F5F Stay Refreshed Power Users Overclocking Suggestions for achieving stability in the final stages of 4790k OC, including uncore, VRIN, and RING BUS adjustments.

Suggestions for achieving stability in the final stages of 4790k OC, including uncore, VRIN, and RING BUS adjustments.

Suggestions for achieving stability in the final stages of 4790k OC, including uncore, VRIN, and RING BUS adjustments.

G
gavin_shaka
Senior Member
535
03-17-2016, 02:14 AM
#1
SPECS:
4790k
Noctua DH15S
5 other fans including industrial top fan
Gigabyte z97mx Gaming 5
16gb Ballistix Tactical (OC to 2133 with 10 9 10 24 CR1 and ram enhancement set to "stability" in BIOS, 1.52 Volts)
Fractal Define Mini Case
EVO SSD
Aim is to achieve 4.9 Ghz stable on lowest voltages possible, the thing I'm confused about is the voltage leap from 4.8 - 4.9 and the general instability I'm getting as I tweak the overclock around 4.9 relative to the stability I have found at 4.8.
I know things like uncore, VRIN and RING BUS tweaks can all aid the stability and temperatures of the overall overclock. I've lost an old link that explained a lot of this, so hoping the community can chime in.
I've been overclocking my 4790k for some time at a few different speeds between 4.7 and 4.9Ghz. The last stable long term overclock I had I left it on 4.7 at around or a bit below 1.26Vcore I believe.
For some reason all my BIOS settings reset at some point, I honestly can't remember why, when or how this might have happened. The point being that I don't know what my old settings were, and no matter.
Now but I'm trying a new approach to squeeze better single threaded performance for some flight sims I'm playing (il-2 sturmovik series, BoX and CloD).
I have achieved good results so far, please note I have hyper threading off, all c states off, turbo off and power management to performance in Windows.
This overclock is passing the stress tests
(p95, AIDA64), but temps going a little high for my liking at this stage (peaks 85).. I don't mind high temps in stress test but hoping for under 80. Also I'm getting crashes while gaming, after 30-60 mins. Possibly could be related to GPU but not so likely IMO.
49 core multiplier
40 uncore
Vcore 1.356
VRIN 1.78
RING BUS 1.151
This overclock is perfectly stable
, and as you can see it's much lower voltages, and I'm getting temps peaking at 75 in stress test, which is great as far as I'm concerned.
48x
40uc
Vcore 1.28
VRIN 1.71
RING BUS 1.14
I'm tweaking it lower and lower, but the other OC seems to be going higher and higher...
Can anyone suggest what might be my bottleneck for stability and temperatures?
Why is the leap so big for the Vcore from 4.8-4.9? I expect I should be able to get a much lower Vcore for 4.9 if I can find the right settings for the other variables...
I guess I can try to drop the uncore to about 35ish for some tests, and also drop ring bus along with that, so that should take some stress out of the system, though I wanted to keep uncore around 40 in an ideal world. I'm not quite sure how to judge dropping the ring but relative to everything else though... how low can it go?
NB - I don't mind hitting slightly above 1.3 Vcore, and I don't mind hitting 80C in stress tests, as they are unnatural and not representative of what I'm doing every day. I only need another year out of this chip...
thanks for reading!
G
gavin_shaka
03-17-2016, 02:14 AM #1

