xref: /libCEED/doc/sphinx/source/releasenotes.md (revision 3f21f6b10abeb5d85d3454ea5cd38498737dc88a)
1# Changes/Release Notes
2
3On this page we provide a summary of the main API changes, new features and examples
4for each release of libCEED.
5
6(main)=
7
8## Current `main` branch
9
10### Interface changes
11
12- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields.
13- 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`.
14- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added.
15- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients.
16- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces.
17- Warning added when compiling OCCA backend to alert users that this backend is experimental.
18- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`.
19- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name.
20
21### New features
22
23- `CeedScalar` can now be set as `float` or `double` at compile time.
24- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends.
25- 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.
26- Remove need to guard library headers in QFunction source for code generation backends.
27- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present.
28
29### Maintainability
30
31- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage.
32- `Include-what-you-use` makefile target added as `make iwyu`.
33
34(v0-9)=
35
36## v0.9 (Jul 6, 2021)
37
38### Interface changes
39
40- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact.
41
42### New features
43
44- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use.
45- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends.
46- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations.
47- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed).
48
49### Performance improvements
50
51### Examples
52
53- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations.
54- Fluid mechanics example adds GPU support and improves modularity.
55
56### Deprecated backends
57
58- 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.
59
60(v0-8)=
61
62## v0.8 (Mar 31, 2021)
63
64### Interface changes
65
66- Error handling improved to include enumerated error codes for C interface return values.
67- 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.
68
69### New features
70
71- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features.
72- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly.
73- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators.
74
75### Performance improvements
76
77- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`.
78- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`.
79
80### Examples
81
82- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions.
83- {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.
84- {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.
85- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree.
86
87(v0-7)=
88
89## v0.7 (Sep 29, 2020)
90
91### Interface changes
92
93- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors.
94  As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`.
95  These changes improve support for mixed finite element methods.
96- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature.
97- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity.
98- 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.
99- 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.
100- 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.
101- 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.
102  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`.
103- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory.
104- 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.
105
106### New features
107
108- New HIP backend: `/gpu/hip/ref`.
109- CeedQFunction support for user `CUfunction`s in some backends
110
111### Performance improvements
112
113- OCCA backend rebuilt to facilitate future performance enhancements.
114- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`.
115
116### Examples
117
118- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions.
119
120### Deprecated backends
121
122- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`.
123
124(v0-6)=
125
126## v0.6 (Mar 29, 2020)
127
128libCEED v0.6 contains numerous new features and examples, as well as expanded
129documentation in [this new website](https://libceed.readthedocs.io).
130
131### New features
132
133- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly
134  1-1 correspondence with the C interface, plus some convenience features.  For instance,
135  data stored in the {cpp:type}`CeedVector` structure are available without copy as
136  {py:class}`numpy.ndarray`.  Short tutorials are provided in
137  [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/).
138- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point,
139  {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal
140  ({c:func}`CeedOperatorAssembleLinearDiagonal`).  These operations are useful for
141  preconditioning ingredients and are used in the libCEED's multigrid examples.
142- The inverse of separable operators can be obtained using
143  {c:func}`CeedOperatorCreateFDMElementInverse` and applied with
144  {c:func}`CeedOperatorApply`.  This is a useful preconditioning ingredient,
145  especially for Laplacians and related operators.
146- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`,
147  {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`.
148- Make public accessors for various attributes to facilitate writing composable code.
149- New backend: `/cpu/self/memcheck/serial`.
150- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA
151  backends.  (Single source is coming soon for OCCA backends.)
152- Fix some missing edge cases in CUDA backend.
153
154### Performance Improvements
155
156- MAGMA backend performance optimization and non-tensor bases.
157- No-copy optimization in {c:func}`CeedOperatorApply`.
158
159### Interface changes
160
161- Replace {code}`CeedElemRestrictionCreateIdentity` and
162  {code}`CeedElemRestrictionCreateBlocked` with more flexible
163  {c:func}`CeedElemRestrictionCreateStrided` and
164  {c:func}`CeedElemRestrictionCreateBlockedStrided`.
165- Add arguments to {c:func}`CeedQFunctionCreateIdentity`.
166- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification
167  with {cpp:enum}`CeedInterlaceMode`.  This is now an attribute of the
168  {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no
169  longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and
170  {c:func}`CeedElemRestrictionApply`.
171
172### Examples
173
174libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`.
175Notable additions include:
176
177- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of
178  a domain in 1, 2, and 3 dimensions by applying a Laplacian.
179
180- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area
181  of domains (like the cube and sphere) by direct integration on a surface mesh;
182  demonstrates geometric dimension different from topological dimension.
