xref: /libCEED/doc/sphinx/source/releasenotes.md (revision 30601ac0afe2b18e1eb1cf6827bcb5d4bd528d77)
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` statements for additional local files.
26- Remove need to guard library headers in QFunction source for code generation backends.
27
28### Maintainability
29
30- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage.
31- `Include-what-you-use` makefile target added as `make iwyu`.
32
33(v0-9)=
34
35## v0.9 (Jul 6, 2021)
36
37### Interface changes
38
39- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact.
40
41### New features
42
43- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use.
44- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends.
45- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations.
46- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed).
47
48### Performance improvements
49
50### Examples
51
52- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations.
53- Fluid mechanics example adds GPU support and improves modularity.
54
55### Deprecated backends
56
57- 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.
58
59(v0-8)=
60
61## v0.8 (Mar 31, 2021)
62
63### Interface changes
64
65- Error handling improved to include enumerated error codes for C interface return values.
66- 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.
67
68### New features
69
70- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features.
71- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly.
72- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators.
73
74### Performance improvements
75
76- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`.
77- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`.
78
79### Examples
80
81- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions.
82- {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.
83- {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.
84- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree.
85
86(v0-7)=
87
88## v0.7 (Sep 29, 2020)
89
90### Interface changes
91
92- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors.
93  As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`.
94  These changes improve support for mixed finite element methods.
95- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature.
96- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity.
97- 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.
98- 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.
99- 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.
100- 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.
101  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`.
102- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory.
103- 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.
104
105### New features
106
107- New HIP backend: `/gpu/hip/ref`.
108- CeedQFunction support for user `CUfunction`s in some backends
109
110### Performance improvements
111
112- OCCA backend rebuilt to facilitate future performance enhancements.
113- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`.
114
115### Examples
116
117- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions.
118
119### Deprecated backends
120
121- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`.
122
123(v0-6)=
124
125## v0.6 (Mar 29, 2020)
126
127libCEED v0.6 contains numerous new features and examples, as well as expanded
128documentation in [this new website](https://libceed.readthedocs.io).
129
130### New features
131
132- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly
133  1-1 correspondence with the C interface, plus some convenience features.  For instance,
134  data stored in the {cpp:type}`CeedVector` structure are available without copy as
135  {py:class}`numpy.ndarray`.  Short tutorials are provided in
136  [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/).
137- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point,
138  {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal
139  ({c:func}`CeedOperatorAssembleLinearDiagonal`).  These operations are useful for
140  preconditioning ingredients and are used in the libCEED's multigrid examples.
141- The inverse of separable operators can be obtained using
142  {c:func}`CeedOperatorCreateFDMElementInverse` and applied with
143  {c:func}`CeedOperatorApply`.  This is a useful preconditioning ingredient,
144  especially for Laplacians and related operators.
145- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`,
146  {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`.
147- Make public accessors for various attributes to facilitate writing composable code.
148- New backend: `/cpu/self/memcheck/serial`.
149- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA
150  backends.  (Single source is coming soon for OCCA backends.)
151- Fix some missing edge cases in CUDA backend.
152
153### Performance Improvements
154
155- MAGMA backend performance optimization and non-tensor bases.
156- No-copy optimization in {c:func}`CeedOperatorApply`.
157
158### Interface changes
159
160- Replace {code}`CeedElemRestrictionCreateIdentity` and
161  {code}`CeedElemRestrictionCreateBlocked` with more flexible
162  {c:func}`CeedElemRestrictionCreateStrided` and
163  {c:func}`CeedElemRestrictionCreateBlockedStrided`.
164- Add arguments to {c:func}`CeedQFunctionCreateIdentity`.
165- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification
166  with {cpp:enum}`CeedInterlaceMode`.  This is now an attribute of the
167  {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no
168  longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and
169  {c:func}`CeedElemRestrictionApply`.
170
171### Examples
172
173libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`.
174Notable additions include:
175
176- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of
177  a domain in 1, 2, and 3 dimensions by applying a Laplacian.
178
179- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area
180  of domains (like the cube and sphere) by direct integration on a surface mesh;
181  demonstrates geometric dimension different from topological dimension.
182
183- PETSc {ref}`example-petsc-bps`:
184
185  - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support.
186  - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements
187    and transparent CUDA support.
188  - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`):
189    generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric
190    dimension different from topological dimension.
191
192- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid
193  solver with algebraic multigrid coarse solve.
194
195- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly
196  `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`),
197  implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and
198  quasi-2D computational domain support.
