Replies: 2 comments 2 replies
-
Thank you for pointing this out 👍 HPIPM was likely not installed on the machine I last run the benchmark on. (While most solvers can be installed straight from the conda In the meantime, you can also run the benchmark on your machine, and feel free to report any issue you may encounter. |
Beta Was this translation helpful? Give feedback.
-
Hi, first of all, thanks for adding also HPIPM to the solvers and benchmarking it.
Currently HPIPM does not support a termination criterion based on duality gap, but this could be easily added if requested.
I understand that adding preprocessing steps such as condensing or a structure detection is outside the scope of this benchmark and would also raise the question of picking the best formulation for each solver for each specific problem. There is already a "Known limitations" section that already contains factors affecting performance to a much smaller degree such as thermal throttling, so the issues above could definitely find space there too. Thank you, Gianluca |
Beta Was this translation helpful? Give feedback.
-
Hi,
First of all, nice work on this MPC benchmarking test framework for QP solvers!
I noticed that HPIPM is mentioned in the list of QP solvers, but I do not see it in the benchmark results, why is that?
Thank you,
Rien
Beta Was this translation helpful? Give feedback.
All reactions