xref: /libCEED/doc/sphinx/source/releasenotes.md (revision cdf32b9369b06315737c6f9df3317771bd8e77c7)
1bcb2dfaeSJed Brown# Changes/Release Notes
2bcb2dfaeSJed Brown
3bcb2dfaeSJed BrownOn this page we provide a summary of the main API changes, new features and examples
4bcb2dfaeSJed Brownfor each release of libCEED.
5bcb2dfaeSJed Brown
6bcb2dfaeSJed Brown(main)=
7bcb2dfaeSJed Brown
8bcb2dfaeSJed Brown## Current `main` branch
9bcb2dfaeSJed Brown
107e7773b5SJeremy L Thompson### Interface changes
117e7773b5SJeremy L Thompson
127e7773b5SJeremy L Thompson- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields.
13ce4822f6SJeremy L Thompson- Promote to the public API: QFunction and Operator field objects, `CeedQFunctionField` and `CeedOperatorField`, and associated getters, {c:func}`CeedQFunctionGetFields`; {c:func}`CeedQFunctionFieldGetName`; {c:func}`CeedQFunctionFieldGetSize`; {c:func}`CeedQFunctionFieldGetEvalMode`; {c:func}`CeedOperatorGetFields`; {c:func}`CeedOperatorFieldGetElemRestriction`; {c:func}`CeedOperatorFieldGetBasis`; and {c:func}`CeedOperatorFieldGetVector`.
14f04ea552SJeremy L Thompson- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added.
1570a7ffb3SJeremy L Thompson- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients.
164db537f9SJeremy L Thompson- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces.
17dcc1e3ecSJeremy L Thompson- Warning added when compiling OCCA backend to alert users that this backend is experimental.
189a1d3511SJeremy L Thompson- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`.
1943e1b16fSJeremy L Thompson- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name.
209c774eddSJeremy L Thompson- Clarify documentation for {c:func}`CeedVectorTakeArray`; this function will error if {c:func}`CeedVectorSetArray` with `copy_mode == CEED_USE_POINTER` was not previously called for the corresponding `CeedMemType`.
219c774eddSJeremy L Thompson- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitalized arrays; require initalized data for {c:func}`CeedVectorGetArray`.
22*cdf32b93SJeremy L Thompson- Added {c:func}`CeedQFunctionContextRegisterDouble` and {c:func}`CeedQFunctionContextRegisterInt32` with {c:func}`CeedQFunctionContextSetDouble` and {c:func}`CeedQFunctionContextSetInt32` to facilitate easy updating of {ref}`CeedQFunctionContext` data by user defined field names.
23*cdf32b93SJeremy L Thompson- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retreive user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`.
247e7773b5SJeremy L Thompson
25f479eb23SJeremy L Thompson### New features
26f479eb23SJeremy L Thompson
27f479eb23SJeremy L Thompson- `CeedScalar` can now be set as `float` or `double` at compile time.
2830601ac0SJeremy L Thompson- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends.
29fb3c7d02SJeremy L Thompson- Added support for JiT of QFunctions with `#include "relative/path/local-file.h"` statements for additional local files. Note that files included with `""` are searched relative to the current file first, then by compiler paths (as with `<>` includes). To use this feature, one should adhere to relative paths only, not compiler flags like `-I`, which the JiT will not be aware of.
3023dfbf5bSJeremy L Thompson- Remove need to guard library headers in QFunction source for code generation backends.
313f21f6b1SJeremy L Thompson- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present.
32f479eb23SJeremy L Thompson
33bcb2dfaeSJed Brown### Maintainability
34bcb2dfaeSJed Brown
35bcb2dfaeSJed Brown- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage.
36db52d626SJeremy L Thompson- `Include-what-you-use` makefile target added as `make iwyu`.
37bf4cb664SJeremy L Thompson- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase.
38bcb2dfaeSJed Brown
39bcb2dfaeSJed Brown(v0-9)=
40bcb2dfaeSJed Brown
41bcb2dfaeSJed Brown## v0.9 (Jul 6, 2021)
42bcb2dfaeSJed Brown
43bcb2dfaeSJed Brown### Interface changes
44bcb2dfaeSJed Brown
45bcb2dfaeSJed Brown- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact.
46bcb2dfaeSJed Brown
47bcb2dfaeSJed Brown### New features
48bcb2dfaeSJed Brown
49bcb2dfaeSJed Brown- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use.
50bcb2dfaeSJed Brown- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends.
51bcb2dfaeSJed Brown- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations.
52bcb2dfaeSJed Brown- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed).
53bcb2dfaeSJed Brown
54bcb2dfaeSJed Brown### Performance improvements
55bcb2dfaeSJed Brown
56bcb2dfaeSJed Brown### Examples
57bcb2dfaeSJed Brown
58bcb2dfaeSJed Brown- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations.
59bcb2dfaeSJed Brown- Fluid mechanics example adds GPU support and improves modularity.
