xref: /libCEED/doc/sphinx/source/releasenotes.md (revision ce4822f6b3f943b8351eb0be75b7b9f045298d6a)
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.
13*ce4822f6SJeremy 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.
157e7773b5SJeremy L Thompson
16f479eb23SJeremy L Thompson### New features
17f479eb23SJeremy L Thompson
18f479eb23SJeremy L Thompson- `CeedScalar` can now be set as `float` or `double` at compile time.
19f479eb23SJeremy L Thompson
20bcb2dfaeSJed Brown### Maintainability
21bcb2dfaeSJed Brown
22bcb2dfaeSJed Brown- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage.
23bcb2dfaeSJed Brown
24bcb2dfaeSJed Brown(v0-9)=
25bcb2dfaeSJed Brown
26bcb2dfaeSJed Brown## v0.9 (Jul 6, 2021)
27bcb2dfaeSJed Brown
28bcb2dfaeSJed Brown### Interface changes
29bcb2dfaeSJed Brown
30bcb2dfaeSJed Brown- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact.
31bcb2dfaeSJed Brown
32bcb2dfaeSJed Brown### New features
33bcb2dfaeSJed Brown
34bcb2dfaeSJed Brown- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use.
35bcb2dfaeSJed Brown- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends.
36bcb2dfaeSJed Brown- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations.
37bcb2dfaeSJed Brown- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed).
38bcb2dfaeSJed Brown
39bcb2dfaeSJed Brown### Performance improvements
40bcb2dfaeSJed Brown
41bcb2dfaeSJed Brown### Examples
42bcb2dfaeSJed Brown
43bcb2dfaeSJed Brown- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations.
44bcb2dfaeSJed Brown- Fluid mechanics example adds GPU support and improves modularity.
45bcb2dfaeSJed Brown
46bcb2dfaeSJed Brown### Deprecated backends
47bcb2dfaeSJed Brown
48bcb2dfaeSJed 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.
49bcb2dfaeSJed Brown
50bcb2dfaeSJed Brown(v0-8)=
51bcb2dfaeSJed Brown
52bcb2dfaeSJed Brown## v0.8 (Mar 31, 2021)
53bcb2dfaeSJed Brown
54bcb2dfaeSJed Brown### Interface changes
55bcb2dfaeSJed Brown
56bcb2dfaeSJed Brown- Error handling improved to include enumerated error codes for C interface return values.
57bcb2dfaeSJed 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.
58bcb2dfaeSJed Brown
59bcb2dfaeSJed Brown### New features
60bcb2dfaeSJed Brown
61bcb2dfaeSJed Brown- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features.
62bcb2dfaeSJed Brown- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly.
63bcb2dfaeSJed Brown- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators.
64bcb2dfaeSJed Brown
65bcb2dfaeSJed Brown### Performance improvements
66bcb2dfaeSJed Brown
67bcb2dfaeSJed Brown- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`.
68bcb2dfaeSJed Brown- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`.
69bcb2dfaeSJed Brown
70bcb2dfaeSJed Brown### Examples
71bcb2dfaeSJed Brown
72bcb2dfaeSJed Brown- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions.
73bcb2dfaeSJed 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.
74bcb2dfaeSJed 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.
75bcb2dfaeSJed Brown- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree.
76bcb2dfaeSJed Brown
77bcb2dfaeSJed Brown(v0-7)=
78bcb2dfaeSJed Brown
79bcb2dfaeSJed Brown## v0.7 (Sep 29, 2020)
80bcb2dfaeSJed Brown
81bcb2dfaeSJed Brown### Interface changes
82bcb2dfaeSJed Brown
83bcb2dfaeSJed Brown- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors.
84bcb2dfaeSJed 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`.
85bcb2dfaeSJed Brown  These changes improve support for mixed finite element methods.
86bcb2dfaeSJed Brown- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature.
87bcb2dfaeSJed Brown- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity.
88bcb2dfaeSJed 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.
89bcb2dfaeSJed 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.
90bcb2dfaeSJed 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.
91bcb2dfaeSJed 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.
92bcb2dfaeSJed 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`.
93bcb2dfaeSJed Brown- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory.
94bcb2dfaeSJed 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.
95bcb2dfaeSJed Brown
96bcb2dfaeSJed Brown### New features
97bcb2dfaeSJed Brown
98bcb2dfaeSJed Brown- New HIP backend: `/gpu/hip/ref`.
