Skip to content

Latest commit

 

History

History
34 lines (27 loc) · 966 Bytes

README.md

File metadata and controls

34 lines (27 loc) · 966 Bytes

LULESH GPU Optimization

This repository contains multiple LULESH variants and experiments focused on GPU acceleration, including CUDA, OpenACC, and OpenMP versions, along with benchmarking scripts and results summaries.

Contents

  • lulesh-cuda/: CUDA implementation and build files
  • LULESH-openacc/: OpenACC implementation
  • LULESH-codex-optimization/, LULESH-codex-translation/, LULESH-codex-translation-again/: Codex-assisted variants and logs
  • LULESH-intel-translation/: Intel migration tool and translated variants
  • HeCBench-lulesh-omp/: HeCBench OpenMP variant

Quick Start

Build in a specific subdirectory (example):

cd lulesh-cuda
make

Run the binary with an example problem size:

./lulesh -i 100 -s 200

Scripts

Run every implementation in this repo (each subdirectory's build/run flow) via:

./run_all.sh

Profile data transfers and runtime behavior with Nsight Systems:

./run_nsys_transfers.sh