hmm i tried a way to overclock the CPU, and now when i look in Speedfan, the "CPU" goes up to about 80 Celcius, though the 4 cores tend to be much lower, around 50-60 Celcius. With a little research, i found that the G0's "TjMax" is 100 Celcius, while on CPU-World.com, they say the CPU's maximum temperature is 60.3 Celcius, so which one is the real CPU's maximum temperature? And should i trust the "Core (0,1,2,3)" temps more than the CPU temp itself?
Briefly, the TjMax is the value for an internal package temperature (Tjunction) above which the CPU starts throttling, expressed in Celsius degrees (please write Celsius
, not "Celcius"), while the TCaseMax is the one you read about on CPU-World, and it's a statistical max value for safe operation of an "external" temperature for the CPU package (the TCase), usually taken by a thermocouple just under the heatspreader: so both are real, and TjMax is usually much higher than the TCaseMax value, but more important there's no direct relationship between those two.
From a practical standpoint, on the Intel platform you may know just something about TjMax through a DTS circuit, as there are no sensor to directly read the TCaseMax accessed by the motherboard, so it's a common sense among overclockers to do not trust any of the mobo sensors. You can read something more here
Unfortunately you can't directly know even Tjunction, but just the DTS readings, which converts the analog signal from an on die diode to digital values, and then reports them as a relative offset from an absolute 0, when the PROCHOT# function is activated: so those figures are strictly not Celsius degrees, but relative measures on an absolute scale.
Summarizing, when you're oc'ing, you have to trust mainly DTS readings, as most mobo called "CPU" sensors are not so trustful.
Use RealTemp to double check SpeedFan figures, in doubt, trust RealTemp rather than SpeedFan: it may worth a reading here
As a rule of thumb, people usually want to stay under 65-75 degrees for DTS figures under load, depending of the stress-testing program/methods used, and providing the relevant readings are properly calibrated.
Regarding your reported CPU temp, 80°C would look like a bit on the high-side, so maybe you're overheating the mobo, rather than the CPU (even a mobo can burn, not just the CPU): I would try to blow the mobo around the Ninja with a small fan, in order to see whether it changes.
Maybe he has actually checked it? I dunno, I've never heard of such limitations for mem sticks.
There are chances it just doesn't sport any (IMVHO useless) automatic XMP profile (which should be just for Intel platforms, not AMD ones).