xref: /libCEED/doc/sphinx/source/releasenotes.md (revision cdf32b9369b06315737c6f9df3317771bd8e77c7)
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 {ref}`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
33### Maintainability
34
35- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage.
36- `Include-what-you-use` makefile target added as `make iwyu`.
37- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase.
38
39(v0-9)=
40
41## v0.9 (Jul 6, 2021)
42
43### Interface changes
44
45- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact.
46
47### New features
48
49- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use.
50- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends.
51- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations.
52- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed).
53
54### Performance improvements
55
56### Examples
57
58- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations.
59- Fluid mechanics example adds GPU support and improves modularity.
60
61### Deprecated backends
62
63- 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.
64
65(v0-8)=
66
67## v0.8 (Mar 31, 2021)
68
69### Interface changes
70
71- Error handling improved to include enumerated error codes for C interface return values.
72- 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.
73
74### New features
75
76- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features.
77- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly.
78- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators.
79
80### Performance improvements
81
82- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`.
83- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`.
84
85### Examples
86
87- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions.
88- {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.
89- {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.
90- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree.
91
92(v0-7)=
93
94## v0.7 (Sep 29, 2020)
95
96### Interface changes
97
98- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors.
99  As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`.
100  These changes improve support for mixed finite element methods.
101- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature.
102- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity.
103- 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.
104- 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.
105- 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.
106- 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.
107  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`.
108- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory.
109- 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.
110
111### New features
112
113- New HIP backend: `/gpu/hip/ref`.
114- CeedQFunction support for user `CUfunction`s in some backends
115
116### Performance improvements
117
118- OCCA backend rebuilt to facilitate future performance enhancements.
119- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`.
120
121### Examples
122
123- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions.
124
125### Deprecated backends
126
127- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`.
128
129(v0-6)=
130
131## v0.6 (Mar 29, 2020)
132
133libCEED v0.6 contains numerous new features and examples, as well as expanded
134documentation in [this new website](https://libceed.readthedocs.io).
135
136### New features
137
138- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly
139  1-1 correspondence with the C interface, plus some convenience features.  For instance,
140  data stored in the {cpp:type}`CeedVector` structure are available without copy as
141  {py:class}`numpy.ndarray`.  Short tutorials are provided in
142  [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/).
143- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point,
144  {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal
145  ({c:func}`CeedOperatorAssembleLinearDiagonal`).  These operations are useful for
146  preconditioning ingredients and are used in the libCEED's multigrid examples.
147- The inverse of separable operators can be obtained using
148  {c:func}`CeedOperatorCreateFDMElementInverse` and applied with
149  {c:func}`CeedOperatorApply`.  This is a useful preconditioning ingredient,
150  especially for Laplacians and related operators.
151- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`,
152  {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`.
153- Make public accessors for various attributes to facilitate writing composable code.
154- New backend: `/cpu/self/memcheck/serial`.
155- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA
156  backends.  (Single source is coming soon for OCCA backends.)
157- Fix some missing edge cases in CUDA backend.
158
159### Performance Improvements
160
161- MAGMA backend performance optimization and non-tensor bases.
162- No-copy optimization in {c:func}`CeedOperatorApply`.
163
164### Interface changes
165
166- Replace {code}`CeedElemRestrictionCreateIdentity` and
167  {code}`CeedElemRestrictionCreateBlocked` with more flexible
168  {c:func}`CeedElemRestrictionCreateStrided` and
169  {c:func}`CeedElemRestrictionCreateBlockedStrided`.
170- Add arguments to {c:func}`CeedQFunctionCreateIdentity`.
171- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification
172  with {cpp:enum}`CeedInterlaceMode`.  This is now an attribute of the
173  {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no
174  longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and
175  {c:func}`CeedElemRestrictionApply`.
176
177### Examples
178
179libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`.
180Notable additions include:
181
182- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of
183  a domain in 1, 2, and 3 dimensions by applying a Laplacian.
184
185- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area
186  of domains (like the cube and sphere) by direct integration on a surface mesh;
187  demonstrates geometric dimension different from topological dimension.
188
189- PETSc {ref}`example-petsc-bps`:
190
191  - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support.
192  - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements
193    and transparent CUDA support.
194  - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`):
195    generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric
196    dimension different from topological dimension.
197
198- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid
199  solver with algebraic multigrid coarse solve.
200
201- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly
202  `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`),
203  implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and
204  quasi-2D computational domain support.