99bcb2dfaeSJed Brown- CeedQFunction support for user `CUfunction`s in some backends
100bcb2dfaeSJed Brown
101bcb2dfaeSJed Brown### Performance improvements
102bcb2dfaeSJed Brown
103bcb2dfaeSJed Brown- OCCA backend rebuilt to facilitate future performance enhancements.
104bcb2dfaeSJed Brown- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`.
105bcb2dfaeSJed Brown
106bcb2dfaeSJed Brown### Examples
107bcb2dfaeSJed Brown
108bcb2dfaeSJed Brown- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions.
109bcb2dfaeSJed Brown
110bcb2dfaeSJed Brown### Deprecated backends
111bcb2dfaeSJed Brown
112bcb2dfaeSJed Brown- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`.
113bcb2dfaeSJed Brown
114bcb2dfaeSJed Brown(v0-6)=
115bcb2dfaeSJed Brown
116bcb2dfaeSJed Brown## v0.6 (Mar 29, 2020)
117bcb2dfaeSJed Brown
118bcb2dfaeSJed BrownlibCEED v0.6 contains numerous new features and examples, as well as expanded
119bcb2dfaeSJed Browndocumentation in [this new website](https://libceed.readthedocs.io).
120bcb2dfaeSJed Brown
121bcb2dfaeSJed Brown### New features
122bcb2dfaeSJed Brown
123bcb2dfaeSJed Brown- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly
124bcb2dfaeSJed Brown  1-1 correspondence with the C interface, plus some convenience features.  For instance,
125bcb2dfaeSJed Brown  data stored in the {cpp:type}`CeedVector` structure are available without copy as
126bcb2dfaeSJed Brown  {py:class}`numpy.ndarray`.  Short tutorials are provided in
127bcb2dfaeSJed Brown  [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/).
128bcb2dfaeSJed Brown- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point,
129bcb2dfaeSJed Brown  {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal
130bcb2dfaeSJed Brown  ({c:func}`CeedOperatorAssembleLinearDiagonal`).  These operations are useful for
131bcb2dfaeSJed Brown  preconditioning ingredients and are used in the libCEED's multigrid examples.
132bcb2dfaeSJed Brown- The inverse of separable operators can be obtained using
133bcb2dfaeSJed Brown  {c:func}`CeedOperatorCreateFDMElementInverse` and applied with
134bcb2dfaeSJed Brown  {c:func}`CeedOperatorApply`.  This is a useful preconditioning ingredient,
135bcb2dfaeSJed Brown  especially for Laplacians and related operators.
136bcb2dfaeSJed Brown- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`,
137bcb2dfaeSJed Brown  {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`.
138bcb2dfaeSJed Brown- Make public accessors for various attributes to facilitate writing composable code.
139bcb2dfaeSJed Brown- New backend: `/cpu/self/memcheck/serial`.
140bcb2dfaeSJed Brown- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA
141bcb2dfaeSJed Brown  backends.  (Single source is coming soon for OCCA backends.)
142bcb2dfaeSJed Brown- Fix some missing edge cases in CUDA backend.
143bcb2dfaeSJed Brown
144bcb2dfaeSJed Brown### Performance Improvements
145bcb2dfaeSJed Brown
146bcb2dfaeSJed Brown- MAGMA backend performance optimization and non-tensor bases.
147bcb2dfaeSJed Brown- No-copy optimization in {c:func}`CeedOperatorApply`.
148bcb2dfaeSJed Brown
149bcb2dfaeSJed Brown### Interface changes
150bcb2dfaeSJed Brown
151bcb2dfaeSJed Brown- Replace {code}`CeedElemRestrictionCreateIdentity` and
152bcb2dfaeSJed Brown  {code}`CeedElemRestrictionCreateBlocked` with more flexible
153bcb2dfaeSJed Brown  {c:func}`CeedElemRestrictionCreateStrided` and
154bcb2dfaeSJed Brown  {c:func}`CeedElemRestrictionCreateBlockedStrided`.
155bcb2dfaeSJed Brown- Add arguments to {c:func}`CeedQFunctionCreateIdentity`.
156bcb2dfaeSJed Brown- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification
157bcb2dfaeSJed Brown  with {cpp:enum}`CeedInterlaceMode`.  This is now an attribute of the
158bcb2dfaeSJed Brown  {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no
159bcb2dfaeSJed Brown  longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and
160bcb2dfaeSJed Brown  {c:func}`CeedElemRestrictionApply`.
