A gate-level simulator based on GPU, using CUDA C/C++.
This project contains two implementations of gate-level simulating, a simulator accelerated via GPU, and a serial simulator based on CPU for comparison.
Installation
This simulator uses boost library and CUDA parallel computing platform for GPU acceleration. You need to install all required dependencies and related tools to use it.
General dependencies such as a C++ compiler and build-essential are necessary for building.
Boost Library
Please visit Boost Library for more introduction and download link. You need to build and install it with the following commands.
cd ./boost_x_xx_x
./bootstrap.sh
sudo ./b2 install
CUDA Toolkit
Please check the GPU model, and make sure your GPU supports CUDA programming in CUDA-enabled products.
lspci | grep -i nvidia
Check the CUDA toolkit and corresponding driver versions in release notes,
and download the release of CUDA toolkit appropriate for your hardware configuration. Take CUDA-9.2 as an example.
sudo ./cuda_9.2_linux.run
To modify environment variables PATH and LD_LIBRARY_PATH.
vim ~/.bashrc
export PATH=$PATH:/usr/local/cuda-9.2/bin
export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda-9.2/lib64
Building from Source
To build serial simulator simply type ‘make’. Note that the include path for boost need to modify to your local path.
make
For the GPU based simulator it might be necessary to make some changes to the Makefile.
vim Makefile
Getting Started
This simulator has two steps (process and simulate). You can use the lib examples/GENERIC_STD_CELL.vlib and files in examples/default-rando/* from the sources as a simple example.
gate-level-simulator
A gate-level simulator based on GPU, using CUDA C/C++.
This project contains two implementations of gate-level simulating, a simulator accelerated via GPU, and a serial simulator based on CPU for comparison.
Installation
This simulator uses boost library and CUDA parallel computing platform for GPU acceleration. You need to install all required dependencies and related tools to use it. General dependencies such as a C++ compiler and build-essential are necessary for building.
Boost Library
Please visit Boost Library for more introduction and download link. You need to build and install it with the following commands.
CUDA Toolkit
Please check the GPU model, and make sure your GPU supports CUDA programming in CUDA-enabled products.
Check the CUDA toolkit and corresponding driver versions in release notes, and download the release of CUDA toolkit appropriate for your hardware configuration. Take CUDA-9.2 as an example.
To modify environment variables PATH and LD_LIBRARY_PATH.
Building from Source
To build serial simulator simply type ‘make’. Note that the include path for boost need to modify to your local path.
For the GPU based simulator it might be necessary to make some changes to the Makefile.
Getting Started
This simulator has two steps (process and simulate). You can use the lib
examples/GENERIC_STD_CELL.vliband files inexamples/default-rando/*from the sources as a simple example.