We have released the following software in our group. All of these are under a BSD/MIT/Illinois license. We ask that you cite the relevant papers if you use any of our software in your research. Please contact us if you want access to something not listed here. Our Github organization is DependableSystemsLab.

Fault Injectors and Error Resilience

  • New LLTFI – LLVM Based Fault Injector for ML and regular programs
  • LLFI – LLVM Based Fault Injector (super-ceded by LLTFI)
  • PINFI – PIN Based Fault Injector
  • GPU-Qin – Fault Injector for GPUs
  • CrashFinder – Identify Long Latency Crashes through Static Analysis
  • Recov – Find locations to minimize checkpoint corruptions
  • LLFI-GPU – LLVM Based Fault Injector for GPUs
  • DNNFI – Fault injector for Deep Learning Neural Network Accelerator Platforms
  • LLFI-IPA – LLVM Based Fault Propagation Analysis for Multi-threaded Programs
  • Trident/vTrident – Analytical modeling for estimation of SDC Rates of Programs
  • TensorFI – Fault Injection for TensorFlow Applications
    TensorFI Version 2 – Support for weight fault injection, TensorFlow 2
  • LetGo – Roll-forward error recovery for parallel programs’ crashes
  • CrashRateEstimator – Estimating crash rate of programs from IR-level injection
  • SolidiFI – Fault Injector for Solidity Smart Contracts
  • GPU-Trident – Analytical model to estimate SDC rate for GPU kernels
  • Ranger – Error Correction for Deep Neural Network (DNN) Applications
  • TF-DM – TensorFlow Data Mutator for training data faults
  • NN-ensemble – Ensemble of Neural Networks to tolerate training data faults
  • TDFM techniques – Empirical study of training data fault mitigation techniques

IoT and Web


  • PCMeetingDashboard – A web application for PC meetings that interfaces with HotCRP. We developed it for the DSN’19 PC meeting, but it was also used in HPDC’19, DSN’20, DSN’21, SRDS’22 PC meetings.