60bcb2dfaeSJed Brown
61bcb2dfaeSJed Brown### Deprecated backends
62bcb2dfaeSJed Brown
63bcb2dfaeSJed Brown- The `/cpu/self/tmpl` and `/cpu/self/tmpl/sub` backends have been removed. These backends were intially added to test the backend inheritance mechanism, but this mechanism is now widely used and tested in multiple backends.
64bcb2dfaeSJed Brown
65bcb2dfaeSJed Brown(v0-8)=
66bcb2dfaeSJed Brown
67bcb2dfaeSJed Brown## v0.8 (Mar 31, 2021)
68bcb2dfaeSJed Brown
69bcb2dfaeSJed Brown### Interface changes
70bcb2dfaeSJed Brown
71bcb2dfaeSJed Brown- Error handling improved to include enumerated error codes for C interface return values.
72bcb2dfaeSJed Brown- Installed headers that will follow semantic versioning were moved to {code}`include/ceed` directory. These headers have been renamed from {code}`ceed-*.h` to {code}`ceed/*.h`. Placeholder headers with the old naming schema are currently provided, but these headers will be removed in the libCEED v0.9 release.
73bcb2dfaeSJed Brown
74bcb2dfaeSJed Brown### New features
75bcb2dfaeSJed Brown
76bcb2dfaeSJed Brown- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features.
77bcb2dfaeSJed Brown- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly.
78bcb2dfaeSJed Brown- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators.
79bcb2dfaeSJed Brown
80bcb2dfaeSJed Brown### Performance improvements
81bcb2dfaeSJed Brown
82bcb2dfaeSJed Brown- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`.
83bcb2dfaeSJed Brown- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`.
84bcb2dfaeSJed Brown
85bcb2dfaeSJed Brown### Examples
86bcb2dfaeSJed Brown
87bcb2dfaeSJed Brown- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions.
88bcb2dfaeSJed Brown- {ref}`example-petsc-elasticity` example updated with Neo-Hookean hyperelasticity in current configuration as well as improved Neo-Hookean hyperelasticity exploring storage vs computation tradeoffs.
89bcb2dfaeSJed Brown- {ref}`example-petsc-navier-stokes` example updated with isentropic traveling vortex test case, an analytical solution to the Euler equations that is useful for testing boundary conditions, discretization stability, and order of accuracy.
90bcb2dfaeSJed Brown- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree.
91bcb2dfaeSJed Brown
92bcb2dfaeSJed Brown(v0-7)=
93bcb2dfaeSJed Brown
94bcb2dfaeSJed Brown## v0.7 (Sep 29, 2020)
95bcb2dfaeSJed Brown
96bcb2dfaeSJed Brown### Interface changes
97bcb2dfaeSJed Brown
98bcb2dfaeSJed Brown- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors.
99bcb2dfaeSJed Brown  As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`.
100bcb2dfaeSJed Brown  These changes improve support for mixed finite element methods.
101bcb2dfaeSJed Brown- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature.
102bcb2dfaeSJed Brown- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity.
103bcb2dfaeSJed Brown- Linear Operators can be assembled as point-block diagonal matrices with {c:func}`CeedOperatorLinearAssemblePointBlockDiagonal`, provided in row-major form in a {code}`ncomp` by {code}`ncomp` block per node.
104bcb2dfaeSJed Brown- Diagonal assemble interface changed to accept a {ref}`CeedVector` instead of a pointer to a {ref}`CeedVector` to reduce memory movement when interfacing with calling code.
105bcb2dfaeSJed Brown- Added {c:func}`CeedOperatorLinearAssembleAddDiagonal` and {c:func}`CeedOperatorLinearAssembleAddPointBlockDiagonal` for improved future integration with codes such as MFEM that compose the action of {ref}`CeedOperator`s external to libCEED.
106bcb2dfaeSJed Brown- Added {c:func}`CeedVectorTakeAray` to sync and remove libCEED read/write access to an allocated array and pass ownership of the array to the caller.
107bcb2dfaeSJed Brown  This function is recommended over {c:func}`CeedVectorSyncArray` when the {code}`CeedVector` has an array owned by the caller that was set by {c:func}`CeedVectorSetArray`.
108bcb2dfaeSJed Brown- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory.
109bcb2dfaeSJed Brown- Added {c:func}`CeedOperatorMultigridLevelCreate`, {c:func}`CeedOperatorMultigridLevelCreateTensorH1`, and {c:func}`CeedOperatorMultigridLevelCreateH1` to facilitate creation of multigrid prolongation, restriction, and coarse grid operators using a common quadrature space.
110bcb2dfaeSJed Brown
111bcb2dfaeSJed Brown### New features
112bcb2dfaeSJed Brown
113bcb2dfaeSJed Brown- New HIP backend: `/gpu/hip/ref`.
114bcb2dfaeSJed Brown- CeedQFunction support for user `CUfunction`s in some backends
115bcb2dfaeSJed Brown
116bcb2dfaeSJed Brown### Performance improvements
117bcb2dfaeSJed Brown
118bcb2dfaeSJed Brown- OCCA backend rebuilt to facilitate future performance enhancements.
119bcb2dfaeSJed Brown- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`.
