Tweet
Hello!
For the past few days I've been trying to establish a safe OC of my i5-2500k on a P67A-UD3-B3 motherboard.
I'm trying to run it at 4200MHz (100x42). I found it to be stable at Vcore=1.270.
The process of overclocking raised many question of how the mobo applies Vcore and I hope you can help me with at least some of them!
I left all the setting in BIOS on auto, except:
- Turbo: disabled
- PPL overload: disabled
- Load Line Calibration (I was switching on/off, as I describe below)
- Vcore
Part 1. Vcore set manually.
Firstly, I set Vcore manually to 1.260V and rebooted.
Vcore reported by BIOS (in M.I.T. or PC Health) was different than reported by EasyTune6 on idle.
LLC enabled: BIOS 1.260V, EasyTune6 1.284V.
LLC disabled: BIOS 1.248V, EasyTune6 1.284V (still!).
Is the software inaccurate, or does the mobo apply more voltage under Windows?
On stress (prime95) the Vcore reported by ET6 was dropping to 1.212V (with LLC enabled, quite a bit of Vdroop!)
With LLC disabled, it was dropping to 1.19V and gave BSOD.
(Vdroop on my mobo usually is bigger by ~0.02V with LLC disabled, as I tested Vcores from 1.26V to 1.32V)
Then I returned to BIOS, set Vcore to 1.270V and rebooted.
Again various reports:
LCC enabled: BIOS 1.274V, ET6 1.296V (idle).
LCC disabled, BIOS 1.600V, ET6 1.296V (idle).
On stress, with LLC enabled, ET6 was reporting Vcore as low as 1.224V.
(With LLC disabled, it was dropping to 1.200V).
Again, why idle Vcore reported by ET6 is higher than the one set and displayed in the BIOS?
So is my Vdroop 0.072V (1.296-1.224) or 0.050V (1.274-1.224)?
Part 2. Dynamic Vcore
Now the stranger part!
I set Vcore to Normal, to use the Dynamic Vcore (DVID).
CPU Vcore defaulted to 1.320V (this is on clock ratio 42x).
Firstly, I set the DVID to +0.00V.
I left LLC enabled.
PC Health in BIOS showed Vcore of 1.326V (as expected).
Under Windows, when CPU was idle, ET6 reported Vcore 0.96V (also expected).
With some load applied (web browser, prime95 on one or two cores) Vcore was going higher, 1.1V, 1.2V, etc.
On heavy load Vcore went up to 1.31V. Now that was unexpected! Where is Vdroop?
Is there almost no Vdroop, or does my motherboard actually apply Vcore of ~1.38V under heavy load, and I just see the 1.31V after the huge 0.07 Vdroop kicked in?
If there's almost no Vdroop, then WHY? Why so different on static Vcore, and on DVID?
I thought that Vdroop is actually good for CPU, to help stabilize something (TPD or some amperage, I don't really know).
So is Vdroop not needed with DVID, or is it a flaw of Gigabyte's motherboard?
Or maybe DVID works as intended, but the huge Vdroop with static Vcores is a flaw?
So my biggest, most important question: which one should I use for my CPU longevity?
I found out that my OC is stable if I use Dynamic Vcore = -0.05V (then my voltage is between 0.91V idle, and 1.26V load).
Should I use that, or the static Vcore=1.270?
Which one is better for my CPU longevity?
Thanks a lot in advance!
For the past few days I've been trying to establish a safe OC of my i5-2500k on a P67A-UD3-B3 motherboard.
I'm trying to run it at 4200MHz (100x42). I found it to be stable at Vcore=1.270.
The process of overclocking raised many question of how the mobo applies Vcore and I hope you can help me with at least some of them!
I left all the setting in BIOS on auto, except:
- Turbo: disabled
- PPL overload: disabled
- Load Line Calibration (I was switching on/off, as I describe below)
- Vcore
Part 1. Vcore set manually.
Firstly, I set Vcore manually to 1.260V and rebooted.
Vcore reported by BIOS (in M.I.T. or PC Health) was different than reported by EasyTune6 on idle.
LLC enabled: BIOS 1.260V, EasyTune6 1.284V.
LLC disabled: BIOS 1.248V, EasyTune6 1.284V (still!).
Is the software inaccurate, or does the mobo apply more voltage under Windows?
On stress (prime95) the Vcore reported by ET6 was dropping to 1.212V (with LLC enabled, quite a bit of Vdroop!)
With LLC disabled, it was dropping to 1.19V and gave BSOD.
(Vdroop on my mobo usually is bigger by ~0.02V with LLC disabled, as I tested Vcores from 1.26V to 1.32V)
Then I returned to BIOS, set Vcore to 1.270V and rebooted.
Again various reports:
LCC enabled: BIOS 1.274V, ET6 1.296V (idle).
LCC disabled, BIOS 1.600V, ET6 1.296V (idle).
On stress, with LLC enabled, ET6 was reporting Vcore as low as 1.224V.
(With LLC disabled, it was dropping to 1.200V).
Again, why idle Vcore reported by ET6 is higher than the one set and displayed in the BIOS?
So is my Vdroop 0.072V (1.296-1.224) or 0.050V (1.274-1.224)?
Part 2. Dynamic Vcore
Now the stranger part!
I set Vcore to Normal, to use the Dynamic Vcore (DVID).
CPU Vcore defaulted to 1.320V (this is on clock ratio 42x).
Firstly, I set the DVID to +0.00V.
I left LLC enabled.
PC Health in BIOS showed Vcore of 1.326V (as expected).
Under Windows, when CPU was idle, ET6 reported Vcore 0.96V (also expected).
With some load applied (web browser, prime95 on one or two cores) Vcore was going higher, 1.1V, 1.2V, etc.
On heavy load Vcore went up to 1.31V. Now that was unexpected! Where is Vdroop?
Is there almost no Vdroop, or does my motherboard actually apply Vcore of ~1.38V under heavy load, and I just see the 1.31V after the huge 0.07 Vdroop kicked in?
If there's almost no Vdroop, then WHY? Why so different on static Vcore, and on DVID?
I thought that Vdroop is actually good for CPU, to help stabilize something (TPD or some amperage, I don't really know).
So is Vdroop not needed with DVID, or is it a flaw of Gigabyte's motherboard?
Or maybe DVID works as intended, but the huge Vdroop with static Vcores is a flaw?
So my biggest, most important question: which one should I use for my CPU longevity?
I found out that my OC is stable if I use Dynamic Vcore = -0.05V (then my voltage is between 0.91V idle, and 1.26V load).
Should I use that, or the static Vcore=1.270?
Which one is better for my CPU longevity?
Thanks a lot in advance!
Comment