I could go hours into specifics, but adjust the BaseClock (or in your case, the multiplier coan be adjusted as well) to get faster speed.

The faster you go, the more voltage you will need. As a threshold, 1.6 volts is extreme, so always use under that, especially when you are running your build on air. With a 980X you should hopefully be able to achieve 4ghz with your vcore (main voltage) around 1.3-1.4v. it really depends on the chip though. Someone can have the exact same build as you, and get vastly better results, or far worse results.

Best way to do it is work backwards. Set a goal you want to achieve, then test it. Say you set the multiplier so you come out with a speed of 4400 mghz. test it with Prime 95, OCT, or any stress test software you choose.

If it throws you a blue screen, increase your voltage a little, then drop your multiplier, (or modify the Baseclock-doing this though raises the bus frequency)

it would help to know what board you have, as now i can only give you basics.

Check this thread out for a little more info

Core i7 980x - QPI vs multiplier to overclock? - Overclockers UK Forums

And n_w95482 has an overclocked 980X, so you could ask him for more in depth tips.

 

XtremeSystems is also always a good place to go when you want help with overclocking.

 

i have Newegg.com - EVGA 141-GT-E770-A1 LGA 1366 Intel X58 SATA 6Gb/s USB 3.0 Extended ATX Intel Motherboard

it has a lot of nice features and its designed for overclocking. i found stuff on their forums but its quite hard to understand. plus my main concern is if i set something to high that it might blow out something :/.

 

Just raise the multiplier and the voltage. It's essentially that easy.

i do have a very good, non stock cpu cooler and even at a max load the highest temp i saw was high 30's.

Run Prime95 and Intel Burn Test. It's a bit hard to believe it's only in the 30s unless you're running very good cooling and/or have a very cold ambient temperature.

 

here is the cooler i use: Newegg.com - Noctua NH-D14 120mm & 140mm SSO CPU Cooler

and here is my case: Newegg.com - COOLER MASTER HAF X RC-942-KKN1 Black Steel/ Plastic ATX Full Tower Computer Case

when i am just running the pc normally my cpu can run as low as 13 C.

btw the temps i pulled stressing the cpu was with prime 95 stressing all 12 threads at 100% load.

 

For 24/7 use, you'll probably want to keep Vcore at or below 1.4v, and uncore at or below 1.35v. Intel Burn Test, when properly configured, will get the chip much hotter than Prime95.

Out of the two CPUs and motherboards I've used, both of them benefited from a bump in CPU PLL voltage, from 1.8v to 1.84v, when increasing the baseclock above 190 MHz. Increasing QPI PLL slightly may help with higher uncore speeds too (3.2+ GHz).

What I would recommend is starting with a low CPU multiplier and increase the memory multiplier to see how fast it'll run at your desired timings and voltage. Once you find a fast, stable setting, do the opposite. Drop the memory multiplier down low, then play with the CPU to see how it ramps up, particularly once it starts requiring voltages. Run Intel Burn Test for a few passes with a large amount of memory (I used 4400 MB with 6 GB installed).

Once you find about where you want to run the CPU at, crunch the numbers to see what baseclock, CPU multiplier, and RAM multiplier will get you closest to your goals. For me, that was running the baseclock at 194 MHz, CPU multiplier at 22, and RAM multiplier at 8. I also tested 156x27, although that ran the CPU just below what it was capable of at my chosen voltage. When you're at that point, run IBT for a couple of hours and adjust settings as needed for stability.

Some people will say to keep the uncore:QPI speeds at a ratio of 8:9 or less. That seems to mainly apply to earlier Bloomfield chips, if that. I haven't noticed any issues with my 980X at 1:1 or even higher (I tried 3.7+ GHz uncore with QPI at 3.5, and while it wasn't fully stable, it ran pretty well) and even my 920 did 1:1 easily. I've also seen posts stating that even CPU multipliers don't work as well as odd multipliers. I haven't noticed that in my testing.

Here's a screenshot of my stable settings. Notice the high temperatures, that was using a TRUE heatsink (with some of those plastic rings from CD spindles taped together and put under the mounting bracket for extra pressure), fairly fresh Ceramique paste, and a pair of Noctua fans in push/pull. My chip seems to require less voltage than most I've noticed, so I was lucky to get it to run at that speed/voltage (~1.34v for Vcore and uncore).

A note for running Intel Burn Test: You will get higher GFLOPS and lower temperatures by disabling HyperThreading, it's normal behavior for Linpack-based tests. I went up about 10-12 GFLOPS and dropped 10-13 C by disabling HT. If you do decide to keep HT on, manually set IBT to 12 threads.