183
184- PETSc {ref}`example-petsc-bps`:
185
186  - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support.
187  - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements
188    and transparent CUDA support.
189  - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`):
190    generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric
191    dimension different from topological dimension.
192
193- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid
194  solver with algebraic multigrid coarse solve.
195
196- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly
197  `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`),
198  implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and
199  quasi-2D computational domain support.
200
201- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for
202  linear elasticity, small-strain hyperelasticity, and globalized finite-strain
203  hyperelasticity using p-multigrid with algebraic multigrid coarse solve.
204
205(v0-5)=
206
207## v0.5 (Sep 18, 2019)
208
209For this release, several improvements were made. Two new CUDA backends were added to
210the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art
211performance using single-source {ref}`CeedQFunction`. From this release, users
212can define Q-Functions in a single source code independently of the targeted backend
213with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU
214compilation of the user provided {ref}`CeedQFunction` code. To allow a unified
215declaration, the {ref}`CeedQFunction` API has undergone a slight change:
216the `QFunctionField` parameter `ncomp` has been changed to `size`. This change
217requires setting the previous value of `ncomp` to `ncomp*dim` when adding a
218`QFunctionField` with eval mode `CEED EVAL GRAD`.
219
220Additionally, new CPU backends
221were included in this release, such as the `/cpu/self/opt/*` backends (which are
222written in pure C and use partial **E-vectors** to improve performance) and the
223`/cpu/self/ref/memcheck` backend (which relies upon the
224[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user
225{ref}`CeedQFunction` have no undefined values).
226This release also included various performance improvements, bug fixes, new examples,
227and improved tests. Among these improvements, vectorized instructions for
228{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD`
229instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and
230identity Q-Functions were introduced, and the PETSc benchmark problems were expanded
231to include unstructured meshes handling were. For this expansion, the prior version of
232the PETSc BPs, which only included data associated with structured geometries, were
233renamed `bpsraw`, and the new version of the BPs, which can handle data associated
234with any unstructured geometry, were called `bps`. Additionally, other benchmark
235problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively),
236and BP5 and BP6 (the collocated versions---for which the quadrature points are the same
237as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc
238examples. Furthermoew, another standalone libCEED example, called `ex2`, which
239computes the surface area of a given mesh was added to this release.
240
241Backends available in this release:
242
243| CEED resource (`-ceed`)  | Backend                                             |
244|--------------------------|-----------------------------------------------------|
245| `/cpu/self/ref/serial`   | Serial reference implementation                     |
246| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
247| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks            |
248| `/cpu/self/opt/serial`   | Serial optimized C implementation                   |
249| `/cpu/self/opt/blocked`  | Blocked optimized C implementation                  |
250| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
251| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
252| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
253| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
254| `/cpu/occa`              | Serial OCCA kernels                                 |
255| `/gpu/occa`              | CUDA OCCA kernels                                   |
256| `/omp/occa`              | OpenMP OCCA kernels                                 |
257| `/ocl/occa`              | OpenCL OCCA kernels                                 |
258| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
259| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
260| `/gpu/cuda/shared`       | Optimized pure CUDA kernels using shared memory     |
261| `/gpu/cuda/gen`          | Optimized pure CUDA kernels using code generation   |
262| `/gpu/magma`             | CUDA MAGMA kernels                                  |
263
264Examples available in this release:
265
266:::{list-table}
267:header-rows: 1
268:widths: auto
269* - User code
270  - Example
271* - `ceed`
272  - * ex1 (volume)
273    * ex2 (surface)
274* - `mfem`
275  - * BP1 (scalar mass operator)
276    * BP3 (scalar Laplace operator)
277* - `petsc`
278  - * BP1 (scalar mass operator)
279    * BP2 (vector mass operator)
280    * BP3 (scalar Laplace operator)
281    * BP4 (vector Laplace operator)
282    * BP5 (collocated scalar Laplace operator)
283    * BP6 (collocated vector Laplace operator)
284    * Navier-Stokes
285* - `nek5000`
286  - * BP1 (scalar mass operator)
287    * BP3 (scalar Laplace operator)
288:::
289
290(v0-4)=
291
292## v0.4 (Apr 1, 2019)
293
294libCEED v0.4 was made again publicly available in the second full CEED software
295distribution, release CEED 2.0. This release contained notable features, such as
296four new CPU backends, two new GPU backends, CPU backend optimizations, initial
297support for operator composition, performance benchmarking, and a Navier-Stokes demo.
