Skip to content

Parallelization of compute coefficients functions on GPU#63

Merged
huiyuxie merged 11 commits intomainfrom
fix
Dec 6, 2024
Merged

Parallelization of compute coefficients functions on GPU#63
huiyuxie merged 11 commits intomainfrom
fix

Conversation

@huiyuxie
Copy link
Copy Markdown
Member

@huiyuxie huiyuxie commented Oct 4, 2024

This PR moves du and u to the GPU and keep them on GPU during the process of semidiscretization (also fixes #13, i.e., makes the copy process of du and u out of the rhs! function in loops).

@huiyuxie
Copy link
Copy Markdown
Member Author

huiyuxie commented Oct 7, 2024

Remove AMR for now - please add the implementation (especially the resize! function) and tests later.

@huiyuxie huiyuxie changed the title Parallelization for cache components initialization Parallelization for computing coefficients Oct 8, 2024
@huiyuxie huiyuxie added the refactoring Refactoring label Dec 6, 2024
@huiyuxie huiyuxie changed the title Parallelization for computing coefficients Parallelization for compute coefficients functions Dec 6, 2024
@huiyuxie
Copy link
Copy Markdown
Member Author

huiyuxie commented Dec 6, 2024

This also fixes the accuracy tests failure due to the change of code in upstream see trixi-framework/Trixi.jl#2134 (targeting noncons mortars for 2D and 3D).

@huiyuxie huiyuxie changed the title Parallelization for compute coefficients functions Parallelization of compute coefficients functions on GPU Dec 6, 2024
@huiyuxie huiyuxie merged commit ded2864 into main Dec 6, 2024
@huiyuxie huiyuxie deleted the fix branch December 6, 2024 10:46
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

refactoring Refactoring

Projects

None yet

Development

Successfully merging this pull request may close these issues.

Refactor rhs! to move data copying outside of iteration loop

1 participant