Haswell "K" people - What is your VID and best stable overclock?

[Hulk]

It seems as though some people think there might be some correlation between the VID of the Haswell "K" CPU's and how well they overclock. Let's see if there is any validity to this theory.

Post your VID, it's the voltage shown next to your Vcore in the BIOS when you have all default values loaded, no overclock, nothing changed. If you don't see it in your BIOS then you can get a close value of your VID by setting everything to default in the BIOS and then running CPUz under load and see what it reports as your Vcore.

Also post your best stable overclock. Don't worry about the definition of "stable." If you are using your computer day-to-day at your overclock setting it's stable for the purposes of this thread.

I'm pretty sure I have bum chip as my VID is 1.179 and I'm struggling to get 4.2GHz stable.

It would be nice for us guys and gals with high VID chips to know if we are spinning our wheels to get a good overclock, or perhaps there are some o/c setting specific to the higher VID CPU's that may help.

 

[ChronoReverse]

I'll need to check my VID but my chip does 4.3GHz when set to 1.165v.

And it's true stable, including AVX.

 

[nurturedhate]

Not sure what my VID is atm, will check next time I reboot. I'm at 4.7 at 1.2v. Does get into the high 80s under avx loads, don't exactly love that but it usually around the low 60's when gaming. On a macho rev a.

 

[nwo]

The highest I got my i7 4770k stable was 4.3GHz (for mining) but I am using the Asus AISuite to overclock it automatically. It runs at 1.225vcore @4.3 but I leave it at 4.1 @ 1.2 for everyday use. Temps are in the low 70s under full load on a Cooler Master Hyper 212+ with only one CPU fan (no additional 120mm fan).

 

[ChronoReverse]

Not sure what my VID is atm, will check next time I reboot. I'm at 4.7 at 1.2v. Does get into the high 80s under avx loads, don't exactly love that but it usually around the low 60's when gaming. On a macho rev a.

1.2v gets you AVX stable at 4.7GHz? Dang that's nice.

 

[thilanliyan]

What do you guys use to check for stability? AVX stability?

 

[JIM2003]

I'm not sure what my vid is, but I am stable at 1.258V @ 4.5ghz.

 

[ChronoReverse]

What do you guys use to check for stability? AVX stability?

Older versions of Prime95 doing Small FFTs suffice for normal stability.

The latest version uses AVX (and you'll find your temperatures 20c higher). I also run FPU only in AIDA64 (tends to push another 5c above Prime95AVX).


I might be limited by my heatspreader. I have an H110 and temperatures still rocket up past 90c when I try 1.25v or above (while doing AVX stability testing).

 

[rtsurfer]

4.5 Ghz at 1.280V for the 4770K
20 passes of LinX stable.

What do you guys use to check for stability? AVX stability?

Older versions of Prime95 doing Small FFTs suffice for normal stability.

The latest version uses AVX (and you'll find your temperatures 20c higher). I also run FPU only in AIDA64 (tends to push another 5c above Prime95AVX).


I might be limited by my heatspreader. I have an H110 and temperatures still rocket up past 90c when I try 1.25v or above (while doing AVX stability testing).

I believe the latest version of LinX works too.

BEWARE: LinX generates ridiculous amount of heat.

 

[JimmiG]

Mine does 4.5 GHz but needs about 1.365V when running AVX tests. Even with delid things get rather toasty

 

[bgt]

Not very impressive! tested with Prime95, FPU mode. Thermal design is horrible. And this for a K model........shameful really. 3770K is much better.
PS I use a big cooler with a 120mm NB fan.

 

[Jacky60]

<P>4.8Ghz 1.28vcore ring voltage 1.2. Have no ref to VID in bios?</P> Staility tested using Intel Extreme overclocking utitility. Alpenfohn K2 Doom CPU cooler.

 

[ShintaiDK]

Not very impressive! tested with Prime95, FPU mode. Thermal design is horrible. And this for a K model........shameful really. 3770K is much better.
PS I use a big cooler with a 120mm NB fan.

You keep posting the same in every thread. RMA your chip if it throttle at stock settings with the stock cooler. Since you are not doing it, I question your motives.

 

[Khato]

Staility tested using Intel Extreme overclocking utitility.

Just FYI, the Intel Extreme Tuning Utility stability test actually does not exercise AVX by default (and I didn't see a way to enable it?). I decided to try it out this time around and was quite excited at my 'excellent' temperatures and low'ish voltage to hit 4.5 GHz 'stable' according to their test... only to fire up AVX linx and immediately crash.

But that does raise the question of whether or not AVX stability is even necessary at current? Sure if you use some of the few programs that make use of AVX it would be, but I'd imagine for most overclockers the only time AVX instructions are used would be when running stability testing. I do believe I'm going to test out that 'theory' by running at a frequency/voltage that's perfectly stable on a 'normal' stability test but crashes immediately on AVX - if that works it'd nicely avoid the need to delid in order to get AVX thermals under control.

Edit: And to somewhat stay on subject - with my stock VID of 1.138V the best I can do while staying under 95C is 4.3 GHz at 1.26V for AVX stability. (Drops down to 1.2V roughly and under 65C for non-AVX.)

 

[BallaTheFeared]

I've come to the conclusion that I don't need avx stability, unless I plan to encode a video which just requires me to increase voltage in windows prior to running it (or use quick sync).

Otherwise, nothing I do uses it so it's a pointless exercise for me.


Mine is stock vid 1v


Highest stable with 2400MHz memory is 4.9Ghz, highest benchable 5.2GHz.

Daily for me is 4.8GHz with 1.245v

 

[ChronoReverse]

But that does raise the question of whether or not AVX stability is even necessary at current? Sure if you use some of the few programs that make use of AVX it would be, but I'd imagine for most overclockers the only time AVX instructions are used would be when running stability testing.