205
206- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for
207  linear elasticity, small-strain hyperelasticity, and globalized finite-strain
208  hyperelasticity using p-multigrid with algebraic multigrid coarse solve.
209
210(v0-5)=
211
212## v0.5 (Sep 18, 2019)
213
214For this release, several improvements were made. Two new CUDA backends were added to
215the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art
216performance using single-source {ref}`CeedQFunction`. From this release, users
217can define Q-Functions in a single source code independently of the targeted backend
218with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU
219compilation of the user provided {ref}`CeedQFunction` code. To allow a unified
220declaration, the {ref}`CeedQFunction` API has undergone a slight change:
221the `QFunctionField` parameter `ncomp` has been changed to `size`. This change
222requires setting the previous value of `ncomp` to `ncomp*dim` when adding a
223`QFunctionField` with eval mode `CEED EVAL GRAD`.
224
225Additionally, new CPU backends
226were included in this release, such as the `/cpu/self/opt/*` backends (which are
227written in pure C and use partial **E-vectors** to improve performance) and the
228`/cpu/self/ref/memcheck` backend (which relies upon the
229[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user
230{ref}`CeedQFunction` have no undefined values).
231This release also included various performance improvements, bug fixes, new examples,
232and improved tests. Among these improvements, vectorized instructions for
233{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD`
234instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and
235identity Q-Functions were introduced, and the PETSc benchmark problems were expanded
236to include unstructured meshes handling were. For this expansion, the prior version of
237the PETSc BPs, which only included data associated with structured geometries, were
238renamed `bpsraw`, and the new version of the BPs, which can handle data associated
239with any unstructured geometry, were called `bps`. Additionally, other benchmark
240problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively),
241and BP5 and BP6 (the collocated versions---for which the quadrature points are the same
242as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc
243examples. Furthermoew, another standalone libCEED example, called `ex2`, which
244computes the surface area of a given mesh was added to this release.
245
246Backends available in this release:
247
248| CEED resource (`-ceed`)  | Backend                                             |
249|--------------------------|-----------------------------------------------------|
250| `/cpu/self/ref/serial`   | Serial reference implementation                     |
251| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
252| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks            |
253| `/cpu/self/opt/serial`   | Serial optimized C implementation                   |
254| `/cpu/self/opt/blocked`  | Blocked optimized C implementation                  |
255| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
256| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
257| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
258| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
259| `/cpu/occa`              | Serial OCCA kernels                                 |
260| `/gpu/occa`              | CUDA OCCA kernels                                   |
261| `/omp/occa`              | OpenMP OCCA kernels                                 |
262| `/ocl/occa`              | OpenCL OCCA kernels                                 |
263| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
264| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
265| `/gpu/cuda/shared`       | Optimized pure CUDA kernels using shared memory     |
266| `/gpu/cuda/gen`          | Optimized pure CUDA kernels using code generation   |
267| `/gpu/magma`             | CUDA MAGMA kernels                                  |
268
269Examples available in this release:
270
271:::{list-table}
272:header-rows: 1
273:widths: auto
274* - User code
275  - Example
276* - `ceed`
277  - * ex1 (volume)
278    * ex2 (surface)
279* - `mfem`
280  - * BP1 (scalar mass operator)
281    * BP3 (scalar Laplace operator)
282* - `petsc`
283  - * BP1 (scalar mass operator)
284    * BP2 (vector mass operator)
285    * BP3 (scalar Laplace operator)
286    * BP4 (vector Laplace operator)
287    * BP5 (collocated scalar Laplace operator)
288    * BP6 (collocated vector Laplace operator)
289    * Navier-Stokes
290* - `nek5000`
291  - * BP1 (scalar mass operator)
292    * BP3 (scalar Laplace operator)
293:::
294
295(v0-4)=
296
297## v0.4 (Apr 1, 2019)
298
299libCEED v0.4 was made again publicly available in the second full CEED software
300distribution, release CEED 2.0. This release contained notable features, such as
301four new CPU backends, two new GPU backends, CPU backend optimizations, initial
302support for operator composition, performance benchmarking, and a Navier-Stokes demo.
303The new CPU backends in this release came in two families. The `/cpu/self/*/serial`
304backends process one element at a time and are intended for meshes with a smaller number
305of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of
306eight interlaced elements and are intended for meshes with higher numbers of elements.
307The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU
308performance. The `/cpu/self/xsmm/*` backends rely upon the
309[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU
310performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA.
