Question about my CPU temperatures, socket vs package AMD FX 8320

MickDeRaad

Reputable
Jan 30, 2015
22
0
4,520
Good evening,

I have a question about my temperatures. Please see picture below.
I have an FX-8320 at 4.1 GHz + an EVO 212. Motherboard is an MSI Gaming 970 AM3+.

I also have plently of airflow in my case 3x120mm in, 3x120mm out + a small fan pointing at my northbridge (got really hot before, now temps are "ok" there).

Now what I am really worrying about it the big difference in package temperature and cpu temperature as seen by bios (tmpin0). I have set the bios to have it aim at a temperature of 45c (standard was way higher - but my package would get over the limit of me being comfortable with it).
So, while my motherboard things the CPU is at a mere 47c under heavy load, the actual package temperature is hitting the 62c limit. Should this be happening? Could I have applied too much cooling paste? (noctua NT-H1). HWmonitor also seems to show a wrong CPU-fan speed since I put in the bios to run at a minimum of about 53% Please inform me. Also, inform me if I am wrong about the tmpins.

Thank you!

Untitled.png


http://
 


Yeah temps are sure funny. Still wondering about the package and socket temps though - most people seem to say that the socket gets hotter than the package. Am I wrong? something should be wrong here since my motherboard goes after the socket temperature while the package is not even to be seen in my bios.
 
???? seem like me you could read 10,000 threads on this and not get a real straight answer to it .. lol I guess that thermal margin shows how much head room you have under max temp of the chip ?

my deal is if it running full load and not throttling/shutting down/ bsoding its good ??? that's the best I know
 


Well, not if it is damaging my cpu and reducing its lifetime. A few months ago (before the heatwave hit Sweden and I had to reduce it to 4,0Ghz) I had my processor running at 4,4 Ghz and it was not hitting such high temperatures. Seems fishy once again.

My question remains, could this be the effect of too much thermal paste?
 
it could too much too little ?? but any overclock or voltage is going to do the same also it loks from your first post the cpu fans are just going to 38% ?? maybe they need to be ramped up some more ??

that overdrive seems to show your just around 6c below max ?

you know the thing is theres a heck of a lot of guys posting here at toms over this and these chips - just in that I find it funny overall


I just noticed in the post with the overdrive the hw monitor shows the fans at only 720 rpm's ???? seem they should be around max rpm's like 1800 2600 ???
 



Take some things I said with a grain of salt. Apparently my costom fan-settings had been changed to stock which means that the cpu fan (fan0) was "aiming" at a max temp of 70c. I put it on max 45c like I had before so it can compensate for the way higher package temperatures.

as for fan 1 and 2, one of them is a little screamer aiming at my northbridge (put it there with beatiful zipties between my gpu and cpucooler)- and the other is my rear exhaust fan. My Evo 212 is aiming upwards at two 120mm fans pulling air out. Got intake 2 on the front, and one under. should be enough, aight? :)

Will test again now with changed configurations but it should still show about the same situation of the package being far higher. Why does my motherboard not go after this?
 
Changed fan settings in bios and after 10 minutes of small FFT it gave me this. Besides the CPU core voltage, don't mind the rest as HWmonitor shows the wrong voltages.

I gave close attention to AMD Overdrive and I noticed a few drops in frequency (not noticed by HWMonitor) sometimes going under 20 (!) megahertz on core 1 but only 2-3 times without changing the MP for only a split quarter of a second maybe. My cores are however changing their frequency ever so slighty ranging between 4090~4120 every 10 seconds or so. Is this normal while stresstesting?

Also edit on the fans: fan 2 is the 90mm screamer while the fan1 is one of the exhaust. My MB does not have any more fanconnections on the motherboard, the rest is connected on normal 12v.

Untitled3.jpg
 
''4090~4120 every 10 seconds or so. Is this normal while stresstesting?'' I would think so what it would be is total droping out to near idle speeds [??]

hard for me to tell is that new overdrive now showing 16.2 below max ?? if so looking way better now ?? I could go for that
 


Yep looks better. But still, that I HAVE to change the fans MANUALLY to aim at "45c cpu temperature" (which is the lowest I can get since AMD cpus apperently have no feeling of temperature under 40) gives me a tide of worrying. I do have more headroom for more OC since temps are 16c under thermal limit (70c according to AMD, but from what I've read it's more like 62c). This is at 70-80% fan speed on the cpu.

I still would have liked my motherboard to follow package temperatures instead of the socket ones. If I leave it at stock settings my motherboard would fry the cpu without even knowing it itself. Sad.

Thanks for help though, do you have any more things I should know? :)
 


On some games I play the extra performance is nice to have, but not "needed". For instance in Cities Skylines. But for now the bummer is my GTX970 not having more videoram since it limits out on it.