161bcb2dfaeSJed Brown
162bcb2dfaeSJed Brown### Examples
163bcb2dfaeSJed Brown
164bcb2dfaeSJed BrownlibCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`.
165bcb2dfaeSJed BrownNotable additions include:
166bcb2dfaeSJed Brown
167bcb2dfaeSJed Brown- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of
168bcb2dfaeSJed Brown  a domain in 1, 2, and 3 dimensions by applying a Laplacian.
169bcb2dfaeSJed Brown
170bcb2dfaeSJed Brown- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area
171bcb2dfaeSJed Brown  of domains (like the cube and sphere) by direct integration on a surface mesh;
172bcb2dfaeSJed Brown  demonstrates geometric dimension different from topological dimension.
173bcb2dfaeSJed Brown
174bcb2dfaeSJed Brown- PETSc {ref}`example-petsc-bps`:
175bcb2dfaeSJed Brown
176bcb2dfaeSJed Brown  - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support.
177bcb2dfaeSJed Brown  - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements
178bcb2dfaeSJed Brown    and transparent CUDA support.
179bcb2dfaeSJed Brown  - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`):
180bcb2dfaeSJed Brown    generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric
181bcb2dfaeSJed Brown    dimension different from topological dimension.
182bcb2dfaeSJed Brown
183bcb2dfaeSJed Brown- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid
184bcb2dfaeSJed Brown  solver with algebraic multigrid coarse solve.
185bcb2dfaeSJed Brown
186bcb2dfaeSJed Brown- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly
187bcb2dfaeSJed Brown  `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`),
188bcb2dfaeSJed Brown  implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and
189bcb2dfaeSJed Brown  quasi-2D computational domain support.
190bcb2dfaeSJed Brown
191bcb2dfaeSJed Brown- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for
192bcb2dfaeSJed Brown  linear elasticity, small-strain hyperelasticity, and globalized finite-strain
193bcb2dfaeSJed Brown  hyperelasticity using p-multigrid with algebraic multigrid coarse solve.
194bcb2dfaeSJed Brown
195bcb2dfaeSJed Brown(v0-5)=
196bcb2dfaeSJed Brown
197bcb2dfaeSJed Brown## v0.5 (Sep 18, 2019)
198bcb2dfaeSJed Brown
199bcb2dfaeSJed BrownFor this release, several improvements were made. Two new CUDA backends were added to
200bcb2dfaeSJed Brownthe family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art
201bcb2dfaeSJed Brownperformance using single-source {ref}`CeedQFunction`. From this release, users
202bcb2dfaeSJed Browncan define Q-Functions in a single source code independently of the targeted backend
203bcb2dfaeSJed Brownwith the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU
204bcb2dfaeSJed Browncompilation of the user provided {ref}`CeedQFunction` code. To allow a unified
205bcb2dfaeSJed Browndeclaration, the {ref}`CeedQFunction` API has undergone a slight change:
206bcb2dfaeSJed Brownthe `QFunctionField` parameter `ncomp` has been changed to `size`. This change
207bcb2dfaeSJed Brownrequires setting the previous value of `ncomp` to `ncomp*dim` when adding a
208bcb2dfaeSJed Brown`QFunctionField` with eval mode `CEED EVAL GRAD`.
209bcb2dfaeSJed Brown
210bcb2dfaeSJed BrownAdditionally, new CPU backends
211bcb2dfaeSJed Brownwere included in this release, such as the `/cpu/self/opt/*` backends (which are
212bcb2dfaeSJed Brownwritten in pure C and use partial **E-vectors** to improve performance) and the
213bcb2dfaeSJed Brown`/cpu/self/ref/memcheck` backend (which relies upon the
214bcb2dfaeSJed Brown[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user
215bcb2dfaeSJed Brown{ref}`CeedQFunction` have no undefined values).