311The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable
312performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple
313parallelization approach, where each thread treats a finite element. Using just in time
314compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this
315backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend
316achieve good peak performance for 1D, 2D, and low order 3D problems, but performance
317deteriorates very quickly when threads run out of registers.
318
319A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED
320examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`).
321This example solves the time-dependent Navier-Stokes equations of compressible gas
322dynamics in a static Eulerian three-dimensional frame, using structured high-order
323finite/spectral element spatial discretizations and explicit high-order time-stepping
324(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc,
325so that the pointwise physics (defined at quadrature points) is separated from the
326parallelization and meshing concerns.
327
328Backends available in this release:
329
330| CEED resource (`-ceed`)  | Backend                                             |
331|--------------------------|-----------------------------------------------------|
332| `/cpu/self/ref/serial`   | Serial reference implementation                     |
333| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
334| `/cpu/self/tmpl`         | Backend template, defaults to `/cpu/self/blocked`   |
335| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
336| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
337| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
338| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
339| `/cpu/occa`              | Serial OCCA kernels                                 |
340| `/gpu/occa`              | CUDA OCCA kernels                                   |
341| `/omp/occa`              | OpenMP OCCA kernels                                 |
342| `/ocl/occa`              | OpenCL OCCA kernels                                 |
343| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
344| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
345| `/gpu/magma`             | CUDA MAGMA kernels                                  |
346
347Examples available in this release:
348
349:::{list-table}
350:header-rows: 1
351:widths: auto
352* - User code
353  - Example
354* - `ceed`
355  - * ex1 (volume)
356* - `mfem`
357  - * BP1 (scalar mass operator)
358    * BP3 (scalar Laplace operator)
359* - `petsc`
360  - * BP1 (scalar mass operator)
361    * BP3 (scalar Laplace operator)
362    * Navier-Stokes
363* - `nek5000`
364  - * BP1 (scalar mass operator)
365    * BP3 (scalar Laplace operator)
366:::
367
368(v0-3)=
369
370## v0.3 (Sep 30, 2018)
371
372Notable features in this release include active/passive field interface, support for
373non-tensor bases, backend optimization, and improved Fortran interface. This release
374also focused on providing improved continuous integration, and many new tests with code
375coverage reports of about 90%. This release also provided a significant change to the
376public interface: a {ref}`CeedQFunction` can take any number of named input and output
377arguments while {ref}`CeedOperator` connects them to the actual data, which may be
378supplied explicitly to `CeedOperatorApply()` (active) or separately via
379`CeedOperatorSetField()` (passive). This interface change enables reusable libraries
380of CeedQFunctions and composition of block solvers constructed using
381{ref}`CeedOperator`. A concept of blocked restriction was added to this release and
382used in an optimized CPU backend. Although this is typically not visible to the user,
383it enables effective use of arbitrary-length SIMD while maintaining cache locality.
384This CPU backend also implements an algebraic factorization of tensor product gradients
385to perform fewer operations than standard application of interpolation and
386differentiation from nodes to quadrature points. This algebraic formulation
387automatically supports non-polynomial and non-interpolatory bases, thus is more general
388than the more common derivation in terms of Lagrange polynomials on the quadrature points.
389
390Backends available in this release:
391
392| CEED resource (`-ceed`) | Backend                                             |
393|-------------------------|-----------------------------------------------------|
394| `/cpu/self/blocked`     | Blocked reference implementation                    |
395| `/cpu/self/ref`         | Serial reference implementation                     |
396| `/cpu/self/tmpl`        | Backend template, defaults to `/cpu/self/blocked`   |
397| `/cpu/occa`             | Serial OCCA kernels                                 |
398| `/gpu/occa`             | CUDA OCCA kernels                                   |
399| `/omp/occa`             | OpenMP OCCA kernels                                 |
400| `/ocl/occa`             | OpenCL OCCA kernels                                 |
401| `/gpu/magma`            | CUDA MAGMA kernels                                  |
402
403Examples available in this release:
404
405:::{list-table}
406:header-rows: 1
407:widths: auto
408* - User code
409  - Example
410* - `ceed`
411  - * ex1 (volume)
412* - `mfem`
413  - * BP1 (scalar mass operator)
414    * BP3 (scalar Laplace operator)
415* - `petsc`
416  - * BP1 (scalar mass operator)
417    * BP3 (scalar Laplace operator)
418* - `nek5000`
419  - * BP1 (scalar mass operator)
420    * BP3 (scalar Laplace operator)
421:::
422
423(v0-21)=
424
425## v0.21 (Sep 30, 2018)
426
427A MAGMA backend (which relies upon the
428[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this
429release. This initial integration set up the framework of using MAGMA and provided the
430libCEED functionality through MAGMA kernels as one of libCEED’s computational backends.
