[WIP][DISCUSS] Single source QFunction proposal #288

YohannDudouit · 2019-07-01T23:57:02Z

This PR proposes a qFunction format that would allow the same code to be used on host and on device.
The qFunction is in the .qf files, but could be moved to .h files. That would require cpu backends to define the macro CEED_QFUNCTION.

Rescaled NS example

…/libCEED into rescaled-ns-example Conflicts: examples/navier-stokes/navierstokes.c

… into rescaled-ns-example

Restriction Block

* Add all backends to Doxygen * alphabetize * alphabetize

Brief description of the cuda-shared backend in README.md

)

Offsetof fix

Also use CPPFLAGS in case -I/path/to/valgrind is needed.

Update CUDA/OCCA CEED_COPY_VALUES logic

Add ElemRestrGetMultiplicity and Tests

BugFix - Resource strcmp in LIBXSMM backends

[Jed] Commit revised to leave derivatives intact.

* valeria/fix-ns-derivatives: examples/navier-stokes: formatting cleanup navier-stokes: use du[3][3] to clean up indexing Fix derivatives order in du

examples/petsc: fix max reduction type and address

jeremylt

I think this approach is good for both CPU and GPU. I'd like to use this to build the QF gallery.

We do need to update to the new PETSc BPs though.

jeremylt

We need to update to the new PETSc BPs.

Fix typo in t209-f

Fix ref backend priorities

* Update nek5000 examples to use NEKTOOLS_DIR * Remove +x * Nek: Fix a bug in generate-boxes.sh script * Use one script for nek exaamples * Update make-nek-tests.sh * Update nek-examples.sh * Change permissions of make-nek-tests.sh * Update examples Makefile to use new nek script * Update nek5000 examples build process * Trying to make travis happy * Polishing the script * This should make Travis happy * Polish nek-examples.sh * Update examples/Makefile * Update top-level makefile * Delete make-nek-tests.sh & update nek-examples.sh * Update Makefile and examples/Makefile * Update nek-examples.sh * Update TESTARGS in bp .usr files * Update Makefile * Update .travis.yml * Update nek-examples.sh * Add CEED_DIR to tap.sh * Uncomment OCCA_DIR in Makefile * Fix make prove and update tap.sh * Pass NEK5K_DIR and CEED_DIR in prove-all * Set MPI=0 default for Nek examples/tests * Nek5K make updates * Make prove update * Update make-prove to make all ex except Nek5K in parallel * Set Nek5K examples to use MPI by default * Makefile: Improved Nek5k handling * Makefile: build Nek bps in separate directories * Nek: rename dir for f77 file length * debugging * Nek: use single source * Makefile: fix Nek mpif.h issue * Nek: Clean up build proces, improve documentation * Makefile: enable building single Nek example in parallel with other examples * Nek: Fix CC/FC and TESTARGS

README: Update corectness for examples

* First steps toward cuda-gen backend! * Closer to real code generation. * Generated code should be ready for nvrtc. * The code generation skeleton is ready. * Hack with the qfunction to make the operator kernel compile. * Some tweaks in the makefile + Input fields structure change. * Remove using cout. * 1d interp and grad device functions. * 1d readDofs, readQuads, writeDofs, writeQuads. * Remove dead code. * readDofs, readQuads, writeDofs, writeQuads for 2d and 3d * 2d interp and grad * 3d interp and grad * - weight functions for 1d,2d,3d - link the indices to the kernel - link the fields to the kernel - link the basis to the kernel * Add the qFunction reader + inlining * Add qf files for the tests. * Add qf file for ceed/ex1 * Add qf file for mfem/bp1 * All tests pass. * Add qFunction for mfem/bp3, petsc/bp1, and petsc/bp3. * mfem/bp1 passes + remove dead code * Fix a bug in n_quads_out for writeQuads * mfem/bp3 passes. * All tests all examples pass. * Temporary tweaks for mfem benchmarking * Add Context management. * Modify .qf files to take into account the context. * Enable optimizations. * First set of optimization for 2D and 3D. * Makefile tweaks and destructor code. * make style. * Add -MP flag. * Fix linking issues with the tests. * Update .qf files for the tests. * Add .qf files for nek5000 examples. * Use shared memory for B and G matrices. * Fix bug introduced in previous commit.

jeremylt · 2019-07-26T14:29:47Z

Now that CUDA Gen has been merged, can we make this into a PR to master? I believe that we all generally agree on this approach and it's just a matter of ironing out details.