216bcb2dfaeSJed BrownThis release also included various performance improvements, bug fixes, new examples,
217bcb2dfaeSJed Brownand improved tests. Among these improvements, vectorized instructions for
218bcb2dfaeSJed Brown{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD`
219bcb2dfaeSJed Browninstead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and
220bcb2dfaeSJed Brownidentity Q-Functions were introduced, and the PETSc benchmark problems were expanded
221bcb2dfaeSJed Brownto include unstructured meshes handling were. For this expansion, the prior version of
222bcb2dfaeSJed Brownthe PETSc BPs, which only included data associated with structured geometries, were
223bcb2dfaeSJed Brownrenamed `bpsraw`, and the new version of the BPs, which can handle data associated
224bcb2dfaeSJed Brownwith any unstructured geometry, were called `bps`. Additionally, other benchmark
225bcb2dfaeSJed Brownproblems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively),
226bcb2dfaeSJed Brownand BP5 and BP6 (the collocated versions---for which the quadrature points are the same
227bcb2dfaeSJed Brownas the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc
228bcb2dfaeSJed Brownexamples. Furthermoew, another standalone libCEED example, called `ex2`, which
229bcb2dfaeSJed Browncomputes the surface area of a given mesh was added to this release.
230bcb2dfaeSJed Brown
231bcb2dfaeSJed BrownBackends available in this release:
232bcb2dfaeSJed Brown
23368e843eeSJed Brown| CEED resource (`-ceed`)  | Backend                                             |
23468e843eeSJed Brown|--------------------------|-----------------------------------------------------|
23568e843eeSJed Brown| `/cpu/self/ref/serial`   | Serial reference implementation                     |
23668e843eeSJed Brown| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
23768e843eeSJed Brown| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks            |
23868e843eeSJed Brown| `/cpu/self/opt/serial`   | Serial optimized C implementation                   |
23968e843eeSJed Brown| `/cpu/self/opt/blocked`  | Blocked optimized C implementation                  |
24068e843eeSJed Brown| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
24168e843eeSJed Brown| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
24268e843eeSJed Brown| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
24368e843eeSJed Brown| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
24468e843eeSJed Brown| `/cpu/occa`              | Serial OCCA kernels                                 |
24568e843eeSJed Brown| `/gpu/occa`              | CUDA OCCA kernels                                   |
24668e843eeSJed Brown| `/omp/occa`              | OpenMP OCCA kernels                                 |
24768e843eeSJed Brown| `/ocl/occa`              | OpenCL OCCA kernels                                 |
24868e843eeSJed Brown| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
24968e843eeSJed Brown| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
25068e843eeSJed Brown| `/gpu/cuda/shared`       | Optimized pure CUDA kernels using shared memory     |
25168e843eeSJed Brown| `/gpu/cuda/gen`          | Optimized pure CUDA kernels using code generation   |
25268e843eeSJed Brown| `/gpu/magma`             | CUDA MAGMA kernels                                  |
253bcb2dfaeSJed Brown
254bcb2dfaeSJed BrownExamples available in this release:
255bcb2dfaeSJed Brown
25668e843eeSJed Brown:::{list-table}
25768e843eeSJed Brown:header-rows: 1
25868e843eeSJed Brown:widths: auto
25968e843eeSJed Brown* - User code
26068e843eeSJed Brown  - Example
26168e843eeSJed Brown* - `ceed`
26268e843eeSJed Brown  - * ex1 (volume)
26368e843eeSJed Brown    * ex2 (surface)
26468e843eeSJed Brown* - `mfem`
26568e843eeSJed Brown  - * BP1 (scalar mass operator)
26668e843eeSJed Brown    * BP3 (scalar Laplace operator)
26768e843eeSJed Brown* - `petsc`
26868e843eeSJed Brown  - * BP1 (scalar mass operator)
26968e843eeSJed Brown    * BP2 (vector mass operator)
27068e843eeSJed Brown    * BP3 (scalar Laplace operator)
27168e843eeSJed Brown    * BP4 (vector Laplace operator)
27268e843eeSJed Brown    * BP5 (collocated scalar Laplace operator)
27368e843eeSJed Brown    * BP6 (collocated vector Laplace operator)
27468e843eeSJed Brown    * Navier-Stokes
27568e843eeSJed Brown* - `nek5000`
27668e843eeSJed Brown  - * BP1 (scalar mass operator)
27768e843eeSJed Brown    * BP3 (scalar Laplace operator)
27868e843eeSJed Brown:::
279bcb2dfaeSJed Brown
280bcb2dfaeSJed Brown(v0-4)=
281bcb2dfaeSJed Brown
282bcb2dfaeSJed Brown## v0.4 (Apr 1, 2019)
283bcb2dfaeSJed Brown
284bcb2dfaeSJed BrownlibCEED v0.4 was made again publicly available in the second full CEED software
285bcb2dfaeSJed Browndistribution, release CEED 2.0. This release contained notable features, such as
286bcb2dfaeSJed Brownfour new CPU backends, two new GPU backends, CPU backend optimizations, initial
287bcb2dfaeSJed Brownsupport for operator composition, performance benchmarking, and a Navier-Stokes demo.