431As any other backend, the MAGMA backend provides extended basic data structures for
432{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements
433the fundamental CEED building blocks to work with the new data structures.
434In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data
435but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled
436internally, and thus seamlessly to the user, through the functions/methods that are
437provided to support them.
438
439Backends available in this release:
440
441| CEED resource (`-ceed`) | Backend                         |
442|-------------------------|---------------------------------|
443| `/cpu/self`             | Serial reference implementation |
444| `/cpu/occa`             | Serial OCCA kernels             |
445| `/gpu/occa`             | CUDA OCCA kernels               |
446| `/omp/occa`             | OpenMP OCCA kernels             |
447| `/ocl/occa`             | OpenCL OCCA kernels             |
448| `/gpu/magma`            | CUDA MAGMA kernels              |
449
450Examples available in this release:
451
452:::{list-table}
453:header-rows: 1
454:widths: auto
455* - User code
456  - Example
457* - `ceed`
458  - * ex1 (volume)
459* - `mfem`
460  - * BP1 (scalar mass operator)
461    * BP3 (scalar Laplace operator)
462* - `petsc`
463  - * BP1 (scalar mass operator)
464* - `nek5000`
465  - * BP1 (scalar mass operator)
466:::
467
468(v0-2)=
469
470## v0.2 (Mar 30, 2018)
471
472libCEED was made publicly available the first full CEED software distribution, release
473CEED 1.0. The distribution was made available using the Spack package manager to provide
474a common, easy-to-use build environment, where the user can build the CEED distribution
475with all dependencies. This release included a new Fortran interface for the library.
476This release also contained major improvements in the OCCA backend (including a new
477`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to
478compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an
479`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application
480of the Laplace operator) was also added to this release.
481
482Backends available in this release:
483
484| CEED resource (`-ceed`) | Backend                         |
485|-------------------------|---------------------------------|
486| `/cpu/self`             | Serial reference implementation |
487| `/cpu/occa`             | Serial OCCA kernels             |
488| `/gpu/occa`             | CUDA OCCA kernels               |
489| `/omp/occa`             | OpenMP OCCA kernels             |
490| `/ocl/occa`             | OpenCL OCCA kernels             |
491
492Examples available in this release:
493
494:::{list-table}
495:header-rows: 1
496:widths: auto
497* - User code
498  - Example
499* - `ceed`
500  - * ex1 (volume)
501* - `mfem`
502  - * BP1 (scalar mass operator)
503    * BP3 (scalar Laplace operator)
504* - `petsc`
505  - * BP1 (scalar mass operator)
506* - `nek5000`
507  - * BP1 (scalar mass operator)
508:::
509
510(v0-1)=
511
512## v0.1 (Jan 3, 2018)
513
514Initial low-level API of the CEED project. The low-level API provides a set of Finite
515Elements kernels and components for writing new low-level kernels. Examples include:
516vector and sparse linear algebra, element matrix assembly over a batch of elements,
517partial assembly and action for efficient high-order operators like mass, diffusion,
518advection, etc. The main goal of the low-level API is to establish the basis for the
519high-level API. Also, identifying such low-level kernels and providing a reference
520implementation for them serves as the basis for specialized backend implementations.
521This release contained several backends: `/cpu/self`, and backends which rely upon the
522[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`,
523`/gpu/occa`, and `/omp/occa`.
524It also included several examples, in the `examples` folder:
525A standalone code that shows the usage of libCEED (with no external
526dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1
527(with the application of the mass operator); and a `petsc` example to perform BP1
528(with the application of the mass operator).
529
530Backends available in this release:
531
532| CEED resource (`-ceed`) | Backend                         |
533|-------------------------|---------------------------------|
534| `/cpu/self`             | Serial reference implementation |
535| `/cpu/occa`             | Serial OCCA kernels             |
536| `/gpu/occa`             | CUDA OCCA kernels               |
537| `/omp/occa`             | OpenMP OCCA kernels             |
538
539Examples available in this release:
540
541| User code             | Example                           |
542|-----------------------|-----------------------------------|
543| `ceed`                | ex1 (scalar Laplace operator)     |
544| `mfem`                | BP1 (scalar mass operator)        |
545| `petsc`               | BP1 (scalar mass operator)        |
546```
547