• Apparently there was a bug in Prime v30.4 & 30.5 that could lead to incorrect fatal errors. Updated to Prime 30.6 build 4, which should have fixed this issue according to the developer (see here: https://www.mersenneforum.org/showpost.php?p=577388&postcount=274)
• Added a final summary when terminating the script
• The script now prevents sleep/hibernate/standby while it is running

• Added an auto option for the runtimePerCore setting. With this option all FFT sizes of the selected preset will be tested for one core before it continues to the next core, where it will also test all the FFT sizes, etc.
  For y-Cruncher and Aida64 the "auto" mode uses a fixed runtime of 10 minutes per core due to the lack of detectable progress.
• You can now also enter a custom FFT range for the FFTSize setting in the Prime95 section, e.g. something like 36-1344. This eliminates the need to use the CUSTOM mode to define a custom FFT range (although you can still use the CUSTOM mode of course).
• The verbosityMode setting has been renamed to logLevel
• The script should now also work if started as an administrator
• Fixed a couple of bugs with case sensitivity and trailing spaces in the config file
• Rewrote the error detection for Prime95 log files, hopefully with more accurate detection
• Improved error detection when CPU usage was too low

The script should now also work when started as an administrator.
Fixed a bug with the name of the log file.

This should be the final release before v0.8.1.0, i.e. if no other bugs are found, no changes will be made.

Fixed a problem when settings in the config.ini had trailing white spaces.

Fixed some problems with the case sensitivity.

Release Candidate 2 for version 0.8.1.0

• There's now an auto setting for the runtimePerCore, where it will test all FFT sizes before continuing to the next core
• It's now possible to define a custom FFT range in the FFTSize setting (e.g. 8-24)
• Improved error detection when CPU usage was too low
• Changed the verbosityMode setting to logLevel and added two more options

• Updated Prime95 to 30.5 build 2
• Support for Aida64 and Y-Cruncher! You can now define which stress test program to use (Note: see the readme.txt for more infos on Aida64)
• Restructured the config file to support Aida64 and Y-Cruncher. Old config files will not work!
• Added a new "suspendPeriodically" setting, which is activated by default. This setting tries to simulate load changes by periodically suspending and resuming the stress test program
• Added a new "coreTestOrder" setting, which defaults to "Alternate" (for more than 8 cores/2 CCDs) or "Random" (max 8 cores/1 CCD). You can also define a custom order for the sequence in which the cores should be tested (e.g. "5, 7, 5, 1, 0, 7, 4")
• Added more presets for Prime95: "Moderate", "Heavy" & "HeavyShort". See the config file for additional explanation about these presets
• The priority of the stress test process is now set to "High" to prevent other processes from "stealing" processing power and produce false alarms due to low CPU usage
• Starting a stress test program will not steal the focus anymore (or give it back immediately to the last opened window)
• The script will now try to close the stress test program after pressing CTRL+C, if it is still running and using CPU processing power
• The title of the terminal window is now set to "CoreCycler"
• The approximate CPU frequency is now visible in the verbose output (but unfortunately it's not as accurate as e.g. HWInfo64 or Ryzen Master, which is why it's not in the main output)
• All logs are now in the /logs directory
• Added an "analyze_prime95_logfile.ps1" script to the /tools directoy, which can be used to determine the time it took for all FFT sizes to appear at least once for each iteration within a Prime95 log file
• Added "CoreTunerX" to the /tools directory. See https://github.com/CXWorld/CoreTunerX of what it can do

Still intended for beta testing only.

• Updated Prime95 to 30.5 build 2
• The analyze log file script now autodetects if one or two threads were used
• Fixed a bug where Aida64 did not close correctly
• Fixed a bug where the test run for a core was not immediately stopped when an error occurred. This could cause a loop where an error was thrown over and over again until the runtimePerCore was reached

Still intended only for beta testing.
Some more bug fixes and improvements, hopefully without introducing new bugs.
You can now define your own custom testing order by providing a comma separated list to the coreTestOrder setting.

Still intended only for beta testing.
Fixed a couple of issues, especially the affinity not being set correctly for 2 threads and the Y-Cruncher window not recognized correctly (at least I hope so).