288bcb2dfaeSJed BrownThe new CPU backends in this release came in two families. The `/cpu/self/*/serial`
289bcb2dfaeSJed Brownbackends process one element at a time and are intended for meshes with a smaller number
290bcb2dfaeSJed Brownof high order elements. The `/cpu/self/*/blocked` backends process blocked batches of
291bcb2dfaeSJed Browneight interlaced elements and are intended for meshes with higher numbers of elements.
292bcb2dfaeSJed BrownThe `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU
293bcb2dfaeSJed Brownperformance. The `/cpu/self/xsmm/*` backends rely upon the
294bcb2dfaeSJed Brown[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU
295bcb2dfaeSJed Brownperformance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA.
296bcb2dfaeSJed BrownThe `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable
297bcb2dfaeSJed Brownperformance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple
298bcb2dfaeSJed Brownparallelization approach, where each thread treats a finite element. Using just in time
299bcb2dfaeSJed Browncompilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this
300bcb2dfaeSJed Brownbackend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend
301bcb2dfaeSJed Brownachieve good peak performance for 1D, 2D, and low order 3D problems, but performance
302bcb2dfaeSJed Browndeteriorates very quickly when threads run out of registers.
303bcb2dfaeSJed Brown
304bcb2dfaeSJed BrownA new explicit time-stepping Navier-Stokes solver was added to the family of libCEED
305bcb2dfaeSJed Brownexamples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`).
306bcb2dfaeSJed BrownThis example solves the time-dependent Navier-Stokes equations of compressible gas
307bcb2dfaeSJed Browndynamics in a static Eulerian three-dimensional frame, using structured high-order
308bcb2dfaeSJed Brownfinite/spectral element spatial discretizations and explicit high-order time-stepping
309bcb2dfaeSJed Brown(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc,
310bcb2dfaeSJed Brownso that the pointwise physics (defined at quadrature points) is separated from the
311bcb2dfaeSJed Brownparallelization and meshing concerns.
312bcb2dfaeSJed Brown
313bcb2dfaeSJed BrownBackends available in this release:
314bcb2dfaeSJed Brown
31568e843eeSJed Brown| CEED resource (`-ceed`)  | Backend                                             |
31668e843eeSJed Brown|--------------------------|-----------------------------------------------------|
31768e843eeSJed Brown| `/cpu/self/ref/serial`   | Serial reference implementation                     |
31868e843eeSJed Brown| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
31968e843eeSJed Brown| `/cpu/self/tmpl`         | Backend template, defaults to `/cpu/self/blocked`   |
32068e843eeSJed Brown| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
32168e843eeSJed Brown| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
32268e843eeSJed Brown| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
32368e843eeSJed Brown| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
32468e843eeSJed Brown| `/cpu/occa`              | Serial OCCA kernels                                 |
32568e843eeSJed Brown| `/gpu/occa`              | CUDA OCCA kernels                                   |
32668e843eeSJed Brown| `/omp/occa`              | OpenMP OCCA kernels                                 |
32768e843eeSJed Brown| `/ocl/occa`              | OpenCL OCCA kernels                                 |
32868e843eeSJed Brown| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
32968e843eeSJed Brown| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
33068e843eeSJed Brown| `/gpu/magma`             | CUDA MAGMA kernels                                  |
331bcb2dfaeSJed Brown
332bcb2dfaeSJed BrownExamples available in this release:
333bcb2dfaeSJed Brown
33468e843eeSJed Brown:::{list-table}
33568e843eeSJed Brown:header-rows: 1
33668e843eeSJed Brown:widths: auto
33768e843eeSJed Brown* - User code
33868e843eeSJed Brown  - Example
33968e843eeSJed Brown* - `ceed`
34068e843eeSJed Brown  - * ex1 (volume)
34168e843eeSJed Brown* - `mfem`
34268e843eeSJed Brown  - * BP1 (scalar mass operator)
34368e843eeSJed Brown    * BP3 (scalar Laplace operator)
34468e843eeSJed Brown* - `petsc`
34568e843eeSJed Brown  - * BP1 (scalar mass operator)
34668e843eeSJed Brown    * BP3 (scalar Laplace operator)
34768e843eeSJed Brown    * Navier-Stokes
34868e843eeSJed Brown* - `nek5000`