 

13C!? Is your room 45F or something?

 

no the room is currently 75 F, the heatsink is just that good, i can show you a picture if you want, its like a giant block of metal, i should know...it cut a nice bit off my thumb putting it in :/. of course thats the i guess base temp or whatever? i am not looking at individual core temps, those still sit quite a bit under 30 right now and do not get much higher than that either with prime 95 even when it runs for a few hours.

 

no the room is currently 75 F, the heatsink is just that good, i can show you a picture if you want, its like a giant block of metal, i should know...it cut a nice bit off my thumb putting it in :/. of course thats the i guess base temp or whatever? i am not looking at individual core temps, those still sit quite a bit under 30 right now and do not get much higher than that either with prime 95 even when it runs for a few hours.

If your room is ~75F, that's just under 24C. How is your CPU running down to 13C in a ~24C environment using that air as it's cooling method?

If your room is ~75F/24C, this means that, not matter what, your CPU can not be running cooler than that at absolute minimum. It's literally impossible. You are using the air to cool the heatsink on something producing heat, so the temperature of the air is the defining limit. On top of that, it's rare it'll be only as warm as the room, because it's a source of heat itself, so you'll usually be at least ~5C-10C warmer than ambient, and that's at around idle too.

Forgive me for sounding harsh or anything on this one thing, but it's just not possible. If you are seeing those sorts of temperatures, then you're getting inaccurate readings, either due to the program(s), or sensor(s). They're known to be increasingly inaccurate below ~60C or so anyway (which is fine, since temperatures below that aren't a problem), but 13C, and temperatures barely over 30C under Prime95 load? That should have sent up flags to be honest.

 

i am attaching a current image of my temps right now. i have a video and other stuff open right now so the system is not idling, not really been idling lately at all since i been busy so its just a little above normal and i have another app that i have run that can check the sensors to see if they are working ok (think its called real temp). the next time i see it idle and its back to 13 i will show you what i mean. mind you that you will see what i mean when i say cpu temp and not the individual core temps.

btw, i also attached an image of my cooler, its massive and barely even fits but might also explain the low temps as supposedly the thermal compound provided it exceptionally good.

 

It doesn't matter what CPU cooler you have or what programs are showing. You've got either inaccurate sensor(s) and/or inaccurate reports from the program(s). It is physically impossible for your CPU to be running colder than the ambient air temperature, as that is what is cooling it. If your room is ~75F/24C, your CPU will be, at lowest, ~75F/24C, but realistically, it will be even higher (typically ~5C-10C) since it is a source of heat.

Anyway, on the main matter at hand, this thread is for overclocking it, so n_w95482 probably has the best advice for that. I just wanted to point that out.

 

Garnet is right about that. Even with water cooling, it's nearly impossible to beat ambient, unless using chilled water or steam cooling.

My room was 10c (back door open on a winter night) when this was taken, running chilled water.

Even water cant beat ambient.

Only with a CPU pot (Nitrogen) or a contact freezer (Phase Change) can you go below ambient.

 

The CPU core temperatures look about right, use those instead.

 

ok thanks guys, the information is definitely useful, i will see what i can pull together this weekend when i have time to start overclocking everything.

 

Use other programs to test your CPU, and sensors tend to be more inaccurate the farther from TjMax they are, and also notably below ~60C, so you can probably rely on them to still let you know before they're running too hot.

 

It seems like I going to have to disable HT and/or make another profile adjusted for a much lower clock just to make it through the summer. This isn't going to be fun.

 

What are you cooling with anyways?

 

Yeah, I'm curious too. If it's on water, it should be a bit better than that. If you're on air, that's an awfully high frequency.

 

Water, but this room isn't to far from the attic. Ambient temps are going to be higher than it is outside. My old air conditioning unit probably isn't going to cut it either. I'll worry about what I need to do when it starts getting hot. The heat wave is over (85F+), and the weather is cool for now.

 

In that case, disabling HT would probably suffice.

 

Awesome Voltage!

Now I got to do stability tests from 1.44 to 1.46 sometime tomorrow. I forgot where it was before. Auto is bad.

 

Yeah, I learned with my i7 920/Asus P6T that auto voltages were bad. I remember it running 1.4+ volts at 3.7 GHz when 1.3 was enough.

 

1.3? Was that a D0 stepped one?

 

Nope, very early C0 batch (early-mid November 2008). It didn't need too much Vcore at that speed, but I had to dump on the uncore voltage. It needed just under 1.37v. It did 3.5 GHz at stock voltages so it ran out of steam pretty quickly.

 

Ah thought so. With the D0 you should have been able to easily hit 3.7 on 1.25v.