OpenMM Benchmark

Goal

This repo benchmarks the performance of serveral molecular simulation task on different GPUs. We use the benchmark script of the OpenMM library and run simulation with single/mixed/double precisions for eight different tasks, with both CUDA and OpenCL backend.

Results can be found in these CSV files:

• Single Precision
• Double Precision
• Mixed Precision.

Installation

Prerequisites

CUDA 10.0 (You can get it via installing Lambda Stack)

Anaconda

cd && wget https://repo.anaconda.com/archive/Anaconda3-2020.02-Linux-x86_64.sh && bash Anaconda3-2020.02-Linux-x86_64.sh

# Type yes to accept th license terms

# Press Enter to confirm the installation location /home/$USERNAME/anaconda3

# Type no when asked "Do you wish the installer to initialize Anaconda3 by running conda init?"

# Close the current terminal and open a new one
conda config --set auto_activate_base false && rm Anaconda3-2020.02-Linux-x86_64.sh

Create Conda Virtual Environment

In a new terminal:

conda create --name venv_openmm
conda activate venv_openmm
conda install -c omnia/label/cuda100 -c conda-forge openmm
conda install -c eumetsat expect
conda install pandas=1.0.3

Run Benchmark

git clone https://github.com/lambdal/openmm_benchmark.git

cd openmm_benchmark

conda activate venv_openmm

./benchmark.sh <GPU_NAME> <GPU_INDEX>

# Example
./benchmark.sh QuadroRTX8000 0

Compile Results

python compile_results.py