34968e843eeSJed Brown  - * BP1 (scalar mass operator)
35068e843eeSJed Brown    * BP3 (scalar Laplace operator)
35168e843eeSJed Brown:::
352bcb2dfaeSJed Brown
353bcb2dfaeSJed Brown(v0-3)=
354bcb2dfaeSJed Brown
355bcb2dfaeSJed Brown## v0.3 (Sep 30, 2018)
356bcb2dfaeSJed Brown
357bcb2dfaeSJed BrownNotable features in this release include active/passive field interface, support for
358bcb2dfaeSJed Brownnon-tensor bases, backend optimization, and improved Fortran interface. This release
359bcb2dfaeSJed Brownalso focused on providing improved continuous integration, and many new tests with code
360bcb2dfaeSJed Browncoverage reports of about 90%. This release also provided a significant change to the
361bcb2dfaeSJed Brownpublic interface: a {ref}`CeedQFunction` can take any number of named input and output
362bcb2dfaeSJed Brownarguments while {ref}`CeedOperator` connects them to the actual data, which may be
363bcb2dfaeSJed Brownsupplied explicitly to `CeedOperatorApply()` (active) or separately via
364bcb2dfaeSJed Brown`CeedOperatorSetField()` (passive). This interface change enables reusable libraries
365bcb2dfaeSJed Brownof CeedQFunctions and composition of block solvers constructed using
366bcb2dfaeSJed Brown{ref}`CeedOperator`. A concept of blocked restriction was added to this release and
367bcb2dfaeSJed Brownused in an optimized CPU backend. Although this is typically not visible to the user,
368bcb2dfaeSJed Brownit enables effective use of arbitrary-length SIMD while maintaining cache locality.
369bcb2dfaeSJed BrownThis CPU backend also implements an algebraic factorization of tensor product gradients
370bcb2dfaeSJed Brownto perform fewer operations than standard application of interpolation and
371bcb2dfaeSJed Browndifferentiation from nodes to quadrature points. This algebraic formulation
372bcb2dfaeSJed Brownautomatically supports non-polynomial and non-interpolatory bases, thus is more general
373bcb2dfaeSJed Brownthan the more common derivation in terms of Lagrange polynomials on the quadrature points.
374bcb2dfaeSJed Brown
375bcb2dfaeSJed BrownBackends available in this release:
376bcb2dfaeSJed Brown
37768e843eeSJed Brown| CEED resource (`-ceed`) | Backend                                             |
37868e843eeSJed Brown|-------------------------|-----------------------------------------------------|
37968e843eeSJed Brown| `/cpu/self/blocked`     | Blocked reference implementation                    |
38068e843eeSJed Brown| `/cpu/self/ref`         | Serial reference implementation                     |
38168e843eeSJed Brown| `/cpu/self/tmpl`        | Backend template, defaults to `/cpu/self/blocked`   |
38268e843eeSJed Brown| `/cpu/occa`             | Serial OCCA kernels                                 |
38368e843eeSJed Brown| `/gpu/occa`             | CUDA OCCA kernels                                   |
38468e843eeSJed Brown| `/omp/occa`             | OpenMP OCCA kernels                                 |
38568e843eeSJed Brown| `/ocl/occa`             | OpenCL OCCA kernels                                 |
38668e843eeSJed Brown| `/gpu/magma`            | CUDA MAGMA kernels                                  |
387bcb2dfaeSJed Brown
388bcb2dfaeSJed BrownExamples available in this release:
389bcb2dfaeSJed Brown
39068e843eeSJed Brown:::{list-table}
39168e843eeSJed Brown:header-rows: 1
39268e843eeSJed Brown:widths: auto
39368e843eeSJed Brown* - User code
39468e843eeSJed Brown  - Example
39568e843eeSJed Brown* - `ceed`
39668e843eeSJed Brown  - * ex1 (volume)
39768e843eeSJed Brown* - `mfem`
39868e843eeSJed Brown  - * BP1 (scalar mass operator)
39968e843eeSJed Brown    * BP3 (scalar Laplace operator)
40068e843eeSJed Brown* - `petsc`
40168e843eeSJed Brown  - * BP1 (scalar mass operator)
40268e843eeSJed Brown    * BP3 (scalar Laplace operator)
40368e843eeSJed Brown* - `nek5000`
40468e843eeSJed Brown  - * BP1 (scalar mass operator)
40568e843eeSJed Brown    * BP3 (scalar Laplace operator)
40668e843eeSJed Brown:::
407bcb2dfaeSJed Brown
408bcb2dfaeSJed Brown(v0-21)=
409bcb2dfaeSJed Brown
410bcb2dfaeSJed Brown## v0.21 (Sep 30, 2018)
411bcb2dfaeSJed Brown
412bcb2dfaeSJed BrownA MAGMA backend (which relies upon the
413bcb2dfaeSJed Brown[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this
414bcb2dfaeSJed Brownrelease. This initial integration set up the framework of using MAGMA and provided the
415bcb2dfaeSJed BrownlibCEED functionality through MAGMA kernels as one of libCEED’s computational backends.
