![]() ![]() Ideally, I would like to compute some distribution of runtimes for variable inputs. I can then post all results here for those interested. FORTRAN ZEROS OF BESSEL FUNCTION CODEI just installed this and ran my first Fortran code today! However, some help with setting up a proper benchmark would be very helpful so it is as fair as possible (and because I don’t know Fortran). I am very interested in benchmarking these further against the standard gfortran compiler. They now account for around 25% which is a noticeable decrease. After all, I started this development after I discovered that Bessel function calculations were over 75% of the runtime in my research software. With that said the performance gains are very real ( one example). SpecialFunctions.jl wraps several C and Fortran libraries (OpenLibm, OpenSpecFun, MPFR, etc.) which I do not differentiate in those benchmarks as they are just meant to signal the advantages of using Bessels.jl over SpecialFunctions.jl. The benchmarks at Bessels.jl are only meant to exemplify expected performance gains from users coming from SpecialFunctions.jl to Bessels.jl within the Julia community. ![]() Please feel free to split this topic into a new thread as I’d be happy to continue discussing and answer any questions. I’m sorry that this discussion has led to just comparing Fortran to Julia (as fun as that is). I sometimes feel open-source developments can be quite fractured which makes for a less cohesive experience especially for new users. ![]() I think these kinds of posts are very beneficial to the ecosystem and would love to have something like this in the Julia community. Thank you for the blog post Truly an amazing effort. ![]()
0 Comments
Leave a Reply. |
Details
AuthorWrite something about yourself. No need to be fancy, just an overview. ArchivesCategories |