xref: /libCEED/doc/sphinx/source/releasenotes.md (revision cdf95791513f7c35170bef3ba2e19f272fe04533)
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- Clarify documentation for {c:func}`CeedVectorTakeArray`; this function will error if {c:func}`CeedVectorSetArray` with `copy_mode == CEED_USE_POINTER` was not previously called for the corresponding `CeedMemType`.
21- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitalized arrays; require initalized data for {c:func}`CeedVectorGetArray`.
22- Added {c:func}`CeedQFunctionContextRegisterDouble` and {c:func}`CeedQFunctionContextRegisterInt32` with {c:func}`CeedQFunctionContextSetDouble` and {c:func}`CeedQFunctionContextSetInt32` to facilitate easy updating of {c:struct}`CeedQFunctionContext` data by user defined field names.
23- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retreive user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`.
24- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums.
25
26### New features
27
28- `CeedScalar` can now be set as `float` or `double` at compile time.
29- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends.
30- 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.
31- Remove need to guard library headers in QFunction source for code generation backends.
32- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present.
33- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally.
34- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file.
35- Added support for non-tensor H(div) elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor.
36
37### Maintainability
38
39- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage.
40- `Include-what-you-use` makefile target added as `make iwyu`.
41- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase.
42- Put GPU JiTed kernel source code into separate files.
43- Dropped legacy version support in PETSc based examples to better utilize PETSc DMPlex and Mat updates to support libCEED; current minimum PETSc version for the examples is v3.17.
44
45(v0-9)=
46
47## v0.9 (Jul 6, 2021)
48
49### Interface changes
50
51- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact.
52
53### New features
54
55- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use.
56- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends.
57- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations.
58- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed).
59
60### Performance improvements
61
62### Examples
63
64- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations.
65- Fluid mechanics example adds GPU support and improves modularity.
66
67### Deprecated backends
68
69- 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.
70
71(v0-8)=
72
73## v0.8 (Mar 31, 2021)
74
75### Interface changes
76
77- Error handling improved to include enumerated error codes for C interface return values.
78- 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.
79
80### New features
81
82- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features.
83- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly.
84- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators.
85
86### Performance improvements
87
88- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`.
89- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`.
90
91### Examples
92
93- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions.
94- {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.
95- {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.
96- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree.
97
98(v0-7)=
99
100## v0.7 (Sep 29, 2020)
101
102### Interface changes
103
104- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors.
105  As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`.
106  These changes improve support for mixed finite element methods.
107- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature.
108- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity.
109- 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.
110- 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.
111- 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.
112- 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.
113  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`.
114- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory.
115- 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.
116
117### New features
118
119- New HIP backend: `/gpu/hip/ref`.
120- CeedQFunction support for user `CUfunction`s in some backends
121
122### Performance improvements
123
124- OCCA backend rebuilt to facilitate future performance enhancements.
125- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`.
126
127### Examples
128
129- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions.
130
131### Deprecated backends
132
133- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`.
134
135(v0-6)=
136
137## v0.6 (Mar 29, 2020)
138
139libCEED v0.6 contains numerous new features and examples, as well as expanded
140documentation in [this new website](https://libceed.readthedocs.io).
141
142### New features
143
144- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly
145  1-1 correspondence with the C interface, plus some convenience features.  For instance,
146  data stored in the {cpp:type}`CeedVector` structure are available without copy as
147  {py:class}`numpy.ndarray`.  Short tutorials are provided in
148  [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/).
149- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point,
150  {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal
151  ({c:func}`CeedOperatorAssembleLinearDiagonal`).  These operations are useful for
152  preconditioning ingredients and are used in the libCEED's multigrid examples.
153- The inverse of separable operators can be obtained using
154  {c:func}`CeedOperatorCreateFDMElementInverse` and applied with
155  {c:func}`CeedOperatorApply`.  This is a useful preconditioning ingredient,
156  especially for Laplacians and related operators.
157- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`,
158  {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`.
159- Make public accessors for various attributes to facilitate writing composable code.
160- New backend: `/cpu/self/memcheck/serial`.
161- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA
162  backends.  (Single source is coming soon for OCCA backends.)
163- Fix some missing edge cases in CUDA backend.
164
165### Performance Improvements
166
167- MAGMA backend performance optimization and non-tensor bases.
168- No-copy optimization in {c:func}`CeedOperatorApply`.
169
170### Interface changes
171
172- Replace {code}`CeedElemRestrictionCreateIdentity` and
173  {code}`CeedElemRestrictionCreateBlocked` with more flexible
174  {c:func}`CeedElemRestrictionCreateStrided` and
175  {c:func}`CeedElemRestrictionCreateBlockedStrided`.
176- Add arguments to {c:func}`CeedQFunctionCreateIdentity`.
177- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification
178  with {cpp:enum}`CeedInterlaceMode`.  This is now an attribute of the
179  {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no
180  longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and
181  {c:func}`CeedElemRestrictionApply`.
182
183### Examples
184
185libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`.
186Notable additions include:
187
188- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of
189  a domain in 1, 2, and 3 dimensions by applying a Laplacian.
190
191- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area
192  of domains (like the cube and sphere) by direct integration on a surface mesh;
193  demonstrates geometric dimension different from topological dimension.
194
195- PETSc {ref}`example-petsc-bps`:
196
197  - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support.
198  - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements
199    and transparent CUDA support.
200  - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`):
201    generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric
202    dimension different from topological dimension.
203
204- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid
205  solver with algebraic multigrid coarse solve.
206
207- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly
208  `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`),
209  implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and
210  quasi-2D computational domain support.
211
212- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for
213  linear elasticity, small-strain hyperelasticity, and globalized finite-strain
214  hyperelasticity using p-multigrid with algebraic multigrid coarse solve.
215
216(v0-5)=
217
218## v0.5 (Sep 18, 2019)
219
220For this release, several improvements were made. Two new CUDA backends were added to
221the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art
222performance using single-source {ref}`CeedQFunction`. From this release, users
223can define Q-Functions in a single source code independently of the targeted backend
224with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU
225compilation of the user provided {ref}`CeedQFunction` code. To allow a unified
226declaration, the {ref}`CeedQFunction` API has undergone a slight change:
227the `QFunctionField` parameter `ncomp` has been changed to `size`. This change
228requires setting the previous value of `ncomp` to `ncomp*dim` when adding a
229`QFunctionField` with eval mode `CEED EVAL GRAD`.
230
231Additionally, new CPU backends
232were included in this release, such as the `/cpu/self/opt/*` backends (which are
233written in pure C and use partial **E-vectors** to improve performance) and the
234`/cpu/self/ref/memcheck` backend (which relies upon the
235[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user
236{ref}`CeedQFunction` have no undefined values).
237This release also included various performance improvements, bug fixes, new examples,
238and improved tests. Among these improvements, vectorized instructions for
239{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD`
240instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and
241identity Q-Functions were introduced, and the PETSc benchmark problems were expanded
242to include unstructured meshes handling were. For this expansion, the prior version of
243the PETSc BPs, which only included data associated with structured geometries, were
244renamed `bpsraw`, and the new version of the BPs, which can handle data associated
245with any unstructured geometry, were called `bps`. Additionally, other benchmark
246problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively),
247and BP5 and BP6 (the collocated versions---for which the quadrature points are the same
248as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc
249examples. Furthermoew, another standalone libCEED example, called `ex2`, which
250computes the surface area of a given mesh was added to this release.
251
252Backends available in this release:
253
254| CEED resource (`-ceed`)  | Backend                                             |
255|--------------------------|-----------------------------------------------------|
256| `/cpu/self/ref/serial`   | Serial reference implementation                     |
257| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
258| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks            |
259| `/cpu/self/opt/serial`   | Serial optimized C implementation                   |
260| `/cpu/self/opt/blocked`  | Blocked optimized C implementation                  |
261| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
262| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
263| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
264| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
265| `/cpu/occa`              | Serial OCCA kernels                                 |
266| `/gpu/occa`              | CUDA OCCA kernels                                   |
267| `/omp/occa`              | OpenMP OCCA kernels                                 |
268| `/ocl/occa`              | OpenCL OCCA kernels                                 |
269| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
270| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
271| `/gpu/cuda/shared`       | Optimized pure CUDA kernels using shared memory     |
272| `/gpu/cuda/gen`          | Optimized pure CUDA kernels using code generation   |
273| `/gpu/magma`             | CUDA MAGMA kernels                                  |
274
275Examples available in this release:
276
277:::{list-table}
278:header-rows: 1
279:widths: auto
280* - User code
281  - Example
282* - `ceed`
283  - * ex1 (volume)
284    * ex2 (surface)
285* - `mfem`
286  - * BP1 (scalar mass operator)
287    * BP3 (scalar Laplace operator)
288* - `petsc`
289  - * BP1 (scalar mass operator)
290    * BP2 (vector mass operator)
291    * BP3 (scalar Laplace operator)
292    * BP4 (vector Laplace operator)
293    * BP5 (collocated scalar Laplace operator)
294    * BP6 (collocated vector Laplace operator)
295    * Navier-Stokes
296* - `nek5000`
297  - * BP1 (scalar mass operator)
298    * BP3 (scalar Laplace operator)
299:::
300
301(v0-4)=
302
303## v0.4 (Apr 1, 2019)
304
305libCEED v0.4 was made again publicly available in the second full CEED software
306distribution, release CEED 2.0. This release contained notable features, such as
307four new CPU backends, two new GPU backends, CPU backend optimizations, initial
308support for operator composition, performance benchmarking, and a Navier-Stokes demo.
