ALCPU Forums

Hello,

I have the AMD FX-4100 running on the ASRock 970 Extreme4. CPU is water-cooled with the Antec Kuhler H2O 620(2 Fans push/pull). Core Temp isn't displaying the right temps. It keeps showing 11c at idle and skyrockets to 55c+ under load. I know none of these temps are correct but I cant figure out how to correct them. Adjusting the offset at idle to something relatively close is great for idle times. but then when you load up Prime95 the temps still skyrocket immediately to absurd ranges. I am comparing the temps to the ASRock Extreme Tuning Utility. I am sure that Utility is displaying the correct temps. Around 34-36c idle and a steady increase to 42-49c under load.

Can someone help me fix this?

Here are a few screenshots to help out.

Here is the register dump as well. Not sure if it will help or not but here it is.

Am I to assume no one has a fix for this? I know I am not the only one with this problem. Is it something that cannot be fixed? Do I just have to guess what the offset is and call it good? What should I do here?

Bump

bump

does anyone even read these posts anymore?

I'm sorry for the delay, but I doubt that there's anything to fix.
AMD thermal sensors are simply not working right in many occasions.
Core Temp simply reports what it reads from the CPU.

Hi,

I don't know if you found an answer, but I noticed your thread so I wanted to share my experience with you. I have an AMD Phenom II 1100T, on an ASUS board. I have used Coretemp, HWMonitor, SpeedFan, and PC Probe II. Assuming PC Probe is socket temp, then I found what was happening for me, at least so far today, was that my TJMax was wrong in Coretemp.

My TJMax was 90C, but AMD states 62C is the max temp. Assuming the problem is the incorrect TJMax, I discovered this was a 50.4F, or a 10.2C difference. I set the offset to 10.2 and it seems to be what the actual core temperature would be.

When I use coretemp while running prime 95, my temperature jumps to a safe level, but much higher than Speedfan and the rest. I believe this is because the place coretemp is taking the temperature is literally from the core. The others seems to get a temperature reading from the socket instead. In comparison, one is like taking a temp from the fry pan, while the other is taking the temp directly from the element. The temp from the frypan will always be lower than right next to the element.

That's my guess as to what and why, but I could be completely off too. Anyone have any thoughts on this?

hi guys,

i have the same problems it seems. so i started the day with everything running as norm (idle 39c, underload 52c) i went out for a few hrs so turn my machine off, when i got back according to asrock extreme tuner my cpu was idling at 61c and 61c under load. however with a quick manual check (hand behind my exhaust fan) the air was stone cold, like colder than it normally is under idle.

after looking around the net and double checking with razer game booster and then triple checking with speccy i had some worrying findings

ATXU = 61c
RGB = 39c (normal temp reading for RGB and ATXU)
Speccy = 29c

so for a quadruple check i thought i would reboot and check what my BIOS reading was

BIOS 39c

upon reboot to see if the problem had magically disappeared, i had even more of a shock.

ATXU = -1c (yes no joke -1c)
RGB = 39/40c
speccy = 29c

once again i went back to the BIOS and its still 39/40c. i mean what the hell?!?! i really worried as i only bought this machine 2 wees ago.

please bump this thread

my PC specs are as follows:
FX-6100 overclocked to sit at 3.9 constantly
2 x 2GB corsair dominator GT RAM
Asrock 970 extreme 4 MOBO
Gigabyte HD7850 OC 2GB

According to AMD the temp info. that Core Temp and HW Monitor (and likely other programs), are reading is not being properly processed and thus the temp readouts are incorrect.

It's obvious that the CPU temp at idle can't be below ambient. You also can not just use an fixed numeric offset to correct the data as this just skews all the temps and still provides incorrect temps.

A proper algorithm must be used to generate the correct CPU temps. There is nothing wrong with the AMD Deneb, Thuban, Zambezi or Vishera CPU temp sensors. The problem is the temp data being taken from the CPU is not being processed properly by these temp programs. These programs need to be rewritten so that they process the temp data correctly and then the temps will read as they should.

It's also important that people do not confuse the BIOS temps with CPU "core temps", (which is actual the Tcase temp on AMD processors), as the true "core temps" can not be read by software on AMD CPUs. The BIOS temps or "CPUTIN" or "CPU" temps in other programs is the CPU SOCKET temp read by a thermister on the mobo. This is NOT the "core temp" for which AMD states the max 24/7 operating temp. The socket temp runs 10-25C above the CPU "core temp" depending on the load on the CPU and the socket cooling. You should ignore the "CPUTIN" "CPU" or similar temps as they are no longer of relevant since Intel and AMD switched to CPU "core temps" years ago.

The TjMax temp is not the 24/7 temp in most cases. It is the Maximum recommend temp at which the CPU can be run for short periods of time as opposed to 24/7. You may experience errors by running the CPU at or close to TjMax.