Speed decreases gradually after about 30 seconds
Speed decreases gradually after about 30 seconds
I agree with Gaidax, particularly when online gaming or other activities involve unexpected crashes.
I often install software before overclocking, as past experiences show that even minor updates can corrupt data.
It’s wise to stress-test at least to gauge how close your CPU is to instability.
By the way, I switched to Prime95 v26, which reduced my 85C temps to around 70°C. I’m currently running at 5.2GHz and haven’t adjusted the voltage yet. The unit still shows 1.36, possibly due to the setting.
I usually prioritize load-line calibration before experimenting with voltage adjustments.
I did encounter a "reason for throttle" labeled as "VR thermal warning." I can bypass it, but I’d like to understand the correct threshold values and see if cooling options improve.
The board I’m using is an ASUS TUF Z370 Pro Gaming.
Jonathanese, Higher Package Power leads to increased VRM temperatures. Lowering Vcore reduces power and core temperatures, as well as VRM power and heat output. VRMs usually handle higher temps than the CPU they supply, but they should stay below 100°C. Ensure the heatsink screws on your motherboard are tight. Also consider reducing Vcore. parani, Your notes didn't include ambient temperature, which is a significant factor. For fair comparisons, everyone should report their ambient conditions. Load also plays a big role. Check the second scale in my previous post—Cinebench runs at around 90% of TDP, giving surprisingly cool core temps. While some prefer Cinebench for stability tests, remember that all loads vary. I suggest using RealBench or AIDA64 with CPU and FPU tests together. Also, see this guide: Intel CPU Temperature Guide - CT
Notice that's what caught my attention. As someone who works with VRMs and expects high thermal performance, I expected much higher numbers. When the thermal warning appeared, I reviewed the motherboard's reported temps and found the peak was only 69°C on "temperature 2," likely just the TCase. All other readings stayed under 50°C, and the heatsinks seemed a bit above room temperature.
Jonathanese, Many monitoring tools often incorrectly identify or label various thermal sensors on motherboards, VRM, CPU, GPU, IGPU, RAM, HDD, SSD, and other peripherals. This is understandable given the confusion. However, it's important to clarify that there are no "Tcase" thermal sensors on retail processors. Tcase refers to a factory-only thermocouple measurement placed in the center of the external surface of the IHS using "Engineering Samples." It measures IHS temperature, not core temperature. Core temperatures are much higher because of sensor placement relative to heat sources and larger thermal gradients. With so much misinformation circulating, as an engineer you should grasp how this works. Detailed explanations are available in our Guide. You can find the section on CPU temperature here: Intel CPU Temperature Guide - Section 3 - CPU temperature.
Yeah sorry I got it wrong. The Tcase is the part they drill a notch into during manufacturing. What I meant was something like "Tpkg," which refers to the overall CPU temperature. This "Temp 2" board reading aligned closely with it, possibly a bit behind. That’s why I thought it wasn’t the VRM reading. But since it was the only one near the hot spot, I wasn’t sure if the board actually reports VRM temps.
And yes, I checked the article. Great info—thanks for sharing.
Package temperature refers to the hottest Core rather than the overall CPU temperature. Variations in sensor readings across labels often stem from strobe timing and whether data comes from the sensor or the slower PECI bus. You're correct... many boards don't monitor VRM temperature.