The order to execute the code in this sub-project:

1. Code_preprocessing
  To process the raw traces.
    Raw/Raw_DN_****.zip => Processed_HDF5/Processed_DN_****.hdf5
  To check whether there were abnormal traces (fortunately none.)
  To copy the input/output texts of SHA3-512 from the zip file.
  All-in-one script: pre_processing.sh


2. Code_intermediate_values
  To calculate the intermediate values from the I/O texts. 
  All-in-one script: get_values.sh


3. Code_detection_R2
  To find the R-square value between the target sample of each clock cycle and each target intermediate byte.
  All-in-one script: readme.sh


4. Code_extract_ics
  To determine the interesting clock cycle set of each target intermediate byte.
  All-in-one script: ics_script.sh


Please follow the order above since they are dependent on each other.

We executed these codes on a 32-core server with 252 GB memory. However, I believe the actual memory usage won't be over 10 GB. On this machine, we relied on Intel® oneAPIi math kernel library to (automatically) support some multi-threaded calculations in NumPy, but I also believe the code can be executed with one single thread.