DO NOT SQUASH Absorb TSFC #3904
Merged
Conversation
This file contains bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Add flop counting of scheduled impero.
Need to be able to create separate kernel arguments for cell sizes and generic coefficients.
This doesn't actually work with Firedrake anyway.
Since we pass in the impero_c to construct the kernel, we can just count flops there.
Push responsibility for construction of the return argument inside the kernel onto the builder.
The Firedrake code requires a loopy kernel anyway, so cull dead code.
…ates Wence/feature/swarm updates
A new "UFLtoGEMCallback" representing the function to dual evaluate is added. This takes the processed UFL expression and, when called, performs the necessary translation to gem. Also add some helpful documentation to partial_indexed.
This rename/refactor makes the purpose of DualEvaluationCallable much clearer. Also removes the untested symbolic derivative stuff since Kirby posits that you can do this with a quadrature rule (like for moments). This is not supported by FInAT's dual_evaluation anyway.
This fiddling for expression shape not matching element value shape is already taken care of in the FInAT dual_evaluation method of a tensor finite element and would now be incorrect. We also no longer give c_e_d_e a base element in the tensor case.
FInAT element.dual_evaluation now returns (gem_expression, basis_indices)
gem.optimise.contraction now takes care of sum factorisation inside FInAT so there is no need to tell FInAT about factors. Also not yet implemented
FInAT has also been updated
On tensor product cells, interpolation of Q/DQ elements should sum
factorise to give an operation count that is O(p^{d+1}). Check this
works. Additionally check that interpolation of Vector/TensorElement
costs only the product of the value_shape more flops than the
equivalent scalar element.
FInAT Dual Evaluation
Update UFL element interface
Refactor variants
Clean up variants
The previous solution to avoiding loop interchanges did not account for having multiple loop nests.
3 tasks
|
|
connorjward
commented
Dec 5, 2024
connorjward
commented
Dec 5, 2024
connorjward
force-pushed
the
connorjward/add-tsfc
branch
from
cacb13f to
5dad275
Compare
dham
reviewed
Dec 5, 2024
|
Looks good to me. Is it better if we move |
Done. |
ksagiyam
previously approved these changes
Dec 6, 2024
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.
This suggestion is invalid because no changes were made to the code.
Suggestions cannot be applied while the pull request is closed.
Suggestions cannot be applied while viewing a subset of changes.
Only one suggestion per line can be applied in a batch.
Add this suggestion to a batch that can be applied as a single commit.
Applying suggestions on deleted lines is not supported.
You must change the existing code in this line in order to create a valid suggestion.
Outdated suggestions cannot be applied.
This suggestion has been applied or marked resolved.
Suggestions cannot be applied from pending reviews.
Suggestions cannot be applied on multi-line comments.
Suggestions cannot be applied while the pull request is queued to merge.
Suggestion cannot be applied right now. Please check back later.
Happily the changes here are extremely minor.
Goes with firedrakeproject/fiat#104