• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

i7-5960X mild overclocking

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

microfire

Member
Joined
Oct 8, 2001
Doesn't seem to much here on the i7-5960X processor (maybe all the old people here have gotten old, not doing it anymore), figured would start a thread specifically about the i7-5960X and overclocking on air. This is all I have to overclock with at the moment, which is a Noctua NH-D14, trusty and rusty but still pretty much at the top of its game for air. I have found that can overclock the i7-5960X reasonably well on air as long at the voltage stay pretty low. Found that if I start getting to 1.1v then the processor heat will get out of control pretty fast if all cores are loaded. More so if the clock speed is raised up. In the past with older Intel processors raising the clock speed and holding a certain low voltage didn't impact to much on heat, with Haswell-E this seems to have changed.

Personally I only purchased this processor as my budget would allow it. Saved on cost of RAM and put that towards an 8 core CPU, rather than chasing a 6 core overclocking and really never getting there in the end. Besides I still have a 6 core that I use in another building/room.

At some stage I will make the move to water cooling, although if I can get 4GHz stable on air without the heat running out of control, I might not ever bother for the sake of a couple of hundred MHz. I am not looking to break any records, just want to have a nice stable usable system to enjoy.

Right now I am at 1.02v with 3.6 GHz as 24/7 overclock. Sounds pretty lame, but its super quiet. Just looking for the sweet spot a little higher, maybe if I can closer to 4.0 GHz on air. I have my cheap RAM running at 1.35v for 2666 MHz and timings of 15-15-15-32 1T, and no worries about overheating there.

Like to hear from others that are running the same processor at mild speeds without comprising stability, and are not concerned with crazy cooling methods. I will try and keep this updated as much as possible to outline what can be done on a more minimal scale with the i7-5960X

This is all I have got what I believe is 100% stable at this stage (early days). Main goal this time around to run the CPU with all the power savings on whilst obtaining highest reliable overclock within reason. Before getting this CPU I use to turn off all power saving and run at max voltage and speed, continuously even while doing next to nothing, e.g. idle, downloading, walk away for half and hour at random. Want this overclock to be more green on power, but have plenty of power when needed.

DDR4_2666_CL14_1_35v.jpg
 
Last edited:
You can probably run it @4GHZ ~1.15V without issues on air. These CPUs are not really running hot.

There is not many 5960X users just because not many wish to spend $1k on a CPU that they won't use for any game or generally anything else at home. Scalling above 4 cores in games isn't really good and all those who need CPU for work will pick Xeons. 5820K isn't much slower in most applications while it costs nearly 1/3 of the 5960x.

X99 is generally selling really bad no matter what you see on the forums.
 
I think you are about right with 1.15v, the problem is maintaining that speed. In a real world there is no need to running synthetic load applications, but I do want to know where the limit is. Prime95 is old skool and said best not to run it anymore, but it does show up how much heat can be generated for a certain voltage and speed.

Issue I have, well I don't think it is possible to run this processor at 4GHz with air, synthetic load pushes this to near 100c, way to high for my liking. After I took this screenshot the temp was about 98c. Only took about 3 minutes to get to this temp. I can understand this is not a normal load and nothing really will push this high, but then again I never had any issues testing a 6 core i7-4930K using the same method with the same cooler. Although the 8 core is a big beast with 2 extra cores to run and additional 8MB of cache to drive compared to my old 6 core.

Really I think this CPU is a bad choice unless you have something hard-core to cool it. Obviously high-end air cooling cant do anything to tame it. I would think any pre-built water kit would also be out of the question. Look like would need some kind of crazy custom water cooling. I was optimistic about getting at least 4GHz with air, but I think all hope is lost.

Just goes to show, it would likely be better to clock a i7-5820K at higher core, and still match (if not better it) the i7-5960X at a lower core speed. This would be for anyone thinking about going for air cooled system, there are many valid reasons why air would be preferred over water, depending on the instance and usage.

These 8 core CPU really do run hot, it going to take some time and manufacturing cycles before these things can be tamed for general usage. I was targeting 4GHz to be a mild overclock for the i7-5960X on air, guess it is not to be.

Just 0.1v and 400MHz adds about 30c to my air cooling, which see the Noctua NH-D14 unable to cope. And I am not talking about adding 0.1v to 1.2 - 1.3v, this is just adding 0.1v to 1.0v for max of a relatively tiny 1.1v under load!

X99_4000_3600_2666_CL14_1_1v_Air.jpg
 
Last edited:
I was able to get to a max of 3.9GHz stable, when tested with OCCT on air, this is in a closed case. I gave up on the idea of using air, just wanted that bit extra out of this CPU due to the extreme price tag. Ended up buying a Kraken X61 and a high air flow case Corsair 540.

Now running 24/7 speed of 4.3GHz with 1.2v in BIOS

Able to bench at either 4.5-4.6GHz

This is 1:1 core/cache:

R15_4500_4500_2666_CL14_1818.jpg
 
How high you have to set cache voltage for 4500MHz ? Looks like 5960X are better than 5820/5930K. I can't make more than 4200MHz stable and about 4400-4500 at 1.45V+ for benchmarks on my 5820K ( actually on 2 that I was testing ). For 4200MHz my cpu needs ~1.3V cache voltage while couple of 5960X users can make it run at 4500MHz ~1.35V pretty stable.
 
