Artifact for "Robust but Relaxed Probing Model"

This artifact contains the extended PROLEAD version for the evaluation under the RR d-probing model.

Getting Started

We use the nix package manager (https://nixos.org/). After installing nix, simply type:
nix-shell which will automatically install all required dependencies.
Afterwards, PROLEAD can be used within the nix shell. To compile run make clean release.
Afterwards, run ./release/PROLEAD to start the evaluation of a basic example.

Structure

The artifact is structured as follows:
* The examples directory contains all experiments we reported in the paper (see Table 4).
* The inc directory contains all header files of PROLEAD.
* The src directory contains all source files of PROLEAD.
* The ut directory contains unit tests. Specifically, we test the evaluation of gadgets given in the full directory.
* The design.v and config.json files serve as default design and configuration file evaluated by running ./release/PROLEAD.
* The library.json file contains the standard cell library used by PROLEAD.
* The shell.nix file contains the configuration for nix. If you need more information about PROLEAD's dependencies, have a look.

Reproduce the Paper Results

The paper results are given in the examples folder. Every example contains a robust and a relaxed directory.
* The robust directory contains the evaluation results under the robust d-probing model.
* The relaxed directory contains the evaluation results under the RR d-probing model.

Further both directories contain the following files.
* The run.sh file serves as eecution script to evaluate the design.
* The config.json serves as a configuration file.
* The Report.dat file visualizes the result of the evaluation.

To evaluate a specific design, navigate to the folder where the run.sh file is located and type ./run.sh.

Caveat

• We further optimized PROLEAD after the paper deadline and reproduced the results. Hence, the results in this artifact may slightly differ from the paper.
• Independent of the changes, we remark that the exact runtimes may highly depend on the applied machine. Please see Table 4 for an intuition of the requirements of a particular experiment.