xref: /libCEED/doc/sphinx/source/releasenotes.md (revision f9996dfd0db601698d1ad85cf5741d3360a86e68)
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
25### New features
26
27- `CeedScalar` can now be set as `float` or `double` at compile time.
28- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends.
29- 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.
30- Remove need to guard library headers in QFunction source for code generation backends.
31- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present.
32- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally.
33- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file.
34- Added {c:func}`CeedBasisCreateHdiv` convenience constructor and CPU backend support for non-tensor H(div) elements.
35
36### Maintainability
37
38- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage.
39- `Include-what-you-use` makefile target added as `make iwyu`.
40- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase.
41- Put GPU JiTed kernel source code into separate files.
42- 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.
43
44(v0-9)=
45
46## v0.9 (Jul 6, 2021)
47
48### Interface changes
49
50- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact.
51
52### New features
53
54- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use.
55- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends.
56- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations.
57- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed).
58
59### Performance improvements
60
61### Examples
62
63- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations.
64- Fluid mechanics example adds GPU support and improves modularity.
65
66### Deprecated backends
67
68- 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.
69
70(v0-8)=
71
72## v0.8 (Mar 31, 2021)
73
74### Interface changes
75
76- Error handling improved to include enumerated error codes for C interface return values.
77- 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.
78
79### New features
80
81- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features.
82- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly.
83- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators.
84
85### Performance improvements
86
87- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`.
88- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`.
89
90### Examples
91
92- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions.
93- {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.
94- {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.
95- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree.
96
97(v0-7)=
98
99## v0.7 (Sep 29, 2020)
100
101### Interface changes
102
103- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors.
104  As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`.
105  These changes improve support for mixed finite element methods.
106- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature.
107- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity.
108- 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.
109- 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.
110- 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.
111- 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.
112  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`.
113- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory.
114- 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.
115
116### New features
117
118- New HIP backend: `/gpu/hip/ref`.
119- CeedQFunction support for user `CUfunction`s in some backends
120
121### Performance improvements
122
123- OCCA backend rebuilt to facilitate future performance enhancements.
124- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`.
125
126### Examples
127
128- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions.
129
130### Deprecated backends
131
132- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`.
133
134(v0-6)=
135
136## v0.6 (Mar 29, 2020)
137
138libCEED v0.6 contains numerous new features and examples, as well as expanded
139documentation in [this new website](https://libceed.readthedocs.io).
140
141### New features
142
143- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly
144  1-1 correspondence with the C interface, plus some convenience features.  For instance,
145  data stored in the {cpp:type}`CeedVector` structure are available without copy as
146  {py:class}`numpy.ndarray`.  Short tutorials are provided in
147  [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/).
148- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point,
149  {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal
150  ({c:func}`CeedOperatorAssembleLinearDiagonal`).  These operations are useful for
151  preconditioning ingredients and are used in the libCEED's multigrid examples.
152- The inverse of separable operators can be obtained using
153  {c:func}`CeedOperatorCreateFDMElementInverse` and applied with
154  {c:func}`CeedOperatorApply`.  This is a useful preconditioning ingredient,
155  especially for Laplacians and related operators.
156- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`,
157  {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`.
158- Make public accessors for various attributes to facilitate writing composable code.
159- New backend: `/cpu/self/memcheck/serial`.
160- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA
161  backends.  (Single source is coming soon for OCCA backends.)
162- Fix some missing edge cases in CUDA backend.
163
164### Performance Improvements
165
166- MAGMA backend performance optimization and non-tensor bases.
167- No-copy optimization in {c:func}`CeedOperatorApply`.
168
169### Interface changes
170
171- Replace {code}`CeedElemRestrictionCreateIdentity` and
172  {code}`CeedElemRestrictionCreateBlocked` with more flexible
173  {c:func}`CeedElemRestrictionCreateStrided` and
174  {c:func}`CeedElemRestrictionCreateBlockedStrided`.
175- Add arguments to {c:func}`CeedQFunctionCreateIdentity`.
176- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification
177  with {cpp:enum}`CeedInterlaceMode`.  This is now an attribute of the
178  {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no
179  longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and
180  {c:func}`CeedElemRestrictionApply`.
181
182### Examples
183
184libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`.
185Notable additions include:
186
187- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of
188  a domain in 1, 2, and 3 dimensions by applying a Laplacian.
189
190- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area
191  of domains (like the cube and sphere) by direct integration on a surface mesh;
192  demonstrates geometric dimension different from topological dimension.
193
194- PETSc {ref}`example-petsc-bps`:
195
196  - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support.