120bcb2dfaeSJed Brown
121bcb2dfaeSJed Brown### Examples
122bcb2dfaeSJed Brown
123bcb2dfaeSJed Brown- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions.
124bcb2dfaeSJed Brown
125bcb2dfaeSJed Brown### Deprecated backends
126bcb2dfaeSJed Brown
127bcb2dfaeSJed Brown- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`.
128bcb2dfaeSJed Brown
129bcb2dfaeSJed Brown(v0-6)=
130bcb2dfaeSJed Brown
131bcb2dfaeSJed Brown## v0.6 (Mar 29, 2020)
132bcb2dfaeSJed Brown
133bcb2dfaeSJed BrownlibCEED v0.6 contains numerous new features and examples, as well as expanded
134bcb2dfaeSJed Browndocumentation in [this new website](https://libceed.readthedocs.io).
135bcb2dfaeSJed Brown
136bcb2dfaeSJed Brown### New features
137bcb2dfaeSJed Brown
138bcb2dfaeSJed Brown- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly
139bcb2dfaeSJed Brown  1-1 correspondence with the C interface, plus some convenience features.  For instance,
140bcb2dfaeSJed Brown  data stored in the {cpp:type}`CeedVector` structure are available without copy as
141bcb2dfaeSJed Brown  {py:class}`numpy.ndarray`.  Short tutorials are provided in
142bcb2dfaeSJed Brown  [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/).
143bcb2dfaeSJed Brown- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point,
144bcb2dfaeSJed Brown  {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal
145bcb2dfaeSJed Brown  ({c:func}`CeedOperatorAssembleLinearDiagonal`).  These operations are useful for
146bcb2dfaeSJed Brown  preconditioning ingredients and are used in the libCEED's multigrid examples.
147bcb2dfaeSJed Brown- The inverse of separable operators can be obtained using
148bcb2dfaeSJed Brown  {c:func}`CeedOperatorCreateFDMElementInverse` and applied with
149bcb2dfaeSJed Brown  {c:func}`CeedOperatorApply`.  This is a useful preconditioning ingredient,
150bcb2dfaeSJed Brown  especially for Laplacians and related operators.
151bcb2dfaeSJed Brown- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`,
152bcb2dfaeSJed Brown  {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`.
153bcb2dfaeSJed Brown- Make public accessors for various attributes to facilitate writing composable code.
154bcb2dfaeSJed Brown- New backend: `/cpu/self/memcheck/serial`.
155bcb2dfaeSJed Brown- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA
156bcb2dfaeSJed Brown  backends.  (Single source is coming soon for OCCA backends.)
157bcb2dfaeSJed Brown- Fix some missing edge cases in CUDA backend.
158bcb2dfaeSJed Brown
159bcb2dfaeSJed Brown### Performance Improvements
160bcb2dfaeSJed Brown
161bcb2dfaeSJed Brown- MAGMA backend performance optimization and non-tensor bases.
162bcb2dfaeSJed Brown- No-copy optimization in {c:func}`CeedOperatorApply`.
163bcb2dfaeSJed Brown
164bcb2dfaeSJed Brown### Interface changes
165bcb2dfaeSJed Brown
166bcb2dfaeSJed Brown- Replace {code}`CeedElemRestrictionCreateIdentity` and
167bcb2dfaeSJed Brown  {code}`CeedElemRestrictionCreateBlocked` with more flexible
168bcb2dfaeSJed Brown  {c:func}`CeedElemRestrictionCreateStrided` and
169bcb2dfaeSJed Brown  {c:func}`CeedElemRestrictionCreateBlockedStrided`.
170bcb2dfaeSJed Brown- Add arguments to {c:func}`CeedQFunctionCreateIdentity`.
171bcb2dfaeSJed Brown- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification
172bcb2dfaeSJed Brown  with {cpp:enum}`CeedInterlaceMode`.  This is now an attribute of the
173bcb2dfaeSJed Brown  {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no
174bcb2dfaeSJed Brown  longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and
175bcb2dfaeSJed Brown  {c:func}`CeedElemRestrictionApply`.
176bcb2dfaeSJed Brown
177bcb2dfaeSJed Brown### Examples
178bcb2dfaeSJed Brown
179bcb2dfaeSJed BrownlibCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`.
180bcb2dfaeSJed BrownNotable additions include:
181bcb2dfaeSJed Brown
182bcb2dfaeSJed Brown- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of
183bcb2dfaeSJed Brown  a domain in 1, 2, and 3 dimensions by applying a Laplacian.
184bcb2dfaeSJed Brown
185bcb2dfaeSJed Brown- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area
186bcb2dfaeSJed Brown  of domains (like the cube and sphere) by direct integration on a surface mesh;
187bcb2dfaeSJed Brown  demonstrates geometric dimension different from topological dimension.
188bcb2dfaeSJed Brown
189bcb2dfaeSJed Brown- PETSc {ref}`example-petsc-bps`:
190bcb2dfaeSJed Brown
191bcb2dfaeSJed Brown  - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support.