Cache at 4.5GHz was set by Asus auto mode, which pushed it up to a horrible 1.45v. I have the cache set on offset and peaks at 1.29v. The cache running higher voltage was key to running lower vcore of 1.2v for my 5960X. I run 1:1 on core/cache both at 4.3GHz stable everyday.
 
I haven't gotten much time to research and tweak my 5960/Asrock OC yet.
But I did see how low I could go in voltage to keep my 4.5 oc.


x99.jpg
 
I am currently running my 5960x with 1.3v @ 4.4 core and I didn't touch cache voltage so far (didn't have a lot of time to mess up with the BIOS settings).

It is quite a good spot (~60C full load with fans at <=1200rpm), but it seems you need a lot more for 4.5 (like 1.475v)
 
OnDborder: I cant see the voltage for that 4.5GHz but if that is 1.23v then you are doing very well and likely above average chip.

Hippogriff: 1.3v for just 4.4GHz is looking a lot like what I needed, with huge volts for 4.5GHz. Reason for me the sweet spot is 4.3GHz and 1.2v to keep heat and power consumption under control, really is a sensible overclock to run everyday at the very small drop of just a hundred MHz
 
Voltage steps are pain in haswells-e. Most of them can make 4.2GHz ~1.2V but 4.3-4.4 at 1.3V and 4.5GHz is already 1.4V. I saw couple of 5960X running at 4.7GHz ~1.35-1.4V but not many.
 
Voltage steps are pain in haswells-e. Most of them can make 4.2GHz ~1.2V but 4.3-4.4 at 1.3V and 4.5GHz is already 1.4V. I saw couple of 5960X running at 4.7GHz ~1.35-1.4V but not many.

I probably can run at 4.4 even less than the 1.304v I'm using now. Tested yesterday @ 1.275v and XTU gave me an error after 1 hour (no bsod, just an error).

But to run 4.5 all cores stable, 1.4v is not enough. Not even 1.45v. Dunno what is the problem...
 
I am in the middle of doing some Linpack testing. I just finished my third mount of the NH-D14. I found a 1.5-2c difference between keeping the fan tops at the same height as the heat pipes vs fans whose tops are 1.5cm above the top of the fin stacks. In the latter case we are getting more of the meat of the fan on the whole of the fin stack.

My bottom line used to be to get low profile RAM. Now it is to mount the fans as high as the case will allow, to get better cooling.
 
Did you get a bad one? I just have just have my overclock to 4.0 Ghz. It will go faster but I leave it like it is. I do a lot of video editing on it.
 
After 4GHz you will start needing more voltage, it starts to get to the point where the power consumption for the amount speed gained is disproportional and not worth the extra power requirements and cooling to keep it under control.
I have since backed off my overclock to 4GHz at just 1.1v, and I have changed my fans on the Karken X61 that I use to 2x 140 Noctua's which spin very slow. The entire PC is very quiet.

btw. I can still run 4.3GHz anytime I want it to as long as I want to, if I feel like it.
 
I would try to OC the ram by getting something @ 2666 and then tighten the timings to 2133 with a much lower CAS RATE like 12 or 11 as i feel 2133 is still the sweet spot for X79 and X99 Platforms.

It would be interesting to see what the overall results would be??

Ajay.
 
2133 is a sweet spot for gaming and general performance, sure.

But why pay a premium for 2666 to tighten down to 2133 MHz? That doesn't make much sense to me. Also, at 2133 MHz, you can buy NEW sticks, DDR3, at CL9 anyway. So 11/12 is incredibly loose for 2133 timings for DDR3.

DDR4 is a different story...which is what we should be talking about exclusively here as X99 does not support DDR3 in the first place. Id get 2666 and see how much I can tighten things down, then bother to lower speed. Use the speed DDR4 gives you!!!

Not sure its common to tighten down 2666 to CL11 from CL16 though, is it? I mean I have seen Woomack do it, but he is a 1%er...not a common folk. LOL! :)
 
Sorry for my poor explanation what i have learned with OCing Ram is that you can lower the CAS rate on a higher speed set of ram sticks and still have the same sort of performance overall.

I.E. 2400Mhz @ C15 would be similar to 2133Mhz @ C12 if you understand what i am trying hard to say! So since most DDR4 has a high CAS RATE it would be interesting to see what a set of ram set to CAS 10 @ 2133Mhz would be capable of doing. :shrug:

Ajay.

P.S. I hope you know what i mean ED! ;)
 
I understand, just disagree with that methodology for DDR4 is all. I would get 2666 and tighten them down as far as I could then lowering speeds to DDR3 levels...

So since most DDR4 has a high CAS RATE it would be interesting to see what a set of ram set to CAS 10 @ 2133Mhz would be capable of doing.
There are plenty of 2133 Mhz DDR3 at CL9 already. They are no different performance wise than DD4 2133 Mhz at CL10 IF it can tighten down that far for the average Joe (read outside of Woomack, LOL). Part of the point of DDR4 outside of lower voltage was higher bandwidth and speed. So lowering it to low DDR3 speeds is counter intuitive.
 
Back