416bcb2dfaeSJed BrownAs any other backend, the MAGMA backend provides extended basic data structures for
417bcb2dfaeSJed Brown{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements
418bcb2dfaeSJed Brownthe fundamental CEED building blocks to work with the new data structures.
419bcb2dfaeSJed BrownIn general, the MAGMA-specific data structures keep the libCEED pointers to CPU data
420bcb2dfaeSJed Brownbut also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled
421bcb2dfaeSJed Browninternally, and thus seamlessly to the user, through the functions/methods that are
422bcb2dfaeSJed Brownprovided to support them.
423bcb2dfaeSJed Brown
424bcb2dfaeSJed BrownBackends available in this release:
425bcb2dfaeSJed Brown
42668e843eeSJed Brown| CEED resource (`-ceed`) | Backend                         |
42768e843eeSJed Brown|-------------------------|---------------------------------|
42868e843eeSJed Brown| `/cpu/self`             | Serial reference implementation |
42968e843eeSJed Brown| `/cpu/occa`             | Serial OCCA kernels             |
43068e843eeSJed Brown| `/gpu/occa`             | CUDA OCCA kernels               |
43168e843eeSJed Brown| `/omp/occa`             | OpenMP OCCA kernels             |
43268e843eeSJed Brown| `/ocl/occa`             | OpenCL OCCA kernels             |
43368e843eeSJed Brown| `/gpu/magma`            | CUDA MAGMA kernels              |
434bcb2dfaeSJed Brown
435bcb2dfaeSJed BrownExamples available in this release:
436bcb2dfaeSJed Brown
43768e843eeSJed Brown:::{list-table}
43868e843eeSJed Brown:header-rows: 1
43968e843eeSJed Brown:widths: auto
44068e843eeSJed Brown* - User code
44168e843eeSJed Brown  - Example
44268e843eeSJed Brown* - `ceed`
44368e843eeSJed Brown  - * ex1 (volume)
44468e843eeSJed Brown* - `mfem`
44568e843eeSJed Brown  - * BP1 (scalar mass operator)
44668e843eeSJed Brown    * BP3 (scalar Laplace operator)
44768e843eeSJed Brown* - `petsc`
44868e843eeSJed Brown  - * BP1 (scalar mass operator)
44968e843eeSJed Brown* - `nek5000`
45068e843eeSJed Brown  - * BP1 (scalar mass operator)
45168e843eeSJed Brown:::
452bcb2dfaeSJed Brown
453bcb2dfaeSJed Brown(v0-2)=
454bcb2dfaeSJed Brown
455bcb2dfaeSJed Brown## v0.2 (Mar 30, 2018)
456bcb2dfaeSJed Brown
457bcb2dfaeSJed BrownlibCEED was made publicly available the first full CEED software distribution, release
458bcb2dfaeSJed BrownCEED 1.0. The distribution was made available using the Spack package manager to provide
459bcb2dfaeSJed Browna common, easy-to-use build environment, where the user can build the CEED distribution
460bcb2dfaeSJed Brownwith all dependencies. This release included a new Fortran interface for the library.
461bcb2dfaeSJed BrownThis release also contained major improvements in the OCCA backend (including a new
462bcb2dfaeSJed Brown`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to
463bcb2dfaeSJed Browncompute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an
464bcb2dfaeSJed Brown`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application
465bcb2dfaeSJed Brownof the Laplace operator) was also added to this release.