197  - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements
198    and transparent CUDA support.
199  - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`):
200    generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric
201    dimension different from topological dimension.
202
203- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid
204  solver with algebraic multigrid coarse solve.
205
206- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly
207  `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`),
208  implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and
209  quasi-2D computational domain support.
210
211- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for
212  linear elasticity, small-strain hyperelasticity, and globalized finite-strain
213  hyperelasticity using p-multigrid with algebraic multigrid coarse solve.
214
215(v0-5)=
216
217## v0.5 (Sep 18, 2019)
218
219For this release, several improvements were made. Two new CUDA backends were added to
220the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art
221performance using single-source {ref}`CeedQFunction`. From this release, users
222can define Q-Functions in a single source code independently of the targeted backend
223with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU
224compilation of the user provided {ref}`CeedQFunction` code. To allow a unified
225declaration, the {ref}`CeedQFunction` API has undergone a slight change:
226the `QFunctionField` parameter `ncomp` has been changed to `size`. This change
227requires setting the previous value of `ncomp` to `ncomp*dim` when adding a
228`QFunctionField` with eval mode `CEED EVAL GRAD`.
229
230Additionally, new CPU backends
231were included in this release, such as the `/cpu/self/opt/*` backends (which are
232written in pure C and use partial **E-vectors** to improve performance) and the
233`/cpu/self/ref/memcheck` backend (which relies upon the
234[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user
235{ref}`CeedQFunction` have no undefined values).
236This release also included various performance improvements, bug fixes, new examples,
237and improved tests. Among these improvements, vectorized instructions for
238{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD`
239instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and
240identity Q-Functions were introduced, and the PETSc benchmark problems were expanded
241to include unstructured meshes handling were. For this expansion, the prior version of
242the PETSc BPs, which only included data associated with structured geometries, were
243renamed `bpsraw`, and the new version of the BPs, which can handle data associated
244with any unstructured geometry, were called `bps`. Additionally, other benchmark
245problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively),
246and BP5 and BP6 (the collocated versions---for which the quadrature points are the same
247as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc
248examples. Furthermoew, another standalone libCEED example, called `ex2`, which
249computes the surface area of a given mesh was added to this release.
250
251Backends available in this release:
252
253| CEED resource (`-ceed`)  | Backend                                             |
254|--------------------------|-----------------------------------------------------|
255| `/cpu/self/ref/serial`   | Serial reference implementation                     |
256| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
257| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks            |
258| `/cpu/self/opt/serial`   | Serial optimized C implementation                   |
259| `/cpu/self/opt/blocked`  | Blocked optimized C implementation                  |
260| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
261| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
262| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
263| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
264| `/cpu/occa`              | Serial OCCA kernels                                 |
265| `/gpu/occa`              | CUDA OCCA kernels                                   |
266| `/omp/occa`              | OpenMP OCCA kernels                                 |
267| `/ocl/occa`              | OpenCL OCCA kernels                                 |
268| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
269| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
270| `/gpu/cuda/shared`       | Optimized pure CUDA kernels using shared memory     |
271| `/gpu/cuda/gen`          | Optimized pure CUDA kernels using code generation   |
272| `/gpu/magma`             | CUDA MAGMA kernels                                  |
273
274Examples available in this release:
275
276:::{list-table}
277:header-rows: 1
278:widths: auto
279* - User code
280  - Example
281* - `ceed`
282  - * ex1 (volume)
283    * ex2 (surface)
284* - `mfem`
285  - * BP1 (scalar mass operator)
286    * BP3 (scalar Laplace operator)
287* - `petsc`
288  - * BP1 (scalar mass operator)
289    * BP2 (vector mass operator)
290    * BP3 (scalar Laplace operator)
291    * BP4 (vector Laplace operator)
292    * BP5 (collocated scalar Laplace operator)
293    * BP6 (collocated vector Laplace operator)
294    * Navier-Stokes
295* - `nek5000`
296  - * BP1 (scalar mass operator)
297    * BP3 (scalar Laplace operator)
298:::
299
300(v0-4)=
301
302## v0.4 (Apr 1, 2019)
303
304libCEED v0.4 was made again publicly available in the second full CEED software
305distribution, release CEED 2.0. This release contained notable features, such as
306four new CPU backends, two new GPU backends, CPU backend optimizations, initial
307support for operator composition, performance benchmarking, and a Navier-Stokes demo.
308The new CPU backends in this release came in two families. The `/cpu/self/*/serial`
309backends process one element at a time and are intended for meshes with a smaller number
310of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of
311eight interlaced elements and are intended for meshes with higher numbers of elements.
