AIDA64 CPU diode temp and the 20C offset

[Theagg]

I have read that the latest build of AIDA64 has 'fixed' the CPU offset temp, so now reads correctly, taking the 20C offset into account

I'm using the trial version which has the same build number as the fixed version yet my CPU Diode (and CP{U) temp, still reads identical to the temp in bios (which supposedly hasn't been fixed as of 1002, or am I wrong on that?). And reads the same as older software I have hanging around on my system, eg old versions of CPU-ID and HWinfo.

Which is correct then ? Has the bios (v1002) actually been fixed to read the temp correctly along with AIDA64. Or is something still amiss.

For the record, at a 4ghz OC, idle temp on CPUdiode reads 52-54C (Cooler Swiftech H220x)

[Theagg]

MeanMachine wrote:
Are you on the trial version of AIDA64.? I have the pro version and it allowed me to update.
CPUid HWMonitor was TMPINO was 65C now 45C. TMPIN1 is 27C as is TMPIN2.
My Bios is 1002 and reads 56C so no change there for me.

I have come to the same conclusion, trial version of AIDA is not updated, so I will go with the Tdie value in HWINFO64, 🙂

[Darth_Ender]

Using the Asus utilities that come from Asus and AMD, i see 1800x idle temps that match the bios. They appear to be fine when not overclocked. Everything sitting around ambient when idle and increase as expected under load.

When overclocked to 4Ghz (no change to bclk), the cpu temp reports both in bios and via sensors on windows and linux that the temp is around 50C idle and will quickly jump to 60 on any random activity and up to 85 under load.

The other temp sensors stay around ambient.

The sensors in both OS's seem to match what the bios shows.

I have an IR temp gun. When i get the motivation to care again, i'll just take the temp of the water block after the system has been idle for hours and compare to the reported cpu temp and then put it under load for a couple hours or so and check the temp of the block again. They should be very close to eachother by then. Put an end to the guessing game for me anyway. If two hours under load and the block isn't roughly the same temp as the cpu then either the software is lying or my block isn't making good contact (it is).

[Ocyrus99]

I'm on bios 0082 with a 1700x and I'm getting really strange temp readings. RM is reporting idle temps between 8c and 15c which obviously doent make any sense. Aida and hwinfo report from high 20s to mid 30s at idle which seems more reasonable. I'm clocked at 3.85 and am using p-states with an h110i. Bad temp sensor in cpu or is this happening to others?

Sent from my SM-G935W8 using Tapatalk

[Darth_Ender]

It's more likely that these developers are adding a fixed offset. Where as the actual offset needed is a function that needs to be applied.

So that while it's near idle, it's raw data is nearly accurate. But as the temp increases, the raw data requires more and more of an offset, approaching 20C by the time you reach what normally would be peak temps around 60C when the raw data reports 80.. something like that.

[Ocyrus99]

Thanks for the reply. If I use the sensor Elmor is suggesting in the OC thread (the CPU reading under c6h in HWinfo64) I get around 30 idle and 48-50 degrees at full load. I'll go with those numbers for now.

[Dazdigo]

I know there is a T Offset setting in the BIOS however I can't find any documentation on it. Can this be used to remove this 25 C offset that AMD put on their X line of AMD4 CPUs?