298The new CPU backends in this release came in two families. The `/cpu/self/*/serial`
299backends process one element at a time and are intended for meshes with a smaller number
300of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of
301eight interlaced elements and are intended for meshes with higher numbers of elements.
302The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU
303performance. The `/cpu/self/xsmm/*` backends rely upon the
304[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU
305performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA.
306The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable
307performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple
308parallelization approach, where each thread treats a finite element. Using just in time
309compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this
310backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend
311achieve good peak performance for 1D, 2D, and low order 3D problems, but performance
312deteriorates very quickly when threads run out of registers.
313
314A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED
315examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`).
316This example solves the time-dependent Navier-Stokes equations of compressible gas
317dynamics in a static Eulerian three-dimensional frame, using structured high-order
318finite/spectral element spatial discretizations and explicit high-order time-stepping
319(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc,
320so that the pointwise physics (defined at quadrature points) is separated from the
321parallelization and meshing concerns.
322
323Backends available in this release:
324
325| CEED resource (`-ceed`)  | Backend                                             |
326|--------------------------|-----------------------------------------------------|
327| `/cpu/self/ref/serial`   | Serial reference implementation                     |
328| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
329| `/cpu/self/tmpl`         | Backend template, defaults to `/cpu/self/blocked`   |
330| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
331| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
332| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
333| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
334| `/cpu/occa`              | Serial OCCA kernels                                 |
335| `/gpu/occa`              | CUDA OCCA kernels                                   |
336| `/omp/occa`              | OpenMP OCCA kernels                                 |
337| `/ocl/occa`              | OpenCL OCCA kernels                                 |
338| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
339| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
340| `/gpu/magma`             | CUDA MAGMA kernels                                  |
341
342Examples available in this release:
343
344:::{list-table}
345:header-rows: 1
346:widths: auto
347* - User code
348  - Example
349* - `ceed`
350  - * ex1 (volume)
351* - `mfem`
352  - * BP1 (scalar mass operator)
353    * BP3 (scalar Laplace operator)
354* - `petsc`
355  - * BP1 (scalar mass operator)
356    * BP3 (scalar Laplace operator)
357    * Navier-Stokes
358* - `nek5000`
359  - * BP1 (scalar mass operator)
360    * BP3 (scalar Laplace operator)
361:::
362
363(v0-3)=
364
365## v0.3 (Sep 30, 2018)
366
367Notable features in this release include active/passive field interface, support for
368non-tensor bases, backend optimization, and improved Fortran interface. This release
369also focused on providing improved continuous integration, and many new tests with code
370coverage reports of about 90%. This release also provided a significant change to the
371public interface: a {ref}`CeedQFunction` can take any number of named input and output
372arguments while {ref}`CeedOperator` connects them to the actual data, which may be
373supplied explicitly to `CeedOperatorApply()` (active) or separately via
374`CeedOperatorSetField()` (passive). This interface change enables reusable libraries
375of CeedQFunctions and composition of block solvers constructed using
376{ref}`CeedOperator`. A concept of blocked restriction was added to this release and
377used in an optimized CPU backend. Although this is typically not visible to the user,
378it enables effective use of arbitrary-length SIMD while maintaining cache locality.
379This CPU backend also implements an algebraic factorization of tensor product gradients
380to perform fewer operations than standard application of interpolation and
381differentiation from nodes to quadrature points. This algebraic formulation
382automatically supports non-polynomial and non-interpolatory bases, thus is more general
383than the more common derivation in terms of Lagrange polynomials on the quadrature points.
384
385Backends available in this release:
386
387| CEED resource (`-ceed`) | Backend                                             |
388|-------------------------|-----------------------------------------------------|
389| `/cpu/self/blocked`     | Blocked reference implementation                    |
390| `/cpu/self/ref`         | Serial reference implementation                     |
391| `/cpu/self/tmpl`        | Backend template, defaults to `/cpu/self/blocked`   |
392| `/cpu/occa`             | Serial OCCA kernels                                 |
393| `/gpu/occa`             | CUDA OCCA kernels                                   |
394| `/omp/occa`             | OpenMP OCCA kernels                                 |
395| `/ocl/occa`             | OpenCL OCCA kernels                                 |
396| `/gpu/magma`            | CUDA MAGMA kernels                                  |
397
398Examples available in this release:
399
400:::{list-table}
401:header-rows: 1
402:widths: auto
403* - User code
404  - Example
405* - `ceed`
406  - * ex1 (volume)
407* - `mfem`
408  - * BP1 (scalar mass operator)
409    * BP3 (scalar Laplace operator)
410* - `petsc`
411  - * BP1 (scalar mass operator)
412    * BP3 (scalar Laplace operator)
413* - `nek5000`
414  - * BP1 (scalar mass operator)
415    * BP3 (scalar Laplace operator)
416:::
417
418(v0-21)=
419
420## v0.21 (Sep 30, 2018)
421
422A MAGMA backend (which relies upon the
423[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this
424release. This initial integration set up the framework of using MAGMA and provided the
425libCEED functionality through MAGMA kernels as one of libCEED’s computational backends.