Conflicts: Makefile backends/cuda-gen/ceed-cuda-gen-operator-build.cpp backends/cuda-gen/ceed-cuda-gen-operator.c backends/cuda-gen/ceed-cuda-gen-qfunction.c backends/cuda-gen/ceed-cuda-gen.c backends/cuda-gen/ceed-cuda-gen.h backends/cuda/ceed-cuda-qfunction.c backends/cuda/ceed-cuda.c examples/ceed/ex1.qf examples/mfem/bp1.qf examples/mfem/bp3.qf examples/nek5000/bp1.qf examples/petsc/bp1.c examples/petsc/bp1.qf examples/petsc/bp3.qf examples/petsc/bps.c tests/t400-qfunction-f.qf tests/t400-qfunction.qf tests/t401-qfunction-f.qf tests/t401-qfunction.qf tests/t500-operator-f.qf tests/t500-operator.qf tests/t501-operator-f.qf tests/t501-operator.qf tests/t502-operator-f.qf tests/t502-operator.qf

jeremylt · 2019-08-02T22:56:46Z

include/ceed.h

 #endif
+#ifndef CEED_QFUNCTION
+#  define CEED_QFUNCTION static
+#endif


I think I'd change this to @jedbrown's suggestion of

#define CEED_QFUNCTION(name) \ static const char name ## _loc[] = __FILE__ ":" #name; \ static int name

so that we can use

CeedQFunctionCreateInterior(ceed, 1, mass, mass_loc, &qf_mass);

This lets us have qfunction files name differently than the base file without my crazy machinations it the PETSc unified BP code.

YohannDudouit · 2019-08-02T23:01:38Z

Closed in favour of #304

Valeria Barra and others added 30 commits May 1, 2019 12:32

Rescaled NS example

13681b3

Rescaled NS example

f7c73ca

Rescaled NS example

Merge branch 'rescaled-ns-example' of https://github.com/valeriabarra…

dca4711

…/libCEED into rescaled-ns-example Conflicts: examples/navier-stokes/navierstokes.c

Add PetscObjectSetName to set solution's vec name

9ccb6b9

Merge branch 'rescaled-ns-example' of github.com:valeriabarra/libCEED…

70cdef1

… into rescaled-ns-example

Rename field name Total Energy -> EnergyDensity

5cf694b

Rescale time-stepper options with units of time

47868d5

Some tidying

0cdf8f0

Add a reference non-tensor BasisApply for cuda backends.

9ad4535

Remove useless function declaration.

469f022

make style

abe33e5

Add ElemRestrictionApplyBlock

be9261b

Add restriction by block to /cpu/self/*/blocked

a765294

Include full evec blocked backend

045b9c4

Use blocking in optimized serial backends

89c6efa

Update t208-f to use offset

9fbf56a

Add block paramenter example in doc

1f37b40

Merge pull request #228 from CEED/rstr-block

d4fd279

Restriction Block

Use function pointers in CeedSetBackendFunction

a7a2e80

Dynamically allocate Ceed function table

6e79d47

Add all backends to Doxygen (#251)

7e95bb9

* Add all backends to Doxygen * alphabetize * alphabetize

Brief description of the cuda-shared backend in README.md

a17ac2e

Merge pull request #255 from CEED/yohann/cuda-shared-readme

db777ff

Brief description of the cuda-shared backend in README.md

Revert "Brief description of the cuda-shared backend in README.md" (#256

8ffbcc6

)

