Power usage of the CPU varies with temperature levels.
Power usage of the CPU varies with temperature levels.
I just ran a test for fun. I used prime95 for 30 minutes twice—once with all fans at full speed, the other at 100RPM. The coolant temperature difference was around 2K. I kept voltage at 1.3V and CPU speed at 4.2GHz, then checked the power usage. The first run showed 138.79W, the second 141.65W. These values came from HFInfo64, measured after the coolant settled. My question: Are these differences real or just due to limited precision in the readings? I'm using an R7 3800X and ran a custom version with FFT sizes of up to 4K.
As temperature rises, resistance increases, leading to greater power usage needed for the same output.
The delta T reflects the water temperature. Since you didn’t log the CPU temperature, they should still be proportional, right?
I didn't expect it to be so straightforward to quantify. I assumed a roughly 1% power gap around 10°C.
The water delta is also affected by the surrounding room temperature. When transistors become hot, they dissipate energy, which can lead to a marginally increased current draw and a higher measurement. Running cooler reduces leakage, resulting in lower power consumption.
I understand that, my intention was simply to say the CPU needs to be 2°C hotter when the water is 2°C warmer under identical heat conditions