3 ms·
Far from exaggeration. In fact CPLEX and Express decided to go nuclear and remove their solvers from the benchmark set, so now we will be in the dark, regarding
by whatever1 8y ago
Far from exaggeration. In fact CPLEX and Express decided to go nuclear and remove their solvers from the benchmark set, so now we will be in the dark, regarding relative performance
http://plato.asu.edu/bench.html http://plato.asu.edu/bench.html
- shaklee3 8y agoThis is pretty sad. Unlike databases and other software, MIPS solvers have very good benchmarking standards, and though they don't cover every case, you can typically see which solvers are best for certain types of problems. Having met the Gurobi founders, they genuinely don't seem like they'd do something like this on purpose. Now Mitellman's only benchmarks for this category will be the slower solvers.
- leethargo 8y agoI guess there will still be third-party benchmarks, in particular as part of the papers that go along the updates to the MIPLIB sets of benchmark instances?