Closed AlexLipp closed 9 months ago
This error may simply be due to a manual scaling factor in the plotting section of runtime_benchmark.py
. This figure shows the results of the scaling factor of 1e-3 being removed from the SCS plotting commands. Now the runtimes are comparable in time in accordance with other tests.
Currently the runtime benchmarking script indicates that the SCS solver is three orders of magnitude faster than ECOS and GUROBI. This is highly improbable and likely indicates a problem with the benchmarking. A possibility is a conversion from ms to seconds somewhere in the processing chain.
funmixer
needs to be fixed.