199
200- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for
201  linear elasticity, small-strain hyperelasticity, and globalized finite-strain
202  hyperelasticity using p-multigrid with algebraic multigrid coarse solve.
203
204(v0-5)=
205
206## v0.5 (Sep 18, 2019)
207
208For this release, several improvements were made. Two new CUDA backends were added to
209the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art
210performance using single-source {ref}`CeedQFunction`. From this release, users
211can define Q-Functions in a single source code independently of the targeted backend
212with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU
213compilation of the user provided {ref}`CeedQFunction` code. To allow a unified
214declaration, the {ref}`CeedQFunction` API has undergone a slight change:
215the `QFunctionField` parameter `ncomp` has been changed to `size`. This change
216requires setting the previous value of `ncomp` to `ncomp*dim` when adding a
217`QFunctionField` with eval mode `CEED EVAL GRAD`.
218
219Additionally, new CPU backends
220were included in this release, such as the `/cpu/self/opt/*` backends (which are
221written in pure C and use partial **E-vectors** to improve performance) and the
222`/cpu/self/ref/memcheck` backend (which relies upon the
223[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user
224{ref}`CeedQFunction` have no undefined values).
225This release also included various performance improvements, bug fixes, new examples,
226and improved tests. Among these improvements, vectorized instructions for
227{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD`
228instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and
229identity Q-Functions were introduced, and the PETSc benchmark problems were expanded
230to include unstructured meshes handling were. For this expansion, the prior version of
231the PETSc BPs, which only included data associated with structured geometries, were
232renamed `bpsraw`, and the new version of the BPs, which can handle data associated
233with any unstructured geometry, were called `bps`. Additionally, other benchmark
234problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively),
235and BP5 and BP6 (the collocated versions---for which the quadrature points are the same
236as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc
237examples. Furthermoew, another standalone libCEED example, called `ex2`, which
238computes the surface area of a given mesh was added to this release.
239
240Backends available in this release:
241
242| CEED resource (`-ceed`)  | Backend                                             |
243|--------------------------|-----------------------------------------------------|
244| `/cpu/self/ref/serial`   | Serial reference implementation                     |
245| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
246| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks            |
247| `/cpu/self/opt/serial`   | Serial optimized C implementation                   |
248| `/cpu/self/opt/blocked`  | Blocked optimized C implementation                  |
249| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
250| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
251| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
252| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
253| `/cpu/occa`              | Serial OCCA kernels                                 |
254| `/gpu/occa`              | CUDA OCCA kernels                                   |
255| `/omp/occa`              | OpenMP OCCA kernels                                 |
256| `/ocl/occa`              | OpenCL OCCA kernels                                 |
257| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
258| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
259| `/gpu/cuda/shared`       | Optimized pure CUDA kernels using shared memory     |
260| `/gpu/cuda/gen`          | Optimized pure CUDA kernels using code generation   |
261| `/gpu/magma`             | CUDA MAGMA kernels                                  |
262
263Examples available in this release:
264
265:::{list-table}
266:header-rows: 1
267:widths: auto
268* - User code
269  - Example
270* - `ceed`
271  - * ex1 (volume)
272    * ex2 (surface)
273* - `mfem`
274  - * BP1 (scalar mass operator)
275    * BP3 (scalar Laplace operator)
276* - `petsc`
277  - * BP1 (scalar mass operator)
278    * BP2 (vector mass operator)
279    * BP3 (scalar Laplace operator)
280    * BP4 (vector Laplace operator)
281    * BP5 (collocated scalar Laplace operator)
282    * BP6 (collocated vector Laplace operator)
283    * Navier-Stokes
284* - `nek5000`
285  - * BP1 (scalar mass operator)
286    * BP3 (scalar Laplace operator)
287:::
288
289(v0-4)=
290
291## v0.4 (Apr 1, 2019)
292
293libCEED v0.4 was made again publicly available in the second full CEED software
294distribution, release CEED 2.0. This release contained notable features, such as
295four new CPU backends, two new GPU backends, CPU backend optimizations, initial
296support for operator composition, performance benchmarking, and a Navier-Stokes demo.
297The new CPU backends in this release came in two families. The `/cpu/self/*/serial`
298backends process one element at a time and are intended for meshes with a smaller number
299of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of
300eight interlaced elements and are intended for meshes with higher numbers of elements.
301The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU
302performance. The `/cpu/self/xsmm/*` backends rely upon the
303[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU
304performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA.