Add description of cuda-shared backend in README.md

854b039

Check for devel Valgrind

bdc3149

Improved documentation for SetBackendFunction

cb37edd

VecCreate -> VectorCreate

f8902d9

Merge pull request #232 from CEED/offsetof-fix

a7724da

Offsetof fix

Makefile: avoid leaving -.o file in valgrind check

b8fb44e

Also use CPPFLAGS in case -I/path/to/valgrind is needed.

jeremylt and others added 3 commits July 3, 2019 23:40

Merge pull request #289 from CEED/cuda-occa-copy-vals

f405f80

Update CUDA/OCCA CEED_COPY_VALUES logic

Add ElemRestrGetMult and tests

1469ee4

Merge pull request #291 from CEED/rstr-mult

56cf2fb

Add ElemRestrGetMultiplicity and Tests

YohannDudouit mentioned this pull request Jul 10, 2019

Full jit compiled operator: cuda-gen backend #275

Merged

jeremylt and others added 7 commits July 12, 2019 15:13

Fix resource strcmp in xsmm backends

e0fc044

Update CPU backends to give default for /cpu/self/***

6f7d248

Merge pull request #295 from CEED/cpu_priority_fix

04ad072

BugFix - Resource strcmp in LIBXSMM backends

examples/navier-stokes: formatting cleanup

6278642

[Jed] Commit revised to leave derivatives intact.

Merge branch 'valeria/fix-ns-derivatives' [PR #292]

e24ee97

* valeria/fix-ns-derivatives: examples/navier-stokes: formatting cleanup navier-stokes: use du[3][3] to clean up indexing Fix derivatives order in du

examples/petsc: fix max reduction type and address

00742fa

Merge pull request #296 from CEED/jed/fix-error-reduction

01b3627

examples/petsc: fix max reduction type and address

jeremylt approved these changes Jul 17, 2019

View reviewed changes

jeremylt requested changes Jul 17, 2019

View reviewed changes

jeremylt and others added 8 commits July 18, 2019 11:23

tests: fix typo in t209-f

deafd6b

backends: fix ref backend priorities

706bc5e

Merge pull request #297 from CEED/jeremy/209-f-fix

3a6bb9f

Fix typo in t209-f

Merge pull request #298 from CEED/jeremy/ref-priority-fix

bd2096d

Fix ref backend priorities

README: Update corectness for examples

01a7252

Merge pull request #300 from CEED/jeremy/doc-fix

6f67eec

README: Update corectness for examples

YohannDudouit added 5 commits July 26, 2019 17:48

Remove previous nek qf files.

0b902cb

Remove .cu files.

f442366

Remove .qf files.

40d6fce

Remove dead code in the tests.

1175c3d

jeremylt reviewed Aug 2, 2019

View reviewed changes

jeremylt closed this Aug 2, 2019

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[WIP][DISCUSS] Single source QFunction proposal #288

[WIP][DISCUSS] Single source QFunction proposal #288

Uh oh!

YohannDudouit commented Jul 1, 2019

Uh oh!

jeremylt left a comment •

edited

Loading

Uh oh!

jeremylt left a comment

Uh oh!

jeremylt commented Jul 26, 2019 •

edited

Loading

Uh oh!

jeremylt Aug 2, 2019

Uh oh!

jeremylt Aug 2, 2019

Uh oh!

YohannDudouit commented Aug 2, 2019

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants

[WIP][DISCUSS] Single source QFunction proposal #288

[WIP][DISCUSS] Single source QFunction proposal #288

Uh oh!

Conversation

YohannDudouit commented Jul 1, 2019

Uh oh!

jeremylt left a comment • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Choose a reason for hiding this comment

Uh oh!

jeremylt left a comment

Choose a reason for hiding this comment

Uh oh!

jeremylt commented Jul 26, 2019 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

jeremylt Aug 2, 2019

Choose a reason for hiding this comment

Uh oh!

jeremylt Aug 2, 2019

Choose a reason for hiding this comment

Uh oh!

YohannDudouit commented Aug 2, 2019

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

7 participants

jeremylt left a comment •

edited

Loading

jeremylt commented Jul 26, 2019 •

edited

Loading