309The new CPU backends in this release came in two families. The `/cpu/self/*/serial`
310backends process one element at a time and are intended for meshes with a smaller number
311of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of
312eight interlaced elements and are intended for meshes with higher numbers of elements.
313The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU
314performance. The `/cpu/self/xsmm/*` backends rely upon the
315[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU
316performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA.
317The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable
318performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple
319parallelization approach, where each thread treats a finite element. Using just in time
320compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this
321backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend
322achieve good peak performance for 1D, 2D, and low order 3D problems, but performance
323deteriorates very quickly when threads run out of registers.
324
325A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED
326examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`).
327This example solves the time-dependent Navier-Stokes equations of compressible gas
328dynamics in a static Eulerian three-dimensional frame, using structured high-order
329finite/spectral element spatial discretizations and explicit high-order time-stepping
330(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc,
331so that the pointwise physics (defined at quadrature points) is separated from the
332parallelization and meshing concerns.
333
334Backends available in this release:
335
336| CEED resource (`-ceed`)  | Backend                                             |
337|--------------------------|-----------------------------------------------------|
338| `/cpu/self/ref/serial`   | Serial reference implementation                     |
339| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
340| `/cpu/self/tmpl`         | Backend template, defaults to `/cpu/self/blocked`   |
341| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
342| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
343| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
344| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
345| `/cpu/occa`              | Serial OCCA kernels                                 |
346| `/gpu/occa`              | CUDA OCCA kernels                                   |
347| `/omp/occa`              | OpenMP OCCA kernels                                 |
348| `/ocl/occa`              | OpenCL OCCA kernels                                 |
349| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
350| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
351| `/gpu/magma`             | CUDA MAGMA kernels                                  |
352
353Examples available in this release:
354
355:::{list-table}
356:header-rows: 1
357:widths: auto
358* - User code
359  - Example
360* - `ceed`
361  - * ex1 (volume)
362* - `mfem`
363  - * BP1 (scalar mass operator)
364    * BP3 (scalar Laplace operator)
365* - `petsc`
366  - * BP1 (scalar mass operator)
367    * BP3 (scalar Laplace operator)
368    * Navier-Stokes
369* - `nek5000`
370  - * BP1 (scalar mass operator)
371    * BP3 (scalar Laplace operator)
372:::
373
374(v0-3)=
375
376## v0.3 (Sep 30, 2018)
377
378Notable features in this release include active/passive field interface, support for
379non-tensor bases, backend optimization, and improved Fortran interface. This release
380also focused on providing improved continuous integration, and many new tests with code
381coverage reports of about 90%. This release also provided a significant change to the
382public interface: a {ref}`CeedQFunction` can take any number of named input and output
383arguments while {ref}`CeedOperator` connects them to the actual data, which may be
384supplied explicitly to `CeedOperatorApply()` (active) or separately via
385`CeedOperatorSetField()` (passive). This interface change enables reusable libraries
386of CeedQFunctions and composition of block solvers constructed using
387{ref}`CeedOperator`. A concept of blocked restriction was added to this release and
388used in an optimized CPU backend. Although this is typically not visible to the user,
389it enables effective use of arbitrary-length SIMD while maintaining cache locality.
390This CPU backend also implements an algebraic factorization of tensor product gradients
391to perform fewer operations than standard application of interpolation and
392differentiation from nodes to quadrature points. This algebraic formulation
393automatically supports non-polynomial and non-interpolatory bases, thus is more general
394than the more common derivation in terms of Lagrange polynomials on the quadrature points.
395
396Backends available in this release:
397
398| CEED resource (`-ceed`) | Backend                                             |
399|-------------------------|-----------------------------------------------------|
400| `/cpu/self/blocked`     | Blocked reference implementation                    |
401| `/cpu/self/ref`         | Serial reference implementation                     |
402| `/cpu/self/tmpl`        | Backend template, defaults to `/cpu/self/blocked`   |
403| `/cpu/occa`             | Serial OCCA kernels                                 |
404| `/gpu/occa`             | CUDA OCCA kernels                                   |
405| `/omp/occa`             | OpenMP OCCA kernels                                 |
406| `/ocl/occa`             | OpenCL OCCA kernels                                 |
407| `/gpu/magma`            | CUDA MAGMA kernels                                  |
408
409Examples available in this release:
410
411:::{list-table}
412:header-rows: 1
413:widths: auto
414* - User code
415  - Example
416* - `ceed`
417  - * ex1 (volume)
418* - `mfem`
419  - * BP1 (scalar mass operator)
420    * BP3 (scalar Laplace operator)
421* - `petsc`
422  - * BP1 (scalar mass operator)
423    * BP3 (scalar Laplace operator)
424* - `nek5000`
425  - * BP1 (scalar mass operator)
426    * BP3 (scalar Laplace operator)
427:::
428
429(v0-21)=
430
431## v0.21 (Sep 30, 2018)
432
433A MAGMA backend (which relies upon the
434[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this
435release. This initial integration set up the framework of using MAGMA and provided the
436libCEED functionality through MAGMA kernels as one of libCEED’s computational backends.
