- Linux x86-64 (Ubuntu 20.04 or later is recommended)
- Python 3.8 with
libpython3.8.so.1.0 - A CUDA-capable NVIDIA GPU
- An NVIDIA driver compatible with the GPU
pip
Create a Python virtual environment and install the required packages:
python3.8 -m venv .venv
source .venv/bin/activate
python -m pip install --upgrade pip
python -m pip install -r requirements.txtrequirements.txt contains all required Python packages. The NVIDIA driver is
a system component and must be installed separately.
./rw_solver \
--input testcase20.sp \
--gt testcase20.csv \
--pred pred_testcase20.csv \
--tol 5e-5 \
--gpu 0,1 \
--ratio 0.1Arguments:
--input: input SPICE file or THUPG.mtxfile--gt: ground-truth CSV file--pred: initial prediction CSV file--tol: relative convergence tolerance; defaults to5e-5--gpu: one GPU ID or comma-separated GPU IDs; defaults to0--ratio: fraction of active nodes to solve; defaults to0.1
The --input, --gt, and --pred arguments are required.