6700K OC - high temps?

Shocked

Limp Gawd
Joined
Nov 1, 2009
Messages
373
Have a 6700K and a MAXIMUS VIII HERO cooled by a Noctua NH-D15S (with second fan) and MX-4. Case is a Corsair Air 540 with 5x Fractal Design HF-14. Using 2x16GB G.Skill TridentZ DDR4-3200 Cas 14 (1.35V).

I initially tried 1.325v in manual mode (with an LLC setting of 6), FCLK set to 1000, and RAM set to XMP settings, but the system locked up after about 10 minutes of Prime95 blend mode. Raised the voltage to 1.34 and the same stress test started hitting 80C, it mostly hovered in the low 70's but it spent enough time around 79-80 to make me nervous enough to call it quits after about 30 minutes.

Temps are a lot lower in AIDA64 with a max of 74C. I was going to try x264 as a middle ground but I've had trouble getting it working (says I don't have AviSynth 260 installed even though I do). However, I've been using P95 for probably 15 years so it's always kind of been the standard I shoot for.

Is this about normal or do I need to try reseating the HSF?
 
Have a 6700K and a MAXIMUS VIII HERO cooled by a Noctua NH-D15S (with second fan) and MX-4. Case is a Corsair Air 540 with 5x Fractal Design HF-14. Using 2x16GB G.Skill TridentZ DDR4-3200 Cas 14 (1.35V).

I initially tried 1.325v in manual mode (with an LLC setting of 6), FCLK set to 1000, and RAM set to XMP settings, but the system locked up after about 10 minutes of Prime95 blend mode. Raised the voltage to 1.34 and the same stress test started hitting 80C, it mostly hovered in the low 70's but it spent enough time around 79-80 to make me nervous enough to call it quits after about 30 minutes.

Temps are a lot lower in AIDA64 with a max of 74C. I was going to try x264 as a middle ground but I've had trouble getting it working (says I don't have AviSynth 260 installed even though I do). However, I've been using P95 for probably 15 years so it's always kind of been the standard I shoot for.

Is this about normal or do I need to try reseating the HSF?
thermal throttle is 100c shutdown is 105c, so ive been told. so if youre under 80 youre gtg!
 
with an LLC setting of 6

That's a lotta LLC man for that voltage. What are you clocked at 4.5 or so? If you wanna test encoding, have you tried handbrake? It supports qsync pretty well. I'd test with and w/o to get feel for it.
 
That's a lotta LLC man for that voltage. What are you clocked at 4.5 or so? If you wanna test encoding, have you tried handbrake? It supports qsync pretty well. I'd test with and w/o to get feel for it.
I was under the impression that 1 is max LLC and 8 is minimum on ASUS boards?
 
It goes from 0-100%, 1-8.
You're right, I read the description in the BIOS backwards. I thought it was odd that 1 would be the max. What would be a more appropriate average LLC for these chips? Many of the recommendations I found for ASUS boards recommended 5-6, which is actually what led me to trying 6.
 
I found that LLC 4 gives a flat voltage curve up to just under 1.4v. That is it gives a 1:1 load voltage from what is input to what is read under load. It's good for clocks up to around 4.7 which basically covers most usable clocks . Beyond that we're talking massive amounts of volts needed and we start getting into preferences more that technical specs. LLC 5-6 is way too much voltage and imo going about it the wrong way for typical overclocks. Also, there's a trick with the XMP setting. Do you know about it? Load defaults, then go to the XMP drop bar, click on XMP but don't hit ok, instead hit escape. That will load the XMP memory profile w/o locking you into it.
 
I found that LLC 4 gives a flat voltage curve up to just under 1.4v. That is it gives a 1:1 load voltage from what is input to what is read under load. It's good for clocks up to around 4.7 which basically covers most usable clocks . Beyond that we're talking massive amounts of volts needed and we start getting into preferences more that technical specs. LLC 5-6 is way too much voltage and imo going about it the wrong way for typical overclocks. Also, there's a trick with the XMP setting. Do you know about it? Load defaults, then go to the XMP drop bar, click on XMP but don't hit ok, instead hit escape. That will load the XMP memory profile w/o locking you into it.
Thanks!
 
You're right, I read the description in the BIOS backwards. I thought it was odd that 1 would be the max. What would be a more appropriate average LLC for these chips? Many of the recommendations I found for ASUS boards recommended 5-6, which is actually what led me to trying 6.
anandtech i believe did a review testing what each setting did but i think eah board is different.

I always do max on ASRock but IIRCC ASRocck doesn't over shoot like ASUS does....or not as badly. 4/5 should be good

nm tweak town. I just googled this. I doubt the right board so google other reviews to confirm

http://www.tweaktown.com/reviews/7506/asus-z170-intel-motherboard-review/index7.html
 
Well, I tried going down to LLC 5, but P95 blend failed within 1-2 minutes (even if P95 is overkill for stress testing, I'd say that's a bad sign). So instead of raising LLC back up to 6 I tried 1.35V with the same results. Raised LLC back up to 6 and P95 ran for 30 minutes before encountering an error.

Could my particular board just suck at voltage regulation? For the most part my temps during the test are hovering in the mid 60's to mid 70's, but will go up to 80-82C for several minutes at a time.
 
Well, I tried going down to LLC 5, but P95 blend failed within 1-2 minutes (even if P95 is overkill for stress testing, I'd say that's a bad sign). So instead of raising LLC back up to 6 I tried 1.35V with the same results. Raised LLC back up to 6 and P95 ran for 30 minutes before encountering an error.

Could my particular board just suck at voltage regulation? For the most part my temps during the test are hovering in the mid 60's to mid 70's, but will go up to 80-82C for several minutes at a time.

If it needs more voltage then don't go around the problem by raising LLC, give it more voltage. P95 failing within a couple minutes = not enough vcore. P95 failing in the half hour or more region = memory or imc. These are generalizations mind you, YMMV. That said, you're solidly in the not enough vcore camp imo. Btw, the first 5 minutes of P95 is testing, you don't get into the heart of it till later. You'll know when your temp starts spiking up. That extra hot time is when its doing real work.
 
If it needs more voltage then don't go around the problem by raising LLC, give it more voltage. P95 failing within a couple minutes = not enough vcore. P95 failing in the half hour or more region = memory or imc. These are generalizations mind you, YMMV. That said, you're solidly in the not enough vcore camp imo. Btw, the first 5 minutes of P95 is testing, you don't get into the heart of it till later. You'll know when your temp starts spiking up. That extra hot time is when its doing real work.
I really did that just as an experiment, I was curious what the difference in temps would be between LLC 5 and 6. It's pretty huge. Now I know I at least have some headroom to raise the vcore.
 
I really did that just as an experiment, I was curious what the difference in temps would be between LLC 5 and 6. It's pretty huge. Now I know I at least have some headroom to raise the vcore.
again look at my link. It kinda shows how ASUS LLC works. You want the LLC that most accurately provides voltage. According to Tweak and Smoke??? LLC 4/5 are most accurate. Reading more reviews via googling LLC will give you a better look at what LLC works best at what voltage.

It is generally best to start with the highest voltage you feel safe with and find the clock that works.....well thats the fast and lazy way to go.
 
Back
Top