores hover around 55 C during heavy benching with an H110i GT.
 
Those are very high temperatures. My highest core hits around 58-60C and all other cores hover around 55 C during heavy benching with an H110i GT.

don't talk about this as a normal case, this CPUs are hot as hell,
what is your vcore/vccin to be around 55c on max load? what is your max load, a game?
 
5820K on ASUS X99-A mobo
4.6Ghz at 1.297V (currently 1 hour of Aida64)
60°c, sometimes jumping to 70°, but they are all over the place. They fluctuate 5° multiple times each second.

Memory at standard 2400Mhz (G-skill)
 
5820K on ASUS X99-A mobo
4.6Ghz at 1.297V (currently 1 hour of Aida64)
60°c, sometimes jumping to 70°, but they are all over the place. They fluctuate 5° multiple times each second.

Ambient temps? Cooler used? ...
 
don't talk about this as a normal case, this CPUs are hot as hell,
what is your vcore/vccin to be around 55c on max load? what is your max load, a game?
Max load is real bench.
My vcore is 1.32 and input voltage is 1.9.

It might be because my 110i GT is using 2 140mm and 2 200mm for push/pull.
 
Ambient temps? Cooler used? ...

Ambient arounnd 22°C, Watercooled with custom loop. (kinda custom, still need to add video card with waterblock and an extra radiator)
I only tried for 10 minutes. 3 reboots to get it at 4.6 at 1.297 (was stable at 1.3 first time, not stable at 1.293, then went to a stable 1.297).
First time overclocking btw. Will also overclock 3 other identical systems (that are air cooled as the only difference), but as they need to be up 24/7 for the next 5 years I will probably aim for a 4.2 Ghz at the lowest voltage I can.
 
Last edited:
Max load is real bench.
My vcore is 1.32 and input voltage is 1.9.

It might be because my 110i GT is using 2 140mm and 2 200mm for push/pull.

you have problem reading temperature or you don't know a software to load the CPU enough :)
trust me with an AIO at 1.32V this CPU can't do 55°c maximum.
 
I am really confused for which one to go for the Haswell-E or the Skylake 6600K setup. In terms of memory courtesy of Hardocp benchmarks of skylake are great. Im assuming DDR4 makes a difference. That is why i want my secondary system to be with DDR4 memory
 
I am really confused for which one to go for the Haswell-E or the Skylake 6600K setup. In terms of memory courtesy of Hardocp benchmarks of skylake are great. Im assuming DDR4 makes a difference. That is why i want my secondary system to be with DDR4 memory


Did you mean 6700K?

comparing a 6 core I7 to a 4 core I5 doesn't make a ton of sense

If all you do is game then the 6600K with a high end GPU will be fine. I chose the 6700k and I am happy with that, but at the same time I am also sure that if I had chosen the 5820K I would have been just as happy. I have used both for gaming and for general PC stuff. I was unable to tell the difference.

People get too caught up in metrics, it is VERY hard to measure how fast something feels. Just like all the people with 2500s claiming they are just as fast and do not seem slow at all, there is a good chance if they played the same games on a 6700k or a 5820k they would notice it feels smoother.
 
Did you mean 6700K?

comparing a 6 core I7 to a 4 core I5 doesn't make a ton of sense

If all you do is game then the 6600K with a high end GPU will be fine. I chose the 6700k and I am happy with that, but at the same time I am also sure that if I had chosen the 5820K I would have been just as happy. I have used both for gaming and for general PC stuff. I was unable to tell the difference.

People get too caught up in metrics, it is VERY hard to measure how fast something feels. Just like all the people with 2500s claiming they are just as fast and do not seem slow at all, there is a good chance if they played the same games on a 6700k or a 5820k they would notice it feels smoother.

so for what setup i should go for????? only the 5820 has a quad channel ddr4 setup and by benchmarks it seems ddr4 makes difference.

and intel makes confusing serial numbers...the 2nd gen was in 2serial numbers 3rd was in 3serial numbe for 4th it was 4serials. then Haswell-E was in 5serial and 5th gen processors also had 5serials ending only in laptops.....where is the desktop 5th gen????? and now its already skylake the 6th gen 6serial code...
 
so for what setup i should go for????? only the 5820 has a quad channel ddr4 setup and by benchmarks it seems ddr4 makes difference.

and intel makes confusing serial numbers...the 2nd gen was in 2serial numbers 3rd was in 3serial numbe for 4th it was 4serials. then Haswell-E was in 5serial and 5th gen processors also had 5serials ending only in laptops.....where is the desktop 5th gen????? and now its already skylake the 6th gen 6serial code...

you have a lot of confusion there..

Extreme Processor Series X79 and X99 (or truly high end socket 2011 or 2011v3) have different serials, for example sandy bridge-E (second generation) was 3820/3930K/3960X, Ivy Bridge-E(third generation) 4820K/4930K/4960X, Haswell-E(fourth generation) 5820K/5930K/5960X they are kinda familiar between gens.. we have a fifth generation which its broadwell and their desktop parts are only mainstrean i7 5775C and i5 5675C as examples... haswell-E isn't fifth generation, is fourth as the name haswell say. every intel extreme platform have a -E at the end of the name to differentiate mainstream products from high end, also the highest end CPU tend to feature a X at the end of the name so i7 980X, i7 3960X, i7 4960X, i7 5960X.

in the other hand, what are you really planing to do with your setup? if its gaming only then you can go with the Skylake i7 6700K (4 cores/8 threads) it have the same DDR4 as Haswell-E and certainly it will make little to none difference in real world gaming..

if you want to do some gaming and also a lot of heavy CPU work then you can go with the more versatile i7 5820K or i7 5930K with their 6 cores and 12 threads it can make significant job in Multi-thread CPU work, but it will offer lower gaming performance out of the box than the 6700K because of the lower clocks than this last one mentioned making it almost mandatory overclock the 5820K/5930K to receive the same single thread performance...
 
E processors seem to defy tech gravity.
I built a 5820k based system this time last year, CPU cost me £285.
Same vendor now charges £330 for it!
 
Back
Top