192bcb2dfaeSJed Brown  - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements
193bcb2dfaeSJed Brown    and transparent CUDA support.
194bcb2dfaeSJed Brown  - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`):
195bcb2dfaeSJed Brown    generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric
196bcb2dfaeSJed Brown    dimension different from topological dimension.
197bcb2dfaeSJed Brown
198bcb2dfaeSJed Brown- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid
199bcb2dfaeSJed Brown  solver with algebraic multigrid coarse solve.
200bcb2dfaeSJed Brown
201bcb2dfaeSJed Brown- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly
202bcb2dfaeSJed Brown  `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`),
203bcb2dfaeSJed Brown  implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and
204bcb2dfaeSJed Brown  quasi-2D computational domain support.
205bcb2dfaeSJed Brown
206bcb2dfaeSJed Brown- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for
207bcb2dfaeSJed Brown  linear elasticity, small-strain hyperelasticity, and globalized finite-strain
208bcb2dfaeSJed Brown  hyperelasticity using p-multigrid with algebraic multigrid coarse solve.
209bcb2dfaeSJed Brown
210bcb2dfaeSJed Brown(v0-5)=
211bcb2dfaeSJed Brown
212bcb2dfaeSJed Brown## v0.5 (Sep 18, 2019)
213bcb2dfaeSJed Brown
214bcb2dfaeSJed BrownFor this release, several improvements were made. Two new CUDA backends were added to
215bcb2dfaeSJed Brownthe family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art
216bcb2dfaeSJed Brownperformance using single-source {ref}`CeedQFunction`. From this release, users
217bcb2dfaeSJed Browncan define Q-Functions in a single source code independently of the targeted backend
218bcb2dfaeSJed Brownwith the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU
219bcb2dfaeSJed Browncompilation of the user provided {ref}`CeedQFunction` code. To allow a unified
220bcb2dfaeSJed Browndeclaration, the {ref}`CeedQFunction` API has undergone a slight change:
221bcb2dfaeSJed Brownthe `QFunctionField` parameter `ncomp` has been changed to `size`. This change
222bcb2dfaeSJed Brownrequires setting the previous value of `ncomp` to `ncomp*dim` when adding a
223bcb2dfaeSJed Brown`QFunctionField` with eval mode `CEED EVAL GRAD`.
224bcb2dfaeSJed Brown
225bcb2dfaeSJed BrownAdditionally, new CPU backends
226bcb2dfaeSJed Brownwere included in this release, such as the `/cpu/self/opt/*` backends (which are
227bcb2dfaeSJed Brownwritten in pure C and use partial **E-vectors** to improve performance) and the
228bcb2dfaeSJed Brown`/cpu/self/ref/memcheck` backend (which relies upon the
229bcb2dfaeSJed Brown[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user
230bcb2dfaeSJed Brown{ref}`CeedQFunction` have no undefined values).
231bcb2dfaeSJed BrownThis release also included various performance improvements, bug fixes, new examples,
232bcb2dfaeSJed Brownand improved tests. Among these improvements, vectorized instructions for
233bcb2dfaeSJed Brown{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD`
234bcb2dfaeSJed Browninstead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and
235bcb2dfaeSJed Brownidentity Q-Functions were introduced, and the PETSc benchmark problems were expanded
236bcb2dfaeSJed Brownto include unstructured meshes handling were. For this expansion, the prior version of
237bcb2dfaeSJed Brownthe PETSc BPs, which only included data associated with structured geometries, were
238bcb2dfaeSJed Brownrenamed `bpsraw`, and the new version of the BPs, which can handle data associated
239bcb2dfaeSJed Brownwith any unstructured geometry, were called `bps`. Additionally, other benchmark
240bcb2dfaeSJed Brownproblems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively),
241bcb2dfaeSJed Brownand BP5 and BP6 (the collocated versions---for which the quadrature points are the same
242bcb2dfaeSJed Brownas the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc
243bcb2dfaeSJed Brownexamples. Furthermoew, another standalone libCEED example, called `ex2`, which
244bcb2dfaeSJed Browncomputes the surface area of a given mesh was added to this release.