437As any other backend, the MAGMA backend provides extended basic data structures for
438{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements
439the fundamental CEED building blocks to work with the new data structures.
440In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data
441but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled
442internally, and thus seamlessly to the user, through the functions/methods that are
443provided to support them.
444
445Backends available in this release:
446
447| CEED resource (`-ceed`) | Backend                         |
448|-------------------------|---------------------------------|
449| `/cpu/self`             | Serial reference implementation |
450| `/cpu/occa`             | Serial OCCA kernels             |
451| `/gpu/occa`             | CUDA OCCA kernels               |
452| `/omp/occa`             | OpenMP OCCA kernels             |
453| `/ocl/occa`             | OpenCL OCCA kernels             |
454| `/gpu/magma`            | CUDA MAGMA kernels              |
455
456Examples available in this release:
457
458:::{list-table}
459:header-rows: 1
460:widths: auto
461* - User code
462  - Example
463* - `ceed`
464  - * ex1 (volume)
465* - `mfem`
466  - * BP1 (scalar mass operator)
467    * BP3 (scalar Laplace operator)
468* - `petsc`
469  - * BP1 (scalar mass operator)
470* - `nek5000`
471  - * BP1 (scalar mass operator)
472:::
473
474(v0-2)=
475
476## v0.2 (Mar 30, 2018)
477
478libCEED was made publicly available the first full CEED software distribution, release
479CEED 1.0. The distribution was made available using the Spack package manager to provide
480a common, easy-to-use build environment, where the user can build the CEED distribution
481with all dependencies. This release included a new Fortran interface for the library.
482This release also contained major improvements in the OCCA backend (including a new
483`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to
484compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an
485`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application
486of the Laplace operator) was also added to this release.
487
488Backends available in this release:
489
490| CEED resource (`-ceed`) | Backend                         |
491|-------------------------|---------------------------------|
492| `/cpu/self`             | Serial reference implementation |
493| `/cpu/occa`             | Serial OCCA kernels             |
494| `/gpu/occa`             | CUDA OCCA kernels               |
495| `/omp/occa`             | OpenMP OCCA kernels             |
496| `/ocl/occa`             | OpenCL OCCA kernels             |
497
498Examples available in this release:
499
500:::{list-table}
501:header-rows: 1
502:widths: auto
503* - User code
504  - Example
505* - `ceed`
506  - * ex1 (volume)
507* - `mfem`
508  - * BP1 (scalar mass operator)
509    * BP3 (scalar Laplace operator)
510* - `petsc`
511  - * BP1 (scalar mass operator)
512* - `nek5000`
513  - * BP1 (scalar mass operator)
514:::
515
516(v0-1)=
517
518## v0.1 (Jan 3, 2018)
519
520Initial low-level API of the CEED project. The low-level API provides a set of Finite
521Elements kernels and components for writing new low-level kernels. Examples include:
522vector and sparse linear algebra, element matrix assembly over a batch of elements,
523partial assembly and action for efficient high-order operators like mass, diffusion,
524advection, etc. The main goal of the low-level API is to establish the basis for the
525high-level API. Also, identifying such low-level kernels and providing a reference
526implementation for them serves as the basis for specialized backend implementations.
527This release contained several backends: `/cpu/self`, and backends which rely upon the
528[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`,
529`/gpu/occa`, and `/omp/occa`.
530It also included several examples, in the `examples` folder:
531A standalone code that shows the usage of libCEED (with no external
532dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1
533(with the application of the mass operator); and a `petsc` example to perform BP1
534(with the application of the mass operator).
535
536Backends available in this release:
537
538| CEED resource (`-ceed`) | Backend                         |
539|-------------------------|---------------------------------|
540| `/cpu/self`             | Serial reference implementation |
541| `/cpu/occa`             | Serial OCCA kernels             |
542| `/gpu/occa`             | CUDA OCCA kernels               |
543| `/omp/occa`             | OpenMP OCCA kernels             |
544
545Examples available in this release:
546
547| User code             | Example                           |
548|-----------------------|-----------------------------------|
549| `ceed`                | ex1 (scalar Laplace operator)     |
550| `mfem`                | BP1 (scalar mass operator)        |
551| `petsc`               | BP1 (scalar mass operator)        |
552```
553