Let's put it this way. Handbrake nightly using AVX only loads my CPU to 50c while AVX stability testing loads it up 70c.

AVX stability isn't realistic just like Furmark is unrealistic.

 

[MBrown]

4.2 @ 1.192v

 

[bgt]

You keep posting the same in every thread. RMA your chip if it throttle at stock settings with the stock cooler. Since you are not doing it, I question your motives.

Motives??????????the point is....@stock it does not throttle(it gets to about 90C) so to bring it in for RMA is pointless. It throttles at 3.7Ghz but thats not stock. You then could say because its a K model you should be able top OC it but 3.6Ghz(stable) is the limit(with Prime95FPU). So thats a very tiny OC for such a proc.

 

[Termie]

I use Intel Burn Test for stability.

My old 4670K at stock hit 1.24V in IBT (I think), and I could only overclock it to about 4.2GHz with close to 1.28V.

My new 4770K at stock hit 1.2V in IBT (I know the VID is lower that my 4670K), and does 4.4GHz with 1.25V.

No doubt in my mind that VID means a lot.

Balla's ridiculous OC above is no doubt possible due to his very low VID. Lucky dog, that one...

Note that Haswell adds additional voltage under AVX loads like IBT (perhaps only when using adaptive overvolting - can't quite remember). My gaming voltage is much lower than my IBT voltage.

 

[Hulk]

Doesn't look like this thread is going anywhere but I'm going ask again if anyone would like to post

Haswell "K" VID (from BIOS next to CPU voltage with stock (default setting) or loaded into Windows and using CPUz under load. Preferably the first method if possible

Your day-to-day overclock GHz and the voltage for it.

I think it would be really interesting to create a small database to see how well the VID corresponds with overclocking ability.

 

[ratjacket]

i7-4770K L316B317 GA-Z87X-OC
stock vcore 1.207 (4.4 is stable at 1.235 with ram on xmp 2133 and uncore auto oc to 4000)

this CPU will go alot higher with less volatge if i chose to accept a lesser version of stability (However I dont like to find out down the track that a demanding app has made me unstable and to be unsure if the cpu has degrdaed or i just failed in my stress testing)

prime 95 small FFT 24hours stable
HT disbaled during stress test (to keep temps down and becuase it crashes my version of prime95 (latest non beta when i got my haswell in October, i will add version number when i get home from work) faster with HT disabled)

I didnt use the beta prime95 with AVX, still gets unrealisticly hot without it, doesnt reach throttle temps though. It gets close with HT enabaled.

I found Adia and Intel XTU to be way too soft of a stress test

not sure about OCCT (AVX mode and non AVX) it would run after prime but i never pushed it for 24+ hours

LinX crashes my system on stock so i cant loop that
I tried OCCT after that to make sure i wasnt realy unstable but it didnt crash with AVX enabled so LinX is juts broke for me

Linpack looks like to much like hard work to learn how to use

Apples for Apples lets post Prime 95 small FFT HT disabled 24hour stable ocs

or not cause im a douche bag that goes OTT with stress testing ( i like to think i have found the middle ground with prime 95 no AVX HT disabled)

 

[timchen]

My i7-4770k shows 1.024v in bios... but with everything set to auto, in windows under load it shows 1.15v @ 3.7 (default multiplier is 39 38 37 37 it seems...)

It seems to me 4.4 is stable at 1.25 manual both for AVX (used ibt to test) and old prime 95. But due to problems with AVX using adaptive voltage (don't want the peak voltage to be higher than 1.3) I have to lower my current setup to 4.3@1.18. Under AVX loads the voltage cranks back to around 1.25.

One thing I found weird is that my cpu seems to hit a huge voltage wall from 4.3 to 4.4. It seems the CPU is stable at 4.3 with 1.18v (or less, didn't test with very fine grain but not stable at 1.15), but need 1.25v to stay stable at 4.4 (1.23 is sort-of stable, last hours before BSOD.) Is this voltage wall a common observed thing? Or is there some other factors that may affect the overclock?

 

[nwo]

One thing I found weird is that my cpu seems to hit a huge voltage wall from 4.3 to 4.4. It seems the CPU is stable at 4.3 with 1.18v (or less, didn't test with very fine grain but not stable at 1.15), but need 1.25v to stay stable at 4.4 (1.23 is sort-of stable, last hours before BSOD.) Is this voltage wall a common observed thing? Or is there some other factors that may affect the overclock?

I dunno about you, but mine won't clock to 4.4GHz stable. I haven't really messed with it too much beyond 4.3GHz though because I fried my 4670k trying to push it past 4.4GHz :whiste:

 

[Morbus]

I'm running at 4.2GHZ 1.2V, but it tops at 4.5GHZ@1.26V.

 

[Arkaign]

The lottery on Haswell is even worse than the lottery with Ivy.

They make FANTASTIC mobile chips, and the non-K parts are all great. The K chips though are a real crap shoot for overclockers. I've yet to find a stable Haswell beyond 4.4Ghz, even with very expensive motherboards and at least average cooling (H100, H110, Noctua DH-14).

IOW, zero reason to move from a SB or Ivy to HW, but the Z87 is enough reason to go HW for a new build, you don't need as much clockspeed to hit the same approximate level of performance.

On a purely average level, 4.25Ghz HW = 4.5Ghz Ivy = 4.75Ghz SB. And all of those are fairly typical decent OCs with $30 air cooling. I finally found a 5Ghz low-heat Ivy, so I'm keeping it. I had a 4.4 4770K, but it was ludicrously hot, and I'm not going to delid (though it's very enlightening).