312The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU
313performance. The `/cpu/self/xsmm/*` backends rely upon the
314[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU
315performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA.
316The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable
317performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple
318parallelization approach, where each thread treats a finite element. Using just in time
319compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this
320backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend
321achieve good peak performance for 1D, 2D, and low order 3D problems, but performance
322deteriorates very quickly when threads run out of registers.
323
324A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED
325examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`).
326This example solves the time-dependent Navier-Stokes equations of compressible gas
327dynamics in a static Eulerian three-dimensional frame, using structured high-order
328finite/spectral element spatial discretizations and explicit high-order time-stepping
329(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc,
330so that the pointwise physics (defined at quadrature points) is separated from the
331parallelization and meshing concerns.
332
333Backends available in this release:
334
335| CEED resource (`-ceed`)  | Backend                                             |
336|--------------------------|-----------------------------------------------------|
337| `/cpu/self/ref/serial`   | Serial reference implementation                     |
338| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
339| `/cpu/self/tmpl`         | Backend template, defaults to `/cpu/self/blocked`   |
340| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
341| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
342| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
343| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
344| `/cpu/occa`              | Serial OCCA kernels                                 |
345| `/gpu/occa`              | CUDA OCCA kernels                                   |
346| `/omp/occa`              | OpenMP OCCA kernels                                 |
347| `/ocl/occa`              | OpenCL OCCA kernels                                 |
348| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
349| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
350| `/gpu/magma`             | CUDA MAGMA kernels                                  |
351
352Examples available in this release:
353
354:::{list-table}
355:header-rows: 1
356:widths: auto
357* - User code
358  - Example
359* - `ceed`
360  - * ex1 (volume)
361* - `mfem`
362  - * BP1 (scalar mass operator)
363    * BP3 (scalar Laplace operator)
364* - `petsc`
365  - * BP1 (scalar mass operator)
366    * BP3 (scalar Laplace operator)
367    * Navier-Stokes
368* - `nek5000`
369  - * BP1 (scalar mass operator)
370    * BP3 (scalar Laplace operator)
371:::
372
373(v0-3)=
374
375## v0.3 (Sep 30, 2018)
376
377Notable features in this release include active/passive field interface, support for
378non-tensor bases, backend optimization, and improved Fortran interface. This release
379also focused on providing improved continuous integration, and many new tests with code
380coverage reports of about 90%. This release also provided a significant change to the
381public interface: a {ref}`CeedQFunction` can take any number of named input and output
382arguments while {ref}`CeedOperator` connects them to the actual data, which may be
383supplied explicitly to `CeedOperatorApply()` (active) or separately via
384`CeedOperatorSetField()` (passive). This interface change enables reusable libraries
385of CeedQFunctions and composition of block solvers constructed using
386{ref}`CeedOperator`. A concept of blocked restriction was added to this release and
387used in an optimized CPU backend. Although this is typically not visible to the user,
388it enables effective use of arbitrary-length SIMD while maintaining cache locality.
389This CPU backend also implements an algebraic factorization of tensor product gradients
390to perform fewer operations than standard application of interpolation and
391differentiation from nodes to quadrature points. This algebraic formulation
392automatically supports non-polynomial and non-interpolatory bases, thus is more general
393than the more common derivation in terms of Lagrange polynomials on the quadrature points.
394
395Backends available in this release:
396
397| CEED resource (`-ceed`) | Backend                                             |
398|-------------------------|-----------------------------------------------------|
399| `/cpu/self/blocked`     | Blocked reference implementation                    |
400| `/cpu/self/ref`         | Serial reference implementation                     |
401| `/cpu/self/tmpl`        | Backend template, defaults to `/cpu/self/blocked`   |
402| `/cpu/occa`             | Serial OCCA kernels                                 |
403| `/gpu/occa`             | CUDA OCCA kernels                                   |
404| `/omp/occa`             | OpenMP OCCA kernels                                 |
405| `/ocl/occa`             | OpenCL OCCA kernels                                 |
406| `/gpu/magma`            | CUDA MAGMA kernels                                  |
407
408Examples available in this release:
409
410:::{list-table}
411:header-rows: 1
412:widths: auto
413* - User code
414  - Example
415* - `ceed`
416  - * ex1 (volume)
417* - `mfem`
418  - * BP1 (scalar mass operator)
419    * BP3 (scalar Laplace operator)
420* - `petsc`
421  - * BP1 (scalar mass operator)
422    * BP3 (scalar Laplace operator)
423* - `nek5000`
424  - * BP1 (scalar mass operator)
425    * BP3 (scalar Laplace operator)
426:::
427
428(v0-21)=
429
430## v0.21 (Sep 30, 2018)
431
432A MAGMA backend (which relies upon the
433[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this
434release. This initial integration set up the framework of using MAGMA and provided the
435libCEED functionality through MAGMA kernels as one of libCEED’s computational backends.
