Threadripper OC question for owners

LurkerLito

2[H]4U
Joined
Dec 5, 2007
Messages
2,719
I have my TR system running and OCed and stable but I'd like to know what software you are measuring your temps with?. I am currently using 2 different ones Aida64 extreme the latest beta and Coretemp. Both are reporting the CPU diode temps properly as far as I can tell, but as I am stress testing my OC settings (4Ghz @ 1.34v) it seems to be have weirdly. It start prim95 and it slowly rises to about 80c then it decreases a bit back down to around 72c then it rises again after a while back to around 80-81 then it drops again. Is this the CPU throttling? I am checking the clock and it's constantly just running at between 3.999 and 4.000 ghz. It seems to go about every 10 minutes or so in this kind of cycle, so I am kind of wondering if this is the CPU throttling or if it's prime95 doing something.
 
Doesn't Prime95 have a differing workload? As in, not the exact same test as the numbers get crunched and go on to the next internal test (per thread).
 
I don't really remember if prime95 varied the workload as it ran. It was like 5-ish years ago when I stress tested my current i7-3770k.

I ran it again at 3.8 and did a shorter stress test and it did a similar thing got to ~71c after 10 minutes then it dropped to 64c and repeated. So I don't think it's throttling but I just want to know if this behavior is normal because I thought prime95 kept a pretty much constant load on the CPU but I could be remembering wrong.
 
It's a constant load, but different work being done. Different work can mean different stress levels. Different stress levels mean different amounts of heat generated.

You still have a 100% load, but different work being done. As a guess, AVX and SSE2 instructions (just a guess) - and that difference in instructions being used to do the work means different temperatures.
 
It's a constant load, but different work being done. Different work can mean different stress levels. Different stress levels mean different amounts of heat generated.

You still have a 100% load, but different work being done. As a guess, AVX and SSE2 instructions (just a guess) - and that difference in instructions being used to do the work means different temperatures.

This is correct in our testing we have found a couple spikes in the load, the first one being around 33 minutes into testing
 
Back
Top