Currently we have a lot of isolated profiling and example programs, but no automatic way to aggregate performance benchmarks. A unified benchmark suite would be useful to catch performance regressions (and track improvements)! Key functionality to cover:
- Integer primality testing and factoring
- Univariate polynomial arithmetic, GCD, factoring
- Multivariate polynomial arithmetic, GCD, factoring
- Linear algebra (determinants, solving, eigenvalues)
- Algebraic numbers
- Special functions
@thofma Do you have some good problems that we could use?
@defeo Do you have some good benchmark problems for FLINT's finite fields and their embeddings? I don't know enough about the applications to tell what kind of parameters would be interesting to profile for.
Currently we have a lot of isolated profiling and example programs, but no automatic way to aggregate performance benchmarks. A unified benchmark suite would be useful to catch performance regressions (and track improvements)! Key functionality to cover:
@thofma Do you have some good problems that we could use?
@defeo Do you have some good benchmark problems for FLINT's finite fields and their embeddings? I don't know enough about the applications to tell what kind of parameters would be interesting to profile for.