426As any other backend, the MAGMA backend provides extended basic data structures for
427{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements
428the fundamental CEED building blocks to work with the new data structures.
429In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data
430but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled
431internally, and thus seamlessly to the user, through the functions/methods that are
432provided to support them.
433
434Backends available in this release:
435
436| CEED resource (`-ceed`) | Backend                         |
437|-------------------------|---------------------------------|
438| `/cpu/self`             | Serial reference implementation |
439| `/cpu/occa`             | Serial OCCA kernels             |
440| `/gpu/occa`             | CUDA OCCA kernels               |
441| `/omp/occa`             | OpenMP OCCA kernels             |
442| `/ocl/occa`             | OpenCL OCCA kernels             |
443| `/gpu/magma`            | CUDA MAGMA kernels              |
444
445Examples available in this release:
446
447:::{list-table}
448:header-rows: 1
449:widths: auto
450* - User code
451  - Example
452* - `ceed`
453  - * ex1 (volume)
454* - `mfem`
455  - * BP1 (scalar mass operator)
456    * BP3 (scalar Laplace operator)
457* - `petsc`
458  - * BP1 (scalar mass operator)
459* - `nek5000`
460  - * BP1 (scalar mass operator)
461:::
462
463(v0-2)=
464
465## v0.2 (Mar 30, 2018)
466
467libCEED was made publicly available the first full CEED software distribution, release
468CEED 1.0. The distribution was made available using the Spack package manager to provide
469a common, easy-to-use build environment, where the user can build the CEED distribution
470with all dependencies. This release included a new Fortran interface for the library.
471This release also contained major improvements in the OCCA backend (including a new
472`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to
473compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an
474`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application
475of the Laplace operator) was also added to this release.
476
477Backends available in this release:
478
479| CEED resource (`-ceed`) | Backend                         |
480|-------------------------|---------------------------------|
481| `/cpu/self`             | Serial reference implementation |
482| `/cpu/occa`             | Serial OCCA kernels             |
483| `/gpu/occa`             | CUDA OCCA kernels               |
484| `/omp/occa`             | OpenMP OCCA kernels             |
485| `/ocl/occa`             | OpenCL OCCA kernels             |
486
487Examples available in this release:
488
489:::{list-table}
490:header-rows: 1
491:widths: auto
492* - User code
493  - Example
494* - `ceed`
495  - * ex1 (volume)
496* - `mfem`
497  - * BP1 (scalar mass operator)
498    * BP3 (scalar Laplace operator)
499* - `petsc`
500  - * BP1 (scalar mass operator)
501* - `nek5000`
502  - * BP1 (scalar mass operator)
503:::
504
505(v0-1)=
506
507## v0.1 (Jan 3, 2018)
508
509Initial low-level API of the CEED project. The low-level API provides a set of Finite
510Elements kernels and components for writing new low-level kernels. Examples include:
511vector and sparse linear algebra, element matrix assembly over a batch of elements,
512partial assembly and action for efficient high-order operators like mass, diffusion,
513advection, etc. The main goal of the low-level API is to establish the basis for the
514high-level API. Also, identifying such low-level kernels and providing a reference
515implementation for them serves as the basis for specialized backend implementations.
516This release contained several backends: `/cpu/self`, and backends which rely upon the
517[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`,
518`/gpu/occa`, and `/omp/occa`.
519It also included several examples, in the `examples` folder:
520A standalone code that shows the usage of libCEED (with no external
521dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1
522(with the application of the mass operator); and a `petsc` example to perform BP1
523(with the application of the mass operator).
524
525Backends available in this release:
526
527| CEED resource (`-ceed`) | Backend                         |
528|-------------------------|---------------------------------|
529| `/cpu/self`             | Serial reference implementation |
530| `/cpu/occa`             | Serial OCCA kernels             |
531| `/gpu/occa`             | CUDA OCCA kernels               |
532| `/omp/occa`             | OpenMP OCCA kernels             |
533
534Examples available in this release:
535
536| User code             | Example                           |
537|-----------------------|-----------------------------------|
538| `ceed`                | ex1 (scalar Laplace operator)     |
539| `mfem`                | BP1 (scalar mass operator)        |
540| `petsc`               | BP1 (scalar mass operator)        |
541```
542