RowHammer PUF evaluation dataset

Dataset for RowHammer PUF evaluation. We provide the data in two formats: micron.zip and zentel.zip contain the raw log output from the FPGA tester that we used, and this logs are used by our data analysis code (see repo) json-parsed-logs.zip contain the same data but in a readable json dictionary, which we recommend using for further studies, where the key corresponds to the row is associated to one or more list of flipped positions (sorted by position). The Python script used to convert between the raw format and the json is availabe in parse_to_json.py and the whole procedure can be run with parse_logs.sh. The data has been obtained has been obtained from two Nexys Video boards, each having a 512 MiB DDR3 component, one from Zentel (A3T4GF40BBF) and the other from Micron (MT41K256M16HA-125:E). Temperature was not controlled during capture. The data has been obtained using a double-sided access pattern and with different parameters: n_act: number of total activations (500k, 1M, 5M, 10M) bg: data pattern used to initialize the target bank The folder structure mirrors these parameters <vendor>/<bg>/<n_act>/and for each combination we have different dataset: control contains control test, where hammering is not performed, where we expect to see no flips rand_rows_uniquniqueness dataset, one test per row for all rows in the bank, used to check similarity betweed different rows. rand_rows_rep/seq_rows_reprepeatability dataset, 100 tests for each row, used to confirm the repeatability of the measurements