Use shared-memory for transpose non-tensor basis kernel to accelerate load of A matrix

H(div) and H(curl) basis support for magma backend

Remove accidental WIP tuning files from #1382

Adjust include groupings

Fix mismatching declaration

Formatting consistency for magma backend with cuda-ref and hip-ref

Includes JiT upgrades for Magma non-tensor basis to only compile for N values which are used at runtime. Adds JiT for Magma non-ten

Formatting consistency for magma backend with cuda-ref and hip-ref

Includes JiT upgrades for Magma non-tensor basis to only compile for N values which are used at runtime. Adds JiT for Magma non-tensor basis CEED_EVAL_WEIGHT mode.

show more ...

Rename files in magma backend for consistency with other libCEED backends

Revert opt/blocked backend delegate, this is no longer needed after #1362

Correct opt/blocked fallback to opt/serial for things like QFunction assembly

Minor improvement to consistency between ref/blocked and opt/blocked backends for readability

cpu - clean up delegation/fallback between CPU backends

Remove unused parameter from CeedTensorContractCreate

magma - fix make tidy issues

memcheck - move codecov markers

ref - fix rstr parent resource name check

ref - move codecov markers

Merge pull request #1340 from CEED/jeremy/fix-ptsc-orients-copy

Fix CEED_COPY_VALUES for ref rstr at points

Non-square operator full assembly (#1316)

* Consistency formatting for operator tests

* Add (failing) test for non-square operator full assembly

* Add support for CPU-based full assembly of no

Non-square operator full assembly (#1316)

* Consistency formatting for operator tests

* Add (failing) test for non-square operator full assembly

* Add support for CPU-based full assembly of non-square CeedOperators

* Fix full assembly of identity quadrature functions and operators with CEED_BASIS_NONE

Also adds a unit test for full assembly which fails prior to the changes in this commit.

* Minor refactor to improve coverage

show more ...

Support CPU shared-memory parallelism with OpenMP (#1279)

* Updates for OpenMP thread safety (one Ceed per thread, potentially with shared output vector)

* Makefile tabs vs. spaces consistency

Support CPU shared-memory parallelism with OpenMP (#1279)

* Updates for OpenMP thread safety (one Ceed per thread, potentially with shared output vector)

* Makefile tabs vs. spaces consistency

* Fix unrelated pragma bug for Intel compilers in `backend.h`

* Address PR feedback: Use _OPENMP macro, simplify OpenMP pragma wrappers

* Address PR feedback: Add new statement macro CeedPragmaThreadPrivate to PREDEFINED entries in Doxyfile

* Add OpenMP support to Intel CI workflow for testing

* Add documentation for OPENMP option and Update releasenotes.md

* Revise OpenMP implementation: Rather than enforcing global variables to be threadprivate, just wrap potential race conditions in a critical block (during registration)

* Avoid returning from OpenMP blocks

* Early break on error when registering backends or QFunctions

* Formatting fixes after rebase, newline after variable declarations

* Address PR feedback: Revert some unintentional changes to debug output

* Update codecov exclusions

show more ...

rstr - fix CEED_COPY_VALUES for ref rstr at points

rstr - renaming for clarity

rstr - add tests for AtPoints

rstr - allow restriction to ordered points evec

rstr - intial CPU implimentation

Small update for cuda-shared/hip-shared consistency