Skip to content

Latest commit

 

History

History
73 lines (50 loc) · 1.56 KB

USAGE_GPU.md

File metadata and controls

73 lines (50 loc) · 1.56 KB
Raw

Benchmarking AutoDock-GPU

1. Clone AutoDock-GPU and scripts

git clone --depth 1 --branch v1.1 https://github.com/ccsb-scripps/AutoDock-GPU.git
cd AutoDock-GPU

git clone https://github.com/L30nardoSV/reproduce-ia3-2021-moldocking-vector.git

2. Apply git patch to AutoDock-GPU v1.1

Regarding execution times, AutoDock-GPU v1.1 only prints full program run-time.

In order to additionally print docking run-times, apply the following git patch:

cp reproduce-ia3-2021-moldocking-vector/adgpu_v1.1_runtimes.patch .
git apply --verbose --ignore-whitespace adgpu_v1.1_runtimes.patch

Expected output once patching:

Checking patch host/src/performdocking.cpp...
Applied patch host/src/performdocking.cpp cleanly.

3. Set environment variables for AutoDock-GPU

Key ideas:

  • Include path: contains "CL" folder, which in turn contains OpenCL header files (e.g., cl.h, opencl.h, etc)
  • Library path: contains the libOpenCL.so file

Example for GPU:

export GPU_INCLUDE_PATH=/usr/local/cuda/include
export GPU_LIBRARY_PATH=/usr/local/cuda/lib64

Example for CPU:

export CPU_INCLUDE_PATH=/opt/intel/opencl/include
export CPU_LIBRARY_PATH=/opt/intel/opencl

4. Compile and copy binaries into main folder

For GPU:

make DEVICE=GPU
cp bin/autodock_gpu_64wi reproduce-ia3-2021-moldocking-vector/

For CPU:

make DEVICE=CPU
cp bin/autodock_cpu_16wi reproduce-ia3-2021-moldocking-vector/

5. Run benchmarks

cd reproduce-ia3-2021-moldocking-vector
./prepare_inputs.sh
./evaluate_autodock_gpu.sh