466bcb2dfaeSJed Brown
467bcb2dfaeSJed BrownBackends available in this release:
468bcb2dfaeSJed Brown
46968e843eeSJed Brown| CEED resource (`-ceed`) | Backend                         |
47068e843eeSJed Brown|-------------------------|---------------------------------|
47168e843eeSJed Brown| `/cpu/self`             | Serial reference implementation |
47268e843eeSJed Brown| `/cpu/occa`             | Serial OCCA kernels             |
47368e843eeSJed Brown| `/gpu/occa`             | CUDA OCCA kernels               |
47468e843eeSJed Brown| `/omp/occa`             | OpenMP OCCA kernels             |
47568e843eeSJed Brown| `/ocl/occa`             | OpenCL OCCA kernels             |
476bcb2dfaeSJed Brown
477bcb2dfaeSJed BrownExamples available in this release:
478bcb2dfaeSJed Brown
47968e843eeSJed Brown:::{list-table}
48068e843eeSJed Brown:header-rows: 1
48168e843eeSJed Brown:widths: auto
48268e843eeSJed Brown* - User code
48368e843eeSJed Brown  - Example
48468e843eeSJed Brown* - `ceed`
48568e843eeSJed Brown  - * ex1 (volume)
48668e843eeSJed Brown* - `mfem`
48768e843eeSJed Brown  - * BP1 (scalar mass operator)
48868e843eeSJed Brown    * BP3 (scalar Laplace operator)
48968e843eeSJed Brown* - `petsc`
49068e843eeSJed Brown  - * BP1 (scalar mass operator)
49168e843eeSJed Brown* - `nek5000`
49268e843eeSJed Brown  - * BP1 (scalar mass operator)
49368e843eeSJed Brown:::
494bcb2dfaeSJed Brown
495bcb2dfaeSJed Brown(v0-1)=
496bcb2dfaeSJed Brown
497bcb2dfaeSJed Brown## v0.1 (Jan 3, 2018)
498bcb2dfaeSJed Brown
499bcb2dfaeSJed BrownInitial low-level API of the CEED project. The low-level API provides a set of Finite
500bcb2dfaeSJed BrownElements kernels and components for writing new low-level kernels. Examples include:
501bcb2dfaeSJed Brownvector and sparse linear algebra, element matrix assembly over a batch of elements,
502bcb2dfaeSJed Brownpartial assembly and action for efficient high-order operators like mass, diffusion,
503bcb2dfaeSJed Brownadvection, etc. The main goal of the low-level API is to establish the basis for the
504bcb2dfaeSJed Brownhigh-level API. Also, identifying such low-level kernels and providing a reference
505bcb2dfaeSJed Brownimplementation for them serves as the basis for specialized backend implementations.
506bcb2dfaeSJed BrownThis release contained several backends: `/cpu/self`, and backends which rely upon the
507bcb2dfaeSJed Brown[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`,
508bcb2dfaeSJed Brown`/gpu/occa`, and `/omp/occa`.
509bcb2dfaeSJed BrownIt also included several examples, in the `examples` folder:
510bcb2dfaeSJed BrownA standalone code that shows the usage of libCEED (with no external
511bcb2dfaeSJed Browndependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1
512bcb2dfaeSJed Brown(with the application of the mass operator); and a `petsc` example to perform BP1
513bcb2dfaeSJed Brown(with the application of the mass operator).
514bcb2dfaeSJed Brown
515bcb2dfaeSJed BrownBackends available in this release:
516bcb2dfaeSJed Brown
51768e843eeSJed Brown| CEED resource (`-ceed`) | Backend                         |
51868e843eeSJed Brown|-------------------------|---------------------------------|
51968e843eeSJed Brown| `/cpu/self`             | Serial reference implementation |
52068e843eeSJed Brown| `/cpu/occa`             | Serial OCCA kernels             |
52168e843eeSJed Brown| `/gpu/occa`             | CUDA OCCA kernels               |
52268e843eeSJed Brown| `/omp/occa`             | OpenMP OCCA kernels             |
523bcb2dfaeSJed Brown
524bcb2dfaeSJed BrownExamples available in this release:
525bcb2dfaeSJed Brown
526bcb2dfaeSJed Brown| User code             | Example                           |
52768e843eeSJed Brown|-----------------------|-----------------------------------|
52868e843eeSJed Brown| `ceed`                | ex1 (scalar Laplace operator)     |
52968e843eeSJed Brown| `mfem`                | BP1 (scalar mass operator)        |
53068e843eeSJed Brown| `petsc`               | BP1 (scalar mass operator)        |
531bcb2dfaeSJed Brown```
532