Hello.
I'm newbie in overclocking topic and I'm somewhat lost in setting UV/OC on my 13600KF @ TUF GAMING Z690-PLUS D4 @ BIOS 4301 (0x12F) @ Arctic Liquid Freezer III 360 and politely request for some help.
I'm using method from user charonme from comments section from this thread: https://www.reddit.com/r/overclocking/comments/1gaqwdt/how_do_i_undervolt_13600kf/ So, AC_LL=DC_LL, Core VIDs=Vcore etc.
What I would like to achieve:
5400 Mhz all P-Core by core usage
4200 Mhz all E-Core by core usage
Reasonable Vcore at 100% load (<1.3 V I think)
What I accomplished:
By HWiNFO @ Cinebench R23
Core Clocks: P-Core – 5401,3 Mhz, E-Core 4201 Mhz
Core Effective Clocks: P-Core – ~5401,3 Mhz, E-Core – ~4070 Mhz
Core VIDs: 1.296 V
Vcore: 1.296 V
By HWiNFO @ Prime95 @ Small FFTs @ 14 Cores
Core Clocks: P-Core – 5000 Mhz, E-Core 4201 Mhz
Core Effective Clocks: P-Core – ~5000 Mhz, E-Core – ~2840 Mhz
Core VIDs: 1.254 V
Vcore: 1.261 V
Comment – As we can see @ Cinebench R23 E-Core's Core Effective Clocks are about 100Mhz lower than the target and @ Prime95 even lower are both P-Cores and E-Cores. When I tried to achieve actual settings I have to decrease adaptive minus offset and AC_LL=DC_LL for stable benchmarking and CPU Core VID = Motherboard Vcore (@ HWiNFO). I benchmarked after every change and sometimes it show more closer values to target of clocks but without stability (freeze @ benchmark).
After final benchmarks one after another (Cinebench R23, Prime95) HWiNFO only show Max Turbo Limit at Performance Limit Reasons.
My not all BIOS settings (video: https://streamable.com/yo6sbv):
HyperThreading – Off
ASUS MultiCore Enhacement – Disabled – Enforce All Limits
SVID Behavoir – Auto
P-Core Ratio – By core usage all x 54
E-Core Ratio – By core usage all x 42
Global Core SVID Voltage – Adaptive Mode -0.09 V Offset Voltage
Cache SVID Voltage – Adaptive Mode -0.09 V Offset Voltage
AC_LL=DC_LL – 0,38 mOhms
Ring Down Bin – Enabled
Min. CPU Cache Ratio – Auto
Max. Cpu Cache Ratio – Auto
AVX2 – Auto
CPU Load-Line Calibration – Level 5
Max CPU Core Temperature – 85 Celsius
Unlimited ICCMAX – Disabled
CPU Core/Cache Current Limit Max. – 511.75 A
Power Limit 1 – 181 W
Power Limit 2 – 181 W
Package Power Time Window – 56 Sec
IA VR Voltage Limit – 1.400 V
IA CEP – Enabled
SA CEP – Enabled
Software I used for monitoring and benchmark:
CPU-Z v2.15.0
HWiNFO 64 v8.28-5770
Cinebench R23 v23.200
Prime95 v30.19 build 20
Geekbench v6.4.0
Questions:
1. Does this method of OC/UV is reliable?
2. Does low effective clock is related to too big minus adaptive offset?
3. Do I need to ie. lower in number LLC level and decrease more minus adaptive offset to maintain target clocks (one method at time)?
4. Or something different is related to this low clocks?
I'm asking because when I changed LLC and minus adaptive offset sometimes results are suprising for me. Ie. lower offset and lower clocks (or I misconfigured something…)
Additional info:
I "migrated" from yesterday stable for me settings
CPU Ratios – 51 / 39
AC_LL – 0,79 mOhms
DC_LL – 0,79 mOhms
Adaptive Offset – -0,16 V
LLC – Level 3
PSU: Seasonic Focus PLUS GOLD 650W
GPU: AMD 7800XT
Memory: G.Skill @ 3900 18-22-22-42-64 Gear 1
Regards and thank you for any informative comments 🙂