245bcb2dfaeSJed Brown
246bcb2dfaeSJed BrownBackends available in this release:
247bcb2dfaeSJed Brown
24868e843eeSJed Brown| CEED resource (`-ceed`)  | Backend                                             |
24968e843eeSJed Brown|--------------------------|-----------------------------------------------------|
25068e843eeSJed Brown| `/cpu/self/ref/serial`   | Serial reference implementation                     |
25168e843eeSJed Brown| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
25268e843eeSJed Brown| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks            |
25368e843eeSJed Brown| `/cpu/self/opt/serial`   | Serial optimized C implementation                   |
25468e843eeSJed Brown| `/cpu/self/opt/blocked`  | Blocked optimized C implementation                  |
25568e843eeSJed Brown| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
25668e843eeSJed Brown| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
25768e843eeSJed Brown| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
25868e843eeSJed Brown| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
25968e843eeSJed Brown| `/cpu/occa`              | Serial OCCA kernels                                 |
26068e843eeSJed Brown| `/gpu/occa`              | CUDA OCCA kernels                                   |
26168e843eeSJed Brown| `/omp/occa`              | OpenMP OCCA kernels                                 |
26268e843eeSJed Brown| `/ocl/occa`              | OpenCL OCCA kernels                                 |
26368e843eeSJed Brown| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
26468e843eeSJed Brown| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
26568e843eeSJed Brown| `/gpu/cuda/shared`       | Optimized pure CUDA kernels using shared memory     |
26668e843eeSJed Brown| `/gpu/cuda/gen`          | Optimized pure CUDA kernels using code generation   |
26768e843eeSJed Brown| `/gpu/magma`             | CUDA MAGMA kernels                                  |
268bcb2dfaeSJed Brown
269bcb2dfaeSJed BrownExamples available in this release:
270bcb2dfaeSJed Brown
27168e843eeSJed Brown:::{list-table}
27268e843eeSJed Brown:header-rows: 1
27368e843eeSJed Brown:widths: auto
27468e843eeSJed Brown* - User code
27568e843eeSJed Brown  - Example
27668e843eeSJed Brown* - `ceed`
27768e843eeSJed Brown  - * ex1 (volume)
27868e843eeSJed Brown    * ex2 (surface)
27968e843eeSJed Brown* - `mfem`
28068e843eeSJed Brown  - * BP1 (scalar mass operator)
28168e843eeSJed Brown    * BP3 (scalar Laplace operator)
28268e843eeSJed Brown* - `petsc`
28368e843eeSJed Brown  - * BP1 (scalar mass operator)
28468e843eeSJed Brown    * BP2 (vector mass operator)
28568e843eeSJed Brown    * BP3 (scalar Laplace operator)
28668e843eeSJed Brown    * BP4 (vector Laplace operator)
28768e843eeSJed Brown    * BP5 (collocated scalar Laplace operator)
28868e843eeSJed Brown    * BP6 (collocated vector Laplace operator)
28968e843eeSJed Brown    * Navier-Stokes
29068e843eeSJed Brown* - `nek5000`
29168e843eeSJed Brown  - * BP1 (scalar mass operator)
29268e843eeSJed Brown    * BP3 (scalar Laplace operator)
29368e843eeSJed Brown:::
294bcb2dfaeSJed Brown
295bcb2dfaeSJed Brown(v0-4)=
296bcb2dfaeSJed Brown
297bcb2dfaeSJed Brown## v0.4 (Apr 1, 2019)
298bcb2dfaeSJed Brown
299bcb2dfaeSJed BrownlibCEED v0.4 was made again publicly available in the second full CEED software
300bcb2dfaeSJed Browndistribution, release CEED 2.0. This release contained notable features, such as
301bcb2dfaeSJed Brownfour new CPU backends, two new GPU backends, CPU backend optimizations, initial
302bcb2dfaeSJed Brownsupport for operator composition, performance benchmarking, and a Navier-Stokes demo.
303bcb2dfaeSJed BrownThe new CPU backends in this release came in two families. The `/cpu/self/*/serial`
304bcb2dfaeSJed Brownbackends process one element at a time and are intended for meshes with a smaller number
305bcb2dfaeSJed Brownof high order elements. The `/cpu/self/*/blocked` backends process blocked batches of
306bcb2dfaeSJed Browneight interlaced elements and are intended for meshes with higher numbers of elements.
307bcb2dfaeSJed BrownThe `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU
308bcb2dfaeSJed Brownperformance. The `/cpu/self/xsmm/*` backends rely upon the
309bcb2dfaeSJed Brown[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU
310bcb2dfaeSJed Brownperformance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA.
311bcb2dfaeSJed BrownThe `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable
312bcb2dfaeSJed Brownperformance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple
313bcb2dfaeSJed Brownparallelization approach, where each thread treats a finite element. Using just in time
314bcb2dfaeSJed Browncompilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this
315bcb2dfaeSJed Brownbackend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend
316bcb2dfaeSJed Brownachieve good peak performance for 1D, 2D, and low order 3D problems, but performance
317bcb2dfaeSJed Browndeteriorates very quickly when threads run out of registers.
318bcb2dfaeSJed Brown
319bcb2dfaeSJed BrownA new explicit time-stepping Navier-Stokes solver was added to the family of libCEED
320bcb2dfaeSJed Brownexamples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`).
321bcb2dfaeSJed BrownThis example solves the time-dependent Navier-Stokes equations of compressible gas
322bcb2dfaeSJed Browndynamics in a static Eulerian three-dimensional frame, using structured high-order
323bcb2dfaeSJed Brownfinite/spectral element spatial discretizations and explicit high-order time-stepping
324bcb2dfaeSJed Brown(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc,
325bcb2dfaeSJed Brownso that the pointwise physics (defined at quadrature points) is separated from the
326bcb2dfaeSJed Brownparallelization and meshing concerns.
