Ecosyste.ms: Awesome

An open API service indexing awesome lists of open source software.

https://github.com/l30nardosv/reproduce-parcosi-moleculardocking

Reproducing paper: "Benchmarking the Performance of Irregular Computations in AutoDock-GPU Molecular Docking"
https://github.com/l30nardosv/reproduce-parcosi-moleculardocking

autodock-gpu cpu cuda gpu molecular-docking molecular-docking-scripts opencl paper reproducible-research

Last synced: about 2 months ago
JSON representation

Reproducing paper: "Benchmarking the Performance of Irregular Computations in AutoDock-GPU Molecular Docking"

Host: GitHub
URL: https://github.com/l30nardosv/reproduce-parcosi-moleculardocking
Owner: L30nardoSV
License: cc0-1.0
Created: 2021-05-10T09:10:27.000Z (over 3 years ago)
Default Branch: master
Last Pushed: 2022-11-24T10:00:02.000Z (about 2 years ago)
Last Synced: 2023-06-05T15:35:28.484Z (over 1 year ago)
Topics: autodock-gpu, cpu, cuda, gpu, molecular-docking, molecular-docking-scripts, opencl, paper, reproducible-research
Language: Shell
Homepage:
Size: 283 KB
Stars: 1
Watchers: 1
Forks: 2
Open Issues: 0
Metadata Files:
- Readme: README.md
- License: LICENSE

# Reproducing paper: _Benchmarking the Performance of Irregular Computations in AutoDock-GPU Molecular Docking_

This repository contains the scripts and additional files for reproducing the results presented in the paper accepted at Parallel Computing:

https://doi.org/10.1016/j.parco.2021.102861

## Instructions

### 1. Program preparation

Clone the [AutoDock-GPU v1.3](https://github.com/ccsb-scripps/AutoDock-GPU/releases/tag/v1.3) (used in this paper):

```
> git clone --depth 1 --branch v1.3 https://github.com/ccsb-scripps/AutoDock-GPU.git
> cd AutoDock-GPU
```

Set environmental variables (adapt the paths to your systems' installation):

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

Clone _this_ repository:

```
> git clone https://github.com/L30nardoSV/reproduce-parcosi-moleculardocking.git
```

Compile AutoDock-GPU and rename the produced binaries conveniently (for the scripts to work):

```
> make DEVICE=OCLGPU NUMWI=32
> mv bin/autodock_gpu_32wi bin/autodock_gpu_32wi_ocl
... Do the same for NUMWI = {32, 64, 128, 256}

> make DEVICE=GPU NUMWI=256
> mv bin/autodock_gpu_256wi bin/autodock_gpu_256wi_cuda
... Do the same for NUMWI = {32, 64, 128, 256}
```

Move above binaries into the test folder:

```
> cp bin/autodock_gpu_64wi_ocl reproduce-parcosi-moleculardocking/
... Repeat for all OpenCL/CUDA cases above
```

### 2. Performance evaluation

Input dataset is provided as a git submodule. Clone [that repository](https://gitlab.com/L30nardoSV/ad-gpu_miniset_20.git) automatically:

```
> ./prepare_inputs.sh
```

Evaluate the performance for different OpenCL work group sizes:

```
> ./evaluate_numwi.sh
```

Evaluate the impact of enabling the _autostop_ option:

```
> ./evaluate_autostop.sh
```

Evaluate the impact of the _heuristic_ option:

```
> ./evaluate_heuristics.sh
```

Evaluate the impact of combining of both _autostop_ and _heuristic_ options:

```
> ./evaluate_auto_plus_heur.sh
```