SPECS:
4790k
Noctua DH15S
5 other fans including industrial top fan
Gigabyte z97mx Gaming 5
16gb Ballistix Tactical (OC to 2133 with 10 9 10 24 CR1 and ram enhancement set to "stability" in BIOS, 1.52 Volts)
Fractal Define Mini Case
EVO SSD
Aim is to achieve 4.9 Ghz stable on lowest voltages possible, the thing I'm confused about is the voltage leap from 4.8 - 4.9 and the general instability I'm getting as I tweak the overclock around 4.9 relative to the stability I have found at 4.8.
I know things like uncore, VRIN and RING BUS tweaks can all aid the stability and temperatures of the overall overclock. I've lost an old link that explained a lot of this, so hoping the community can chime in.
I've been overclocking my 4790k for some time at a few different speeds between 4.7 and 4.9Ghz. The last stable long term overclock I had I left it on 4.7 at around or a bit below 1.26Vcore I believe.
For some reason all my BIOS settings reset at some point, I honestly can't remember why, when or how this might have happened. The point being that I don't know what my old settings were, and no matter.
Now but I'm trying a new approach to squeeze better single threaded performance for some flight sims I'm playing (il-2 sturmovik series, BoX and CloD).
I have achieved good results so far, please note I have hyper threading off, all c states off, turbo off and power management to performance in Windows.
This overclock is passing the stress tests
(p95, AIDA64), but temps going a little high for my liking at this stage (peaks 85).. I don't mind high temps in stress test but hoping for under 80. Also I'm getting crashes while gaming, after 30-60 mins. Possibly could be related to GPU but not so likely IMO.
49 core multiplier
40 uncore
Vcore 1.356
VRIN 1.78
RING BUS 1.151
This overclock is perfectly stable
, and as you can see it's much lower voltages, and I'm getting temps peaking at 75 in stress test, which is great as far as I'm concerned.
48x
40uc
Vcore 1.28
VRIN 1.71
RING BUS 1.14
I'm tweaking it lower and lower, but the other OC seems to be going higher and higher...
Can anyone suggest what might be my bottleneck for stability and temperatures?
Why is the leap so big for the Vcore from 4.8-4.9? I expect I should be able to get a much lower Vcore for 4.9 if I can find the right settings for the other variables...
I guess I can try to drop the uncore to about 35ish for some tests, and also drop ring bus along with that, so that should take some stress out of the system, though I wanted to keep uncore around 40 in an ideal world. I'm not quite sure how to judge dropping the ring but relative to everything else though... how low can it go?
NB - I don't mind hitting slightly above 1.3 Vcore, and I don't mind hitting 80C in stress tests, as they are unnatural and not representative of what I'm doing every day. I only need another year out of this chip...
thanks for reading!

C
CookieStars
Member
220
03-23-2016, 07:42 AM
#2
I'm considering this now, and my goal is to increase the single core overclock, though I'm uncertain about the proper steps.
I came across a forum post suggesting it might be feasible, but not guaranteed on my specific board...
So far, I've experimented with turbo settings, aiming for single core 51, double core 50, and three or four cores at 48, where I currently maintain stability.
Most results appear stable, with some benchmarks improving while others remain consistent.
Until I run more demanding single-thread tests, temperatures spike significantly. I suspect the auto settings for power and current management in turbo are causing issues, and I'm unsure how to adjust them.
It's possible that lowering the two-core turbo to 49 could provide a safer transition.
I'm also not completely confident about how games or applications handle switching between single and multi-core usage—though it seems to happen frequently, so there may be moments when single core turbo is beneficial during gameplay. Benchmarks indicated this, even though they weren't strictly single-core dependent.
It might help to consult a more dedicated overclocking community for these specifics.
C
CookieStars
03-23-2016, 07:42 AM #2

I'm considering this now, and my goal is to increase the single core overclock, though I'm uncertain about the proper steps.
I came across a forum post suggesting it might be feasible, but not guaranteed on my specific board...
So far, I've experimented with turbo settings, aiming for single core 51, double core 50, and three or four cores at 48, where I currently maintain stability.
Most results appear stable, with some benchmarks improving while others remain consistent.
Until I run more demanding single-thread tests, temperatures spike significantly. I suspect the auto settings for power and current management in turbo are causing issues, and I'm unsure how to adjust them.
It's possible that lowering the two-core turbo to 49 could provide a safer transition.
I'm also not completely confident about how games or applications handle switching between single and multi-core usage—though it seems to happen frequently, so there may be moments when single core turbo is beneficial during gameplay. Benchmarks indicated this, even though they weren't strictly single-core dependent.
It might help to consult a more dedicated overclocking community for these specifics.