GitHub - jeffreyscholz/GPU-Traveling-Salesman-Solver: Uses the two-opt algorithm as a heuristic for solving TSP instances. You need CUDA and an NVIDIA GPU to run this.

jeffreyscholz / GPU-Traveling-Salesman-Solver Public

Notifications You must be signed in to change notification settings
Fork 1
Star 0

Uses the two-opt algorithm as a heuristic for solving TSP instances. You need CUDA and an NVIDIA GPU to run this.

Notifications

Name		Name	Last commit message	Last commit date
Latest commit History 7 Commits
34-99.tsp		34-99.tsp
LICENSE		LICENSE
Makefile		Makefile
README		README
two_opt.cu		two_opt.cu
util.cu		util.cu
util.h		util.h

Repository files navigation

compile:

make -j2

run:

./two_opt tsp_file

(-j2 option speeds up compilation on multicore machines)

Chances are you will have to change the NVCC option in the makefile to match
the location of your nvcc binary. You will also probably have to change the
NVCCARCH option to match your version of NVCC. This code will probably not
work on versions prior to 3.0.

The tsp format must be in DIMACs format. You can download instances here:
http://dimacs.rutgers.edu/Challenges/TSP/download.html