436As any other backend, the MAGMA backend provides extended basic data structures for
437{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements
438the fundamental CEED building blocks to work with the new data structures.
439In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data
440but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled
441internally, and thus seamlessly to the user, through the functions/methods that are
442provided to support them.
443
444Backends available in this release:
445
446| CEED resource (`-ceed`) | Backend                         |
447|-------------------------|---------------------------------|
448| `/cpu/self`             | Serial reference implementation |
449| `/cpu/occa`             | Serial OCCA kernels             |
450| `/gpu/occa`             | CUDA OCCA kernels               |
451| `/omp/occa`             | OpenMP OCCA kernels             |
452| `/ocl/occa`             | OpenCL OCCA kernels             |
453| `/gpu/magma`            | CUDA MAGMA kernels              |
454
455Examples available in this release:
456
457:::{list-table}
458:header-rows: 1
459:widths: auto
460* - User code
461  - Example
462* - `ceed`
463  - * ex1 (volume)
464* - `mfem`
465  - * BP1 (scalar mass operator)
466    * BP3 (scalar Laplace operator)
467* - `petsc`
468  - * BP1 (scalar mass operator)
469* - `nek5000`
470  - * BP1 (scalar mass operator)
471:::
472
473(v0-2)=
474
475## v0.2 (Mar 30, 2018)
476
477libCEED was made publicly available the first full CEED software distribution, release
478CEED 1.0. The distribution was made available using the Spack package manager to provide
479a common, easy-to-use build environment, where the user can build the CEED distribution
480with all dependencies. This release included a new Fortran interface for the library.
481This release also contained major improvements in the OCCA backend (including a new
482`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to
483compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an
484`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application
485of the Laplace operator) was also added to this release.
486
487Backends available in this release:
488
489| CEED resource (`-ceed`) | Backend                         |
490|-------------------------|---------------------------------|
491| `/cpu/self`             | Serial reference implementation |
492| `/cpu/occa`             | Serial OCCA kernels             |
493| `/gpu/occa`             | CUDA OCCA kernels               |
494| `/omp/occa`             | OpenMP OCCA kernels             |
495| `/ocl/occa`             | OpenCL OCCA kernels             |
496
497Examples available in this release:
498
499:::{list-table}
500:header-rows: 1
501:widths: auto
502* - User code
503  - Example
504* - `ceed`
505  - * ex1 (volume)
506* - `mfem`
507  - * BP1 (scalar mass operator)
508    * BP3 (scalar Laplace operator)
509* - `petsc`
510  - * BP1 (scalar mass operator)
511* - `nek5000`
512  - * BP1 (scalar mass operator)
513:::
514
515(v0-1)=
516
517## v0.1 (Jan 3, 2018)
518
519Initial low-level API of the CEED project. The low-level API provides a set of Finite
520Elements kernels and components for writing new low-level kernels. Examples include:
521vector and sparse linear algebra, element matrix assembly over a batch of elements,
522partial assembly and action for efficient high-order operators like mass, diffusion,
523advection, etc. The main goal of the low-level API is to establish the basis for the
524high-level API. Also, identifying such low-level kernels and providing a reference
525implementation for them serves as the basis for specialized backend implementations.
526This release contained several backends: `/cpu/self`, and backends which rely upon the
527[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`,
528`/gpu/occa`, and `/omp/occa`.
529It also included several examples, in the `examples` folder:
530A standalone code that shows the usage of libCEED (with no external
531dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1
532(with the application of the mass operator); and a `petsc` example to perform BP1
533(with the application of the mass operator).
534
535Backends available in this release:
536
537| CEED resource (`-ceed`) | Backend                         |
538|-------------------------|---------------------------------|
539| `/cpu/self`             | Serial reference implementation |
540| `/cpu/occa`             | Serial OCCA kernels             |
541| `/gpu/occa`             | CUDA OCCA kernels               |
542| `/omp/occa`             | OpenMP OCCA kernels             |
543
544Examples available in this release:
545
546| User code             | Example                           |
547|-----------------------|-----------------------------------|
548| `ceed`                | ex1 (scalar Laplace operator)     |
549| `mfem`                | BP1 (scalar mass operator)        |
550| `petsc`               | BP1 (scalar mass operator)        |
551```
552