[SOLVED] Amd 5950x. high temp low load, low temp high load

Aug 19, 2022
2
0
10
0
trying to understand the temp behaviour of my CPU. The cooling is a corsair h115i. ive set the pump to "extreme" setting, and from Hwinfo graph i can see that the pump is running constantly at approx 2700rpm.

so what puzzles me is that during normal light usage , i get temps spiking into 75C. most of the time at around 40C, then suddenly rises during light load increase. But when i run a stress test of 100% utilisation, temps dont go above 60C. The graph show quite high power draw during light load, but still lower than power at 100% load (which is expected). But temps go higher at light load, even though cooling power should be stable, with constant pump speed and coolant temp. Whats the logic?

Hope this works, logging of temps etc:
https://250bar.filemail.com/d/hkqumlaethelivv
 
trying to understand the temp behaviour of my CPU. The cooling is a corsair h115i. ive set the pump to "extreme" setting, and from Hwinfo graph i can see that the pump is running constantly at approx 2700rpm.

so what puzzles me is that during normal light usage , i get temps spiking into 75C. most of the time at around 40C, then suddenly rises during light load increase. But when i run a stress test of 100% utilisation, temps dont go above 60C. The graph show quite high power draw during light load, but still lower than power at 100% load (which is expected). But temps go higher at light load, even though cooling power should be stable, with constant pump speed and coolant temp. Whats the logic?

Hope this works, logging of temps etc:
https://250bar.filemail.com/d/hkqumlaethelivv
You are likely seeing temps on one most loaded core but Ryzen doesn't report temps per core so it looks like whole CPU is at that temperature. At light loads only one core is used so understandably that one gets hotter than others that can even be in sleep mode.
Ryzen Master reports more relatable temps, it reports it as if it was a single core processor "effective" temps and voltages
All together, quite normal.
 
trying to understand the temp behaviour of my CPU. The cooling is a corsair h115i. ive set the pump to "extreme" setting, and from Hwinfo graph i can see that the pump is running constantly at approx 2700rpm.

so what puzzles me is that during normal light usage , i get temps spiking into 75C. most of the time at around 40C, then suddenly rises during light load increase. But when i run a stress test of 100% utilisation, temps dont go above 60C. The graph show quite high power draw during light load, but still lower than power at 100% load (which is expected). But temps go higher at light load, even though cooling power should be stable, with constant pump speed and coolant temp. Whats the logic?

Hope this works, logging of temps etc:
https://250bar.filemail.com/d/hkqumlaethelivv
You are likely seeing temps on one most loaded core but Ryzen doesn't report temps per core so it looks like whole CPU is at that temperature. At light loads only one core is used so understandably that one gets hotter than others that can even be in sleep mode.
Ryzen Master reports more relatable temps, it reports it as if it was a single core processor "effective" temps and voltages
All together, quite normal.
 
Aug 19, 2022
2
0
10
0
You are likely seeing temps on one most loaded core but Ryzen doesn't report temps per core so it looks like whole CPU is at that temperature. At light loads only one core is used so understandably that one gets hotter than others that can even be in sleep mode.
All together, quite normal.
thanks for answer. still i dont get why one core should be hotter running at close to 100%, with neighbor cores sleeping, than when all cores are running at 100% together.

edit: i guess its probably because tha voltages didnt get lowered in the first case, so the power draw for one single core is higher (with neighbors sleeping), than the power draw for that same core when all cores are running at 100% and voltage do get lowered.
 
Last edited:
thanks for answer. still i dont get why one core should be hotter running at close to 100%, with neighbor cores sleeping, than when all cores are running at 100% together.

edit: i guess its probably because tha voltages didnt get lowered in the first case, so the power draw for one single core is higher (with neighbors sleeping), than the power draw for that same core when all cores are running at 100% and voltage do get lowered.
In multi core usage, load is dynamically distributed between cores switching between them before reaching maximum temp for most loaded core. When running on single core load stays on it. In theory, CPU algorithm should use best (gold) core but that may not happen if OS doesn't react like that.
 

ASK THE COMMUNITY

TRENDING THREADS