ECL-APSP is an all-pairs-shortest-paths CUDA implementation of the Floyd-Warshall algorithm. As far as we know, it is faster than the fastest such codes from the literature. It operates on graphs stored in binary CSR format. Converters to this format can be found here.
Click on the following links to download pre-converted inputs from the SuiteSparse Matrix Collection:
Click on ECL-APSP_10.cu and ECLgraph.h to download the code. Note that ECL-APSP is protected by the license included in the beginning of the code.
The CUDA code can be compiled as follows:
nvcc -O3 -arch=sm_35 ECL-APSP_10.cu -o ecl-apsp
To compute the APSP of the file
This work has been supported in part by the National Science Foundation under Grant No. 1955367 as well as by equipment donations from NVIDIA Corporation.