305The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable
306performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple
307parallelization approach, where each thread treats a finite element. Using just in time
308compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this
309backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend
310achieve good peak performance for 1D, 2D, and low order 3D problems, but performance
311deteriorates very quickly when threads run out of registers.
312
313A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED
314examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`).
315This example solves the time-dependent Navier-Stokes equations of compressible gas
316dynamics in a static Eulerian three-dimensional frame, using structured high-order
317finite/spectral element spatial discretizations and explicit high-order time-stepping
318(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc,
319so that the pointwise physics (defined at quadrature points) is separated from the
320parallelization and meshing concerns.
321
322Backends available in this release:
323
324| CEED resource (`-ceed`)  | Backend                                             |
325|--------------------------|-----------------------------------------------------|
326| `/cpu/self/ref/serial`   | Serial reference implementation                     |
327| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
328| `/cpu/self/tmpl`         | Backend template, defaults to `/cpu/self/blocked`   |
329| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
330| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
331| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
332| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
333| `/cpu/occa`              | Serial OCCA kernels                                 |
334| `/gpu/occa`              | CUDA OCCA kernels                                   |
335| `/omp/occa`              | OpenMP OCCA kernels                                 |
336| `/ocl/occa`              | OpenCL OCCA kernels                                 |
337| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
338| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
339| `/gpu/magma`             | CUDA MAGMA kernels                                  |
340
341Examples available in this release:
342
343:::{list-table}
344:header-rows: 1
345:widths: auto
346* - User code
347  - Example
348* - `ceed`
349  - * ex1 (volume)
350* - `mfem`
351  - * BP1 (scalar mass operator)
352    * BP3 (scalar Laplace operator)
353* - `petsc`
354  - * BP1 (scalar mass operator)
355    * BP3 (scalar Laplace operator)
356    * Navier-Stokes
357* - `nek5000`
358  - * BP1 (scalar mass operator)
359    * BP3 (scalar Laplace operator)
360:::
361
362(v0-3)=
363
364## v0.3 (Sep 30, 2018)
365
366Notable features in this release include active/passive field interface, support for
367non-tensor bases, backend optimization, and improved Fortran interface. This release
368also focused on providing improved continuous integration, and many new tests with code
369coverage reports of about 90%. This release also provided a significant change to the
370public interface: a {ref}`CeedQFunction` can take any number of named input and output
371arguments while {ref}`CeedOperator` connects them to the actual data, which may be
372supplied explicitly to `CeedOperatorApply()` (active) or separately via
373`CeedOperatorSetField()` (passive). This interface change enables reusable libraries
374of CeedQFunctions and composition of block solvers constructed using
375{ref}`CeedOperator`. A concept of blocked restriction was added to this release and
376used in an optimized CPU backend. Although this is typically not visible to the user,
377it enables effective use of arbitrary-length SIMD while maintaining cache locality.
378This CPU backend also implements an algebraic factorization of tensor product gradients
379to perform fewer operations than standard application of interpolation and
380differentiation from nodes to quadrature points. This algebraic formulation
381automatically supports non-polynomial and non-interpolatory bases, thus is more general
382than the more common derivation in terms of Lagrange polynomials on the quadrature points.
383
384Backends available in this release:
385
386| CEED resource (`-ceed`) | Backend                                             |
387|-------------------------|-----------------------------------------------------|
388| `/cpu/self/blocked`     | Blocked reference implementation                    |
389| `/cpu/self/ref`         | Serial reference implementation                     |
390| `/cpu/self/tmpl`        | Backend template, defaults to `/cpu/self/blocked`   |
391| `/cpu/occa`             | Serial OCCA kernels                                 |
392| `/gpu/occa`             | CUDA OCCA kernels                                   |
393| `/omp/occa`             | OpenMP OCCA kernels                                 |
394| `/ocl/occa`             | OpenCL OCCA kernels                                 |
395| `/gpu/magma`            | CUDA MAGMA kernels                                  |
396
397Examples available in this release:
398
399:::{list-table}
400:header-rows: 1
401:widths: auto
402* - User code
403  - Example
404* - `ceed`
405  - * ex1 (volume)
406* - `mfem`
407  - * BP1 (scalar mass operator)
408    * BP3 (scalar Laplace operator)
409* - `petsc`
410  - * BP1 (scalar mass operator)
411    * BP3 (scalar Laplace operator)
412* - `nek5000`
413  - * BP1 (scalar mass operator)
414    * BP3 (scalar Laplace operator)
415:::
416
417(v0-21)=
418
419## v0.21 (Sep 30, 2018)
420
421A MAGMA backend (which relies upon the
422[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this
423release. This initial integration set up the framework of using MAGMA and provided the
424libCEED functionality through MAGMA kernels as one of libCEED’s computational backends.
