If you have ideas and opinions, go see: Petsc-future: Languages (#645) · Issues · PETSc / petsc · GitLab

The DOE codes (PETSC, Trilinos, Sundials) have 100s of person-years behind them and are proven to be scalable on top-5 computers. Having these things easily accessible from Juila can’t be bad.

[image] mohamed82008: I mean do we really need PETSc in Julia? Seems like we just need to port any distributed data structures and functions they have to DistributedArrays.jl if they are not already there. Then we get the rest of the features for free because of the generic nature of the relevan…

Wrapping the general routines for easily building parallel matrices (the core Petsc vec/mat functionality) and handling distributed meshes / automatic partitioning of unknowns is the feature that I miss. I agree that if that were done and robustly maintained we might see other Julia libraries be bui…

[image] isaacsas: That said, I haven’t looked at the Krylov solver implementations in Petsc; are well-optimized parallel Krylov solvers/preconditioners really completely identical to a serial solver except in the array / matrix type, or do they require further algorithmic changes to work optimal…

I see this comment there (also they’re really thinking ahead with “Fortran 20008” :slight_smile: Fortran 2018 already added some TS 29113: Further Interoperability of Fortran with C ): Julia has a lot going for it, but I’m not aware of successful packages written in Julia to be called from C/C++/Fo…

[image] mohamed82008: I would love to see a benchmark of all the parallel linear algebra options available in Julia including PETSc. Then a list of which packages are incompatible with which types of parallelism can be made which the community can then address one by one. A benchmark is a good p…

I’d focus on the following comment by Jed Brown’s comment : Julia has a lot going for it, but I’m not aware of successful packages written in Julia to be called from C/C++/Fortran. Does anyone has a non trivial example how to integrate Julia into C / C++ which works across Windows, Linux and macO…

[image] RoyiAvital: Julia has a lot going for it, but I’m not aware of successful packages written in Julia to be called from C/C++/Fortran. Does anyone has a non trivial example how to integrate Julia into C / C++ which works across Windows, Linux and macOS? I guess something such does ex…

I think people miss how crucial this is. In many places there is an existing code in C / C++. No one will write in from scratch in Julia. The question is what will be done with new modules? If there is an easy way to embed Julia in new modules it means it is a candidate. Since there is none, it …

[image] RoyiAvital: I think people miss how crucial this is. In many places there is an existing code in C / C++. No one will write in from scratch in Julia. The question is what will be done with new modules? If there is an easy way to embed Julia in new modules it means it is a candidate. …

Mention of Julia for the re-implementation of PETSC

Offtopic

Eben60 September 9, 2020, 1:02pm 22

Or see Data-exchange Julia-LabView for another use case: Experiment control by LabVIEW, and e.g. signal analysis with Julia.

Topic		Replies	Views
PETSC.jl MPI wrapper General Usage hpc , parallel , mpi	8	1315	June 7, 2025
Losing Science Marketshare to Non-Python Languages Offtopic	30	3250	September 21, 2022
Finite Element Computations On A Cluster Using Petsc.jl Numerics question , package	2	2360	April 21, 2017
Julia motivation: why weren't Numpy, Scipy, Numba, good enough? Community history	123	84216	September 21, 2018
What package[s] are state-of-the art OR attract you to Julia, and make you stay there (not easily replicateable in e.g. Python, R, MATLAB)? General Usage	61	15957	April 10, 2020

Mention of Julia for the re-implementation of PETSC

Related topics