Related to Bug 1758 (and https://stackoverflow.com/questions/58366939/eigen3-boost-mpfr-segfault-when-alignment-is-disabled), it may actually not worth to use blocking algorithms for custom types, especially if sub-optimal cache access is less expensive than copying. This not only applies to GEMM/GEMV products but also, e.g., triangular solvers.
Yes that's one of the many items I have had in my mind for years...
-- GitLab Migration Automatic Message -- This bug has been migrated to gitlab.com's GitLab instance and has been closed from further activity. You can subscribe and participate further through the new bug through this link to our GitLab instance: https://gitlab.com/libeigen/eigen/issues/1759.