Checks · Windows

What to do when a Core Temp reading is missing

These checks come from problems people actually report: a blank core, a question mark, an antivirus block, a language file that does nothing, and a 64-bit plugin that will not load. The steps assume Core Temp 1.21. Get that build from the download page once the installer files are attached.

• No temperature, or a question mark
• Antivirus removes or blocks the file
• The language pack does nothing
• A plugin fails on 64-bit Windows
• The overheat warning keeps returning
• The log file lands in the wrong folder
• Habits that keep the reading useful

No temperature, or a question mark

A question mark means that sample was not marked valid. An empty row is a different failure, but the first checks are the same.

1. Confirm the processor is Intel, AMD, or VIA x86. Core Temp does not read a motherboard diode in its place.
2. On an older AMD K10-class chip, expect one package temperature, not a true set for each core. Phenom and related Opteron parts are the usual case.
3. On some 45 nm Intel Core 2 chips, temperatures below about 30°C or 40°C are not reported. A question mark at idle can be the sensor, not a broken install.
4. If a Ryzen system showed temperatures before sleep or a reboot and then froze or went blank, install 1.20.1 or newer. That correction is included in 1.21.
5. Close other programs that also load a CPU sensor driver, then start Core Temp again. Two drivers competing for the same registers is a known way to get a dead reading.
6. If the window says "Not responding" at startup, that hang was corrected in 1.20.1. Update before you keep ending the process.
7. Still stuck: note the CPU model from the top of the window and compare it with the sensor method. A chip with no digital thermal sensor will not grow a row for each core.

Antivirus removes or blocks the file

Core Temp includes a driver so it can read CPU registers. Scanners have called that driver a trojan in the past, including older Bitdefender detections that were later withdrawn. Version 1.21 also reduces the impact of an older 64-bit driver issue, JVN#23328936.

1. Check the file name and version before you restore anything. A repack from an unrelated download site is not this program, and it has installed extra software for other people.
2. Prefer the portable zip once it is attached on the download page. It skips the installer screen that offers a Goodgame Empire shortcut.
3. If your scanner quarantined the file, send it to that vendor as a false positive only after you are sure of the source. Do not turn off the whole antivirus product to finish a setup.
4. Read the trust page for what the author has said about keyloggers and spyware, and for what this site still has not attached.

The language pack does nothing

People download a language file, place it next to the program, and the menus stay in English. Temperature logging can still work while the language does not.

1. Match the language file to the same Core Temp version you are running. A file from an older build often does nothing on 1.21.
2. Put the file where that version expects language files, then restart Core Temp. Copying it into a folder you made up, such as a new "languages" directory, is the report that failed.
3. If the menus stay in English, use the English interface. The readings do not depend on the language pack. There is no separate temperature mode hiding behind the translation.

A plugin fails on 64-bit Windows

The 64-bit build is less friendly to older plugins. Shared memory is often the part that still works when a DLL does not load.

1. Use a plugin built for the same bitness as Core Temp. A 32-bit DLL will not load into the 64-bit program.
2. On Ryzen, if plugins show a wrong Tj. Max while the main window looks right, update to 1.20.1 or newer. That build corrected Fahrenheit Tj. Max passed to shared memory and plugins.
3. If you only need the numbers, read the shared memory fields listed on the features page instead of forcing an old gadget to start.

The overheat warning keeps returning

Early builds could open a new "CPU is overheating" message on every sample. You can turn the warning off, raise the limit, or let the program hibernate or shut down once.

1. Open the overheat settings and set one limit you actually want, in Celsius or Fahrenheit, matching the scale the window is using.
2. Choose hibernation or shutdown if nobody is at the PC. A message box alone does nothing while you are away.
3. On 1.21 the message is the top-most window. If you never see it, you are on an older build, or another full-screen exclusive program is covering the desktop.
4. Check the Windows Event Log if you turned on overheat events. That record remains after the message is closed.

The log file lands in the wrong folder

Version 1.21 corrects a custom log path that was ignored when the program started. Older builds could keep writing to the default folder after you picked another one.

1. Update to 1.21, then set the log folder again and restart Core Temp so the startup path is the one you chose.
2. Pick a logging interval that is separate from the temperature read interval. A very short interval creates a large text file quickly.
3. Open the text file in a spreadsheet when you want a curve. The grapher add-on is optional and is not required for the log itself.

Habits that keep the reading useful

• Compare cores with each other under the same load. A single hot core, as on the Core i7-2600K screenshot where one core peaked at 65°C, tells you more than the average.
• Read Tj. Max from the window before you decide a number is too hot. 95°C on the 9950X screenshot is the junction limit for that chip, not a ceiling that applies to every processor.
• Do not add a large sensor offset to force the number you hoped for. Older Core Temp builds removed offset once because it made the reading less honest, and later builds have been careful with it.
• Keep one sensor tool loaded. Running Core Temp beside another driver-based monitor is a common cause of a blank window.