327bcb2dfaeSJed Brown
328bcb2dfaeSJed BrownBackends available in this release:
329bcb2dfaeSJed Brown
33068e843eeSJed Brown| CEED resource (`-ceed`)  | Backend                                             |
33168e843eeSJed Brown|--------------------------|-----------------------------------------------------|
33268e843eeSJed Brown| `/cpu/self/ref/serial`   | Serial reference implementation                     |
33368e843eeSJed Brown| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
33468e843eeSJed Brown| `/cpu/self/tmpl`         | Backend template, defaults to `/cpu/self/blocked`   |
33568e843eeSJed Brown| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
33668e843eeSJed Brown| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
33768e843eeSJed Brown| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
33868e843eeSJed Brown| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
33968e843eeSJed Brown| `/cpu/occa`              | Serial OCCA kernels                                 |
34068e843eeSJed Brown| `/gpu/occa`              | CUDA OCCA kernels                                   |
34168e843eeSJed Brown| `/omp/occa`              | OpenMP OCCA kernels                                 |
34268e843eeSJed Brown| `/ocl/occa`              | OpenCL OCCA kernels                                 |
34368e843eeSJed Brown| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
34468e843eeSJed Brown| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
34568e843eeSJed Brown| `/gpu/magma`             | CUDA MAGMA kernels                                  |
346bcb2dfaeSJed Brown
347bcb2dfaeSJed BrownExamples available in this release:
348bcb2dfaeSJed Brown
34968e843eeSJed Brown:::{list-table}
35068e843eeSJed Brown:header-rows: 1
35168e843eeSJed Brown:widths: auto
35268e843eeSJed Brown* - User code
35368e843eeSJed Brown  - Example
35468e843eeSJed Brown* - `ceed`
35568e843eeSJed Brown  - * ex1 (volume)
35668e843eeSJed Brown* - `mfem`
35768e843eeSJed Brown  - * BP1 (scalar mass operator)
35868e843eeSJed Brown    * BP3 (scalar Laplace operator)
35968e843eeSJed Brown* - `petsc`
36068e843eeSJed Brown  - * BP1 (scalar mass operator)
36168e843eeSJed Brown    * BP3 (scalar Laplace operator)
36268e843eeSJed Brown    * Navier-Stokes
36368e843eeSJed Brown* - `nek5000`
36468e843eeSJed Brown  - * BP1 (scalar mass operator)
36568e843eeSJed Brown    * BP3 (scalar Laplace operator)
36668e843eeSJed Brown:::
367bcb2dfaeSJed Brown
368bcb2dfaeSJed Brown(v0-3)=
369bcb2dfaeSJed Brown
370bcb2dfaeSJed Brown## v0.3 (Sep 30, 2018)
371bcb2dfaeSJed Brown
372bcb2dfaeSJed BrownNotable features in this release include active/passive field interface, support for
373bcb2dfaeSJed Brownnon-tensor bases, backend optimization, and improved Fortran interface. This release
374bcb2dfaeSJed Brownalso focused on providing improved continuous integration, and many new tests with code
375bcb2dfaeSJed Browncoverage reports of about 90%. This release also provided a significant change to the
376bcb2dfaeSJed Brownpublic interface: a {ref}`CeedQFunction` can take any number of named input and output
377bcb2dfaeSJed Brownarguments while {ref}`CeedOperator` connects them to the actual data, which may be
378bcb2dfaeSJed Brownsupplied explicitly to `CeedOperatorApply()` (active) or separately via
379bcb2dfaeSJed Brown`CeedOperatorSetField()` (passive). This interface change enables reusable libraries
380bcb2dfaeSJed Brownof CeedQFunctions and composition of block solvers constructed using
381bcb2dfaeSJed Brown{ref}`CeedOperator`. A concept of blocked restriction was added to this release and
382bcb2dfaeSJed Brownused in an optimized CPU backend. Although this is typically not visible to the user,
383bcb2dfaeSJed Brownit enables effective use of arbitrary-length SIMD while maintaining cache locality.
384bcb2dfaeSJed BrownThis CPU backend also implements an algebraic factorization of tensor product gradients
385bcb2dfaeSJed Brownto perform fewer operations than standard application of interpolation and
386bcb2dfaeSJed Browndifferentiation from nodes to quadrature points. This algebraic formulation
387bcb2dfaeSJed Brownautomatically supports non-polynomial and non-interpolatory bases, thus is more general
388bcb2dfaeSJed Brownthan the more common derivation in terms of Lagrange polynomials on the quadrature points.