425As any other backend, the MAGMA backend provides extended basic data structures for
426{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements
427the fundamental CEED building blocks to work with the new data structures.
428In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data
429but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled
430internally, and thus seamlessly to the user, through the functions/methods that are
431provided to support them.
432
433Backends available in this release:
434
435| CEED resource (`-ceed`) | Backend                         |
436|-------------------------|---------------------------------|
437| `/cpu/self`             | Serial reference implementation |
438| `/cpu/occa`             | Serial OCCA kernels             |
439| `/gpu/occa`             | CUDA OCCA kernels               |
440| `/omp/occa`             | OpenMP OCCA kernels             |
441| `/ocl/occa`             | OpenCL OCCA kernels             |
442| `/gpu/magma`            | CUDA MAGMA kernels              |
443
444Examples available in this release:
445
446:::{list-table}
447:header-rows: 1
448:widths: auto
449* - User code
450  - Example
451* - `ceed`
452  - * ex1 (volume)
453* - `mfem`
454  - * BP1 (scalar mass operator)
455    * BP3 (scalar Laplace operator)
456* - `petsc`
457  - * BP1 (scalar mass operator)
458* - `nek5000`
459  - * BP1 (scalar mass operator)
460:::
461
462(v0-2)=
463
464## v0.2 (Mar 30, 2018)
465
466libCEED was made publicly available the first full CEED software distribution, release
467CEED 1.0. The distribution was made available using the Spack package manager to provide
468a common, easy-to-use build environment, where the user can build the CEED distribution
469with all dependencies. This release included a new Fortran interface for the library.
470This release also contained major improvements in the OCCA backend (including a new
471`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to
472compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an
473`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application
474of the Laplace operator) was also added to this release.
475
476Backends available in this release:
477
478| CEED resource (`-ceed`) | Backend                         |
479|-------------------------|---------------------------------|
480| `/cpu/self`             | Serial reference implementation |
481| `/cpu/occa`             | Serial OCCA kernels             |
482| `/gpu/occa`             | CUDA OCCA kernels               |
483| `/omp/occa`             | OpenMP OCCA kernels             |
484| `/ocl/occa`             | OpenCL OCCA kernels             |
485
486Examples available in this release:
487
488:::{list-table}
489:header-rows: 1
490:widths: auto
491* - User code
492  - Example
493* - `ceed`
494  - * ex1 (volume)
495* - `mfem`
496  - * BP1 (scalar mass operator)
497    * BP3 (scalar Laplace operator)
498* - `petsc`
499  - * BP1 (scalar mass operator)
500* - `nek5000`
501  - * BP1 (scalar mass operator)
502:::
503
504(v0-1)=
505
506## v0.1 (Jan 3, 2018)
507
508Initial low-level API of the CEED project. The low-level API provides a set of Finite
509Elements kernels and components for writing new low-level kernels. Examples include:
510vector and sparse linear algebra, element matrix assembly over a batch of elements,
511partial assembly and action for efficient high-order operators like mass, diffusion,
512advection, etc. The main goal of the low-level API is to establish the basis for the
513high-level API. Also, identifying such low-level kernels and providing a reference
514implementation for them serves as the basis for specialized backend implementations.
515This release contained several backends: `/cpu/self`, and backends which rely upon the
516[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`,
517`/gpu/occa`, and `/omp/occa`.
518It also included several examples, in the `examples` folder:
519A standalone code that shows the usage of libCEED (with no external
520dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1
521(with the application of the mass operator); and a `petsc` example to perform BP1
522(with the application of the mass operator).
523
524Backends available in this release:
525
526| CEED resource (`-ceed`) | Backend                         |
527|-------------------------|---------------------------------|
528| `/cpu/self`             | Serial reference implementation |
529| `/cpu/occa`             | Serial OCCA kernels             |
530| `/gpu/occa`             | CUDA OCCA kernels               |
531| `/omp/occa`             | OpenMP OCCA kernels             |
532
533Examples available in this release:
534
535| User code             | Example                           |
536|-----------------------|-----------------------------------|
537| `ceed`                | ex1 (scalar Laplace operator)     |
538| `mfem`                | BP1 (scalar mass operator)        |
539| `petsc`               | BP1 (scalar mass operator)        |
540```
541