The unusual multiplier boundary?
The unusual multiplier boundary?
Hi.
I've been experimenting with OC on my 12700K CPU and Z690 Hero motherboard for a week. I tried various methods and finally settled on a configuration that works well. I used 1-4c/53, 5-8c/52, 3c@53, 5c@52, and e-cores@40 with a cache size of 45. Everything functioned properly, keeping temperatures below 90°C under full load. During BIOS tuning, I adjusted the offset in v/f, but after rebooting I encountered an unusual issue. The highest boost setting now appears as the lowest, which is why I chose 52 for this setup. I considered IA CEP, but it doesn't seem to be the problem here. C-EP is enabled, yet changing it to 50/49 still leaves the max boost at 49. I'm not sure if any known settings are causing this. I've also tried toggling c-states on and off, but couldn't resolve it. Thanks!
The C states need to be activated for Intel Turbo Boost to function properly. Verify the BIOS settings. Avoid using an AUTO setting for the C states and ensure they are enabled. HWiNFO can also help verify if idle cores are utilizing the core C7 state. ThrottleStop is useful for displaying this data. This behavior is intentional, not a malfunction. Intel designed Turbo Boost to operate correctly when C states are off.
Thank you for your message. I understand you're using Linux and missing features like HWinfo that are available on Windows. Tools such as HWinfo are limited here. If I turn on C-states other than C1E, I'm restricted to 8x performance, and I'm unsure if this is specific to Linux or a kernel issue. Everything seemed normal before enabling C-stats. I've identified the problem: I set an IA VR limit at 1.4v for safety, but CPU-X (checked via CPU-Z) shows spikes above 1.37-1.38v, which may be necessary for higher multipliers. I'm trying to resolve this.
I need to confirm the voltages somehow.
Can I tolerate invisible spikes above 1.4v?
What is the height of these spikes? Should I increase the IA VR to check, perhaps aiming for 1.42v for a short duration?
It seems acceptable given Intel's recommendation of 1.55v.
I'll research the C-state issue once I finish the overclocking process.
I might prefer a different C-state than C1E.
Thanks again for your help.
The issue was CEP configured at 1400mv. Even when vcore didn't reach 1400, there were small spikes above 1400 that got cut off by the monitoring system. Changing CEP to 1425 resolved the problem. Now I'm running at 1,3 idle and 1.27 at full load.