389bcb2dfaeSJed Brown
390bcb2dfaeSJed BrownBackends available in this release:
391bcb2dfaeSJed Brown
39268e843eeSJed Brown| CEED resource (`-ceed`) | Backend                                             |
39368e843eeSJed Brown|-------------------------|-----------------------------------------------------|
39468e843eeSJed Brown| `/cpu/self/blocked`     | Blocked reference implementation                    |
39568e843eeSJed Brown| `/cpu/self/ref`         | Serial reference implementation                     |
39668e843eeSJed Brown| `/cpu/self/tmpl`        | Backend template, defaults to `/cpu/self/blocked`   |
39768e843eeSJed Brown| `/cpu/occa`             | Serial OCCA kernels                                 |
39868e843eeSJed Brown| `/gpu/occa`             | CUDA OCCA kernels                                   |
39968e843eeSJed Brown| `/omp/occa`             | OpenMP OCCA kernels                                 |
40068e843eeSJed Brown| `/ocl/occa`             | OpenCL OCCA kernels                                 |
40168e843eeSJed Brown| `/gpu/magma`            | CUDA MAGMA kernels                                  |
402bcb2dfaeSJed Brown
403bcb2dfaeSJed BrownExamples available in this release:
404bcb2dfaeSJed Brown
40568e843eeSJed Brown:::{list-table}
40668e843eeSJed Brown:header-rows: 1
40768e843eeSJed Brown:widths: auto
40868e843eeSJed Brown* - User code
40968e843eeSJed Brown  - Example
41068e843eeSJed Brown* - `ceed`
41168e843eeSJed Brown  - * ex1 (volume)
41268e843eeSJed Brown* - `mfem`
41368e843eeSJed Brown  - * BP1 (scalar mass operator)
41468e843eeSJed Brown    * BP3 (scalar Laplace operator)
41568e843eeSJed Brown* - `petsc`
41668e843eeSJed Brown  - * BP1 (scalar mass operator)
41768e843eeSJed Brown    * BP3 (scalar Laplace operator)
41868e843eeSJed Brown* - `nek5000`
41968e843eeSJed Brown  - * BP1 (scalar mass operator)
42068e843eeSJed Brown    * BP3 (scalar Laplace operator)
42168e843eeSJed Brown:::
422bcb2dfaeSJed Brown
423bcb2dfaeSJed Brown(v0-21)=
424bcb2dfaeSJed Brown
425bcb2dfaeSJed Brown## v0.21 (Sep 30, 2018)
426bcb2dfaeSJed Brown
427bcb2dfaeSJed BrownA MAGMA backend (which relies upon the
428bcb2dfaeSJed Brown[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this
429bcb2dfaeSJed Brownrelease. This initial integration set up the framework of using MAGMA and provided the
430bcb2dfaeSJed BrownlibCEED functionality through MAGMA kernels as one of libCEED’s computational backends.
431bcb2dfaeSJed BrownAs any other backend, the MAGMA backend provides extended basic data structures for
432bcb2dfaeSJed Brown{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements
433bcb2dfaeSJed Brownthe fundamental CEED building blocks to work with the new data structures.
434bcb2dfaeSJed BrownIn general, the MAGMA-specific data structures keep the libCEED pointers to CPU data
435bcb2dfaeSJed Brownbut also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled
436bcb2dfaeSJed Browninternally, and thus seamlessly to the user, through the functions/methods that are
437bcb2dfaeSJed Brownprovided to support them.
438bcb2dfaeSJed Brown
439bcb2dfaeSJed BrownBackends available in this release:
440bcb2dfaeSJed Brown
44168e843eeSJed Brown| CEED resource (`-ceed`) | Backend                         |
44268e843eeSJed Brown|-------------------------|---------------------------------|
44368e843eeSJed Brown| `/cpu/self`             | Serial reference implementation |
44468e843eeSJed Brown| `/cpu/occa`             | Serial OCCA kernels             |
44568e843eeSJed Brown| `/gpu/occa`             | CUDA OCCA kernels               |
44668e843eeSJed Brown| `/omp/occa`             | OpenMP OCCA kernels             |
44768e843eeSJed Brown| `/ocl/occa`             | OpenCL OCCA kernels             |
44868e843eeSJed Brown| `/gpu/magma`            | CUDA MAGMA kernels              |
449bcb2dfaeSJed Brown
450bcb2dfaeSJed BrownExamples available in this release:
451bcb2dfaeSJed Brown
45268e843eeSJed Brown:::{list-table}
45368e843eeSJed Brown:header-rows: 1
45468e843eeSJed Brown:widths: auto
45568e843eeSJed Brown* - User code
45668e843eeSJed Brown  - Example
45768e843eeSJed Brown* - `ceed`
45868e843eeSJed Brown  - * ex1 (volume)
45968e843eeSJed Brown* - `mfem`
46068e843eeSJed Brown  - * BP1 (scalar mass operator)
46168e843eeSJed Brown    * BP3 (scalar Laplace operator)
46268e843eeSJed Brown* - `petsc`
46368e843eeSJed Brown  - * BP1 (scalar mass operator)
46468e843eeSJed Brown* - `nek5000`
46568e843eeSJed Brown  - * BP1 (scalar mass operator)
46668e843eeSJed Brown:::
467bcb2dfaeSJed Brown
468bcb2dfaeSJed Brown(v0-2)=
469bcb2dfaeSJed Brown
470bcb2dfaeSJed Brown## v0.2 (Mar 30, 2018)
471bcb2dfaeSJed Brown
472bcb2dfaeSJed BrownlibCEED was made publicly available the first full CEED software distribution, release
473bcb2dfaeSJed BrownCEED 1.0. The distribution was made available using the Spack package manager to provide
474bcb2dfaeSJed Browna common, easy-to-use build environment, where the user can build the CEED distribution
475bcb2dfaeSJed Brownwith all dependencies. This release included a new Fortran interface for the library.
