Howdy,
I dropped all-core voltage to -30 offset in curve optimizer and I’m seeing temps go from about 85c max during a cinebench run to hovering at 89.5c-90c and occasionally going slightly over to 90.1c.
HWM graph during last section of runs: https://imgur.com/a/NJq4U0a
I saw a slight performance improvement with a final score of 819pts from a stock score of 808pts. Idle temps dropped from ~50c to about 38c, which is good. But I am unsure if it is “safe” to run this chip so close to 90c when under load. Would it be recommended to reduce the offset to maybe -25 and have it run farther from redline? Is that even how it works?
Thanks for any help.
2 Comments
What kind of cooler do you have? And did you try cleaning and reapplying some thermal paste, a non-cheap one preferably?
90°C is the target temperature for PBO on that CPU, and is what will be targeted if no other limits (frequency, voltage, power) are encountered first. Remember that CO is *not* an undervolt in the strict sense, it’s a way to shift the voltage/frequency target. There isn’t any relative reduction in voltage unless the core has achieved its maximum frequency, which generally won’t occur with heavy all-core workloads. In other words, it’s all normal.
Better cooling will allow you to eliminate the thermal throttle point and your CPU will then balance between power limits or the frequency cap, but the difference in most scenarios will be fairly minor unless you’re currently using a very basic heatsink.