Phynaz
Lifer
- Mar 13, 2006
- 10,140
- 819
- 126
Correction:
My Liquid Temp is 27.8C (idle) - Using Chill Control
My Cores Temp fluctuates from 6.2C to about 12C (idle) - Using Overdrive
Those are some very chilly cores.
Impossibly cold even...
Correction:
My Liquid Temp is 27.8C (idle) - Using Chill Control
My Cores Temp fluctuates from 6.2C to about 12C (idle) - Using Overdrive
Those are some very chilly cores.
Impossibly cold even...
Correction:
My Liquid Temp is 27.8C (idle) - Using Chill Control
My Cores Temp fluctuates from 6.2C to about 12C (idle) - Using Overdrive
This is part of the conundrum with attempting to monitor and report the operating temperatures of piledriver-based CPUs.
We know the numbers are clearly wrong at idle, so how wrong are the numbers at full load?
Since it seems that there isn't an official response to this, do we have any hypothesis? Have you seen anyone attempting to discover an unofficial offset, for instance?
No ideas Best I could come up with was to use a thermal meter and see how hot the stock HSF got:
This is at stock clocks and stock volts. CoreTemp reported the CPU was ~55°C, my IR gun reported the external surface of the fins on the HSF were 51°C.
To me this is not possible. A mere 4°C delta from the CPU's silicon to the external surface of the HSF fin is just not a reasonable outcome.
The CPU silicon ought to be much much hotter than just 55°C in order to drive a 51°C temperature on the HSF.
The aprox temperatures are Coretemp and +15 C. So if your chip has 80C in load=real is around 95 C. Usually the PC take down CPU clocks or blackscreen.
It's exactly this kind of crap temp reporting combined with AMD's bs marketing of OC OC OC which is why I had no qualms recommending the guy with the toasted 8350 (difft thread) RMA it.
So while intel has chips that can properly report their temps and also throttle down when they hit their limit without issue, AMD cannot do either. AMD also has the added bonus of not telling people what the threshold limits are. Imagine that. So you don't know how hot it is, you can't check for throttling, and you don't know how much leeway you have. EVER ?? **BUT** AMD markets their chips to OC OC OC when they know fully well that is a recipe for disaster.
No ideas Best I could come up with was to use a thermal meter and see how hot the stock HSF got:
Anyone with ASUS motherboards, just use the Probe II application from ASUS AI Suite II. I have found it to be the most accurate for AMD temperature readings.
The CPU properly reports the temps, it is the software that is not calibrated correctly. Most application developers(coretemp etc) doesnt have the temperature and diode data needed to incorporate them in to their software and you get wrong/false readings from them.
have you tried the same with some Intel CPU to compare the difference from
the reported temp?
[FONT="]Hello guys,
I have an FX-9590 and been thinking about overclocking it. It will be my first time as I normally stick with default settings which meet my gaming needs. I have a question for you; how many of you have overclocked your FX-9590? And how far did you push it? Any tips for me?
Here are my specs:
CPU: FX-9590
GPU: HD7990
MB: ASRock 990FX Extreme 9
MEM: 16GB (4x4GB AMD Radeon RG2133)
Ill be playing World of Warplanes most of the times, and probably Bioshock Infinite in between [/FONT]
does anyone else smell bacon? why exactly did you come here? to troll?It's exactly this kind of crap temp reporting combined with AMD's bs marketing of OC OC OC which is why I had no qualms recommending the guy with the toasted 8350 (difft thread) RMA it.
So while intel has chips that can properly report their temps and also throttle down when they hit their limit without issue, AMD cannot do either. AMD also has the added bonus of not telling people what the threshold limits are. Imagine that. So you don't know how hot it is, you can't check for throttling, and you don't know how much leeway you have. EVER ?? **BUT** AMD markets their chips to OC OC OC when they know fully well that is a recipe for disaster.
That is what I use, but all you are really saying is that the software itself is free of errors, it still relies on the CPU correctly knowing and reporting the silicon temperature.
And my silicon temperature most certainly is not 8C when my ambient temperature is 28C.
The CPU clearly does NOT report the temps properly. Not at idle, not at half-load, and not at full load.
The problem here is a lot of people seem to have this completely unsubstantiated idea that the error at higher temperatures must be less than the error at low temperatures...that has never been proven for AMD piledriver CPUs.
I have, I did the same with an Intel CPU (2600K) using an NH-D14 and using the same IR gun and same temperature monitoring/reporting software and the delta between the exterior fin of the HSF and the temperature reported by the CPU was about 20C at full load.
(meaning CoreTemp and RealTemp reported the 2600K was at ~85C when the NH-D14 fins were at ~65C)
Yeah, it does...another factor I've been contemplating with :|