476bcb2dfaeSJed BrownThis release also contained major improvements in the OCCA backend (including a new
477bcb2dfaeSJed Brown`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to
478bcb2dfaeSJed Browncompute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an
479bcb2dfaeSJed Brown`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application
480bcb2dfaeSJed Brownof the Laplace operator) was also added to this release.
481bcb2dfaeSJed Brown
482bcb2dfaeSJed BrownBackends available in this release:
483bcb2dfaeSJed Brown
48468e843eeSJed Brown| CEED resource (`-ceed`) | Backend                         |
48568e843eeSJed Brown|-------------------------|---------------------------------|
48668e843eeSJed Brown| `/cpu/self`             | Serial reference implementation |
48768e843eeSJed Brown| `/cpu/occa`             | Serial OCCA kernels             |
48868e843eeSJed Brown| `/gpu/occa`             | CUDA OCCA kernels               |
48968e843eeSJed Brown| `/omp/occa`             | OpenMP OCCA kernels             |
49068e843eeSJed Brown| `/ocl/occa`             | OpenCL OCCA kernels             |
491bcb2dfaeSJed Brown
492bcb2dfaeSJed BrownExamples available in this release:
493bcb2dfaeSJed Brown
49468e843eeSJed Brown:::{list-table}
49568e843eeSJed Brown:header-rows: 1
49668e843eeSJed Brown:widths: auto
49768e843eeSJed Brown* - User code
49868e843eeSJed Brown  - Example
49968e843eeSJed Brown* - `ceed`
50068e843eeSJed Brown  - * ex1 (volume)
50168e843eeSJed Brown* - `mfem`
50268e843eeSJed Brown  - * BP1 (scalar mass operator)
50368e843eeSJed Brown    * BP3 (scalar Laplace operator)
50468e843eeSJed Brown* - `petsc`
50568e843eeSJed Brown  - * BP1 (scalar mass operator)
50668e843eeSJed Brown* - `nek5000`
50768e843eeSJed Brown  - * BP1 (scalar mass operator)
50868e843eeSJed Brown:::
509bcb2dfaeSJed Brown
510bcb2dfaeSJed Brown(v0-1)=
511bcb2dfaeSJed Brown
512bcb2dfaeSJed Brown## v0.1 (Jan 3, 2018)
513bcb2dfaeSJed Brown
514bcb2dfaeSJed BrownInitial low-level API of the CEED project. The low-level API provides a set of Finite
515bcb2dfaeSJed BrownElements kernels and components for writing new low-level kernels. Examples include:
516bcb2dfaeSJed Brownvector and sparse linear algebra, element matrix assembly over a batch of elements,
517bcb2dfaeSJed Brownpartial assembly and action for efficient high-order operators like mass, diffusion,
518bcb2dfaeSJed Brownadvection, etc. The main goal of the low-level API is to establish the basis for the
519bcb2dfaeSJed Brownhigh-level API. Also, identifying such low-level kernels and providing a reference
520bcb2dfaeSJed Brownimplementation for them serves as the basis for specialized backend implementations.
521bcb2dfaeSJed BrownThis release contained several backends: `/cpu/self`, and backends which rely upon the
522bcb2dfaeSJed Brown[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`,
523bcb2dfaeSJed Brown`/gpu/occa`, and `/omp/occa`.
524bcb2dfaeSJed BrownIt also included several examples, in the `examples` folder:
525bcb2dfaeSJed BrownA standalone code that shows the usage of libCEED (with no external
526bcb2dfaeSJed Browndependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1
527bcb2dfaeSJed Brown(with the application of the mass operator); and a `petsc` example to perform BP1
528bcb2dfaeSJed Brown(with the application of the mass operator).
529bcb2dfaeSJed Brown
530bcb2dfaeSJed BrownBackends available in this release:
531bcb2dfaeSJed Brown
53268e843eeSJed Brown| CEED resource (`-ceed`) | Backend                         |
53368e843eeSJed Brown|-------------------------|---------------------------------|
53468e843eeSJed Brown| `/cpu/self`             | Serial reference implementation |
53568e843eeSJed Brown| `/cpu/occa`             | Serial OCCA kernels             |
53668e843eeSJed Brown| `/gpu/occa`             | CUDA OCCA kernels               |
53768e843eeSJed Brown| `/omp/occa`             | OpenMP OCCA kernels             |
538bcb2dfaeSJed Brown
539bcb2dfaeSJed BrownExamples available in this release:
540bcb2dfaeSJed Brown
541bcb2dfaeSJed Brown| User code             | Example                           |
54268e843eeSJed Brown|-----------------------|-----------------------------------|
54368e843eeSJed Brown| `ceed`                | ex1 (scalar Laplace operator)     |
54468e843eeSJed Brown| `mfem`                | BP1 (scalar mass operator)        |
54568e843eeSJed Brown| `petsc`               | BP1 (scalar mass operator)        |
546bcb2dfaeSJed Brown```
547