xref: /libCEED/doc/sphinx/source/releasenotes.md (revision 82fe8277b7bbee464591a66c51e14f4b35f536f3)
1# Changes/Release Notes
2
3On this page we provide a summary of the main API changes, new features and examples for each release of libCEED.
4
5(main)=
6
7## Current `main` branch
8
9### Interface changes
10
11- Added {c:func}`CeedOperatorSetName` for more readable {c:func}`CeedOperatorView` output.
12
13### New features
14
15- Update `/cpu/self/memcheck/*` backends to help verify `CeedQFunctionContext` data sizes provided by user.
16
17### Bugfix
18
19- Fix storing of indices for `CeedElemRestriction` on the host with GPU backends.
20- Fix `CeedElemRestriction` sizing for {c:func}`CeedOperatorAssemblePointBlockDiagonal`.
21- Fix bugs in CPU implementation of {c:func}`CeedOperatorLinearAssemble` when there are different number of active input modes and active output modes.
22
23### Examples
24
25- Added various performance enhancements for {ref}`example-petsc-navier-stokes`
26- Refactored {ref}`example-petsc-navier-stokes` to improve code reuse
27- Added Shock Tube, Channel, and Flat Plate boundary layer problems to {ref}`example-petsc-navier-stokes`
28
29(v0-10-1)=
30
31## v0.10.1 (Apr 11, 2022)
32
33### Interface changes
34
35- Added {c:func}`CeedQFunctionSetUserFlopsEstimate` and {c:func}`CeedOperatorGetFlopsEstimate` to facilitate estimating FLOPs in operator application.
36
37### Bugfix
38
39- Install JiT source files in install directory to fix GPU functionality for installed libCEED.
40
41(v0-10)=
42
43## v0.10 (Mar 21, 2022)
44
45### Interface changes
46
47- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields.
48- 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`.
49- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added.
50- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients.
51- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces.
52- Warning added when compiling OCCA backend to alert users that this backend is experimental.
53- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`.
54- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name.
55- 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`.
56- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitalized arrays; require initalized data for {c:func}`CeedVectorGetArray`.
57- 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.
58- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retreive user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`.
59- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums.
60- Added type `CeedSize` equivalent to `ptrdiff_t` for array sizes in {c:func}`CeedVectorCreate`, {c:func}`CeedVectorGetLength`, `CeedElemRestrictionCreate*`, {c:func}`CeedElemRestrictionGetLVectorSize`, and {c:func}`CeedOperatorLinearAssembleSymbolic`. This is a breaking change.
61- Added {c:func}`CeedOperatorSetQFunctionUpdated` to facilitate QFunction data re-use between operators sharing the same quadrature space, such as in a multigrid hierarchy.
62- Added {c:func}`CeedOperatorGetActiveVectorLengths` to get shape of CeedOperator.
63
64### New features
65
66- `CeedScalar` can now be set as `float` or `double` at compile time.
67- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends.
68- 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.
69- Remove need to guard library headers in QFunction source for code generation backends.
70- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present.
71- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally.
72- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file.
73- Added support for non-tensor H(div) elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor.
74- Added {c:func}`CeedQFunctionSetContextWritable` and read-only access to `CeedQFunctionContext` data as an optional feature to improve GPU performance. By default, calling the `CeedQFunctionUser` during {c:func}`CeedQFunctionApply` is assumed to write into the `CeedQFunctionContext` data, consistent with the previous behavior. Note that if a user asserts that their `CeedQFunctionUser` does not write into the `CeedQFunctionContext` data, they are responsible for the validity of this assertion.
75- Added support for element matrix assembly in GPU backends.
76
77### Maintainability
78
79- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage.
80- `Include-what-you-use` makefile target added as `make iwyu`.
81- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase.
82- Put GPU JiTed kernel source code into separate files.
83- 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.
84
85(v0-9)=
86
87## v0.9 (Jul 6, 2021)
88
89### Interface changes
90
91- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact.
92
93### New features
94
95- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use.
96- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends.
97- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations.
98- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed).
99
100### Performance improvements
101
102### Examples
103
104- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations.
105- Fluid mechanics example adds GPU support and improves modularity.
106
107### Deprecated backends
108
109- 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.
110
111(v0-8)=
112
113## v0.8 (Mar 31, 2021)
114
115### Interface changes
116
117- Error handling improved to include enumerated error codes for C interface return values.
118- 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.
119
120### New features
121
122- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features.
123- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly.
124- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators.
125
126### Performance improvements
127
128- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`.
129- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`.
130
131### Examples
132
133- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions.
134- {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.
135- {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.
136- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree.
137
138(v0-7)=
139
140## v0.7 (Sep 29, 2020)
141
142### Interface changes
143
144- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors.
145  As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`.
146  These changes improve support for mixed finite element methods.
147- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature.
148- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity.
149- 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.
150- 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.
151- 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.
152- 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.
153  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`.
154- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory.
155- 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.
156
157### New features
158
159- New HIP backend: `/gpu/hip/ref`.
160- CeedQFunction support for user `CUfunction`s in some backends
161
162### Performance improvements
163
164- OCCA backend rebuilt to facilitate future performance enhancements.
165- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`.
166
167### Examples
168
169- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions.
170
171### Deprecated backends
172
173- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`.
174
175(v0-6)=
176
177## v0.6 (Mar 29, 2020)
178
179libCEED v0.6 contains numerous new features and examples, as well as expanded
180documentation in [this new website](https://libceed.org).
181
182### New features
183
184- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly
185  1-1 correspondence with the C interface, plus some convenience features.  For instance,
186  data stored in the {cpp:type}`CeedVector` structure are available without copy as
187  {py:class}`numpy.ndarray`.  Short tutorials are provided in
188  [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/).
189- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point,
190  {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal
191  ({c:func}`CeedOperatorAssembleLinearDiagonal`).  These operations are useful for
192  preconditioning ingredients and are used in the libCEED's multigrid examples.
193- The inverse of separable operators can be obtained using
194  {c:func}`CeedOperatorCreateFDMElementInverse` and applied with
195  {c:func}`CeedOperatorApply`.  This is a useful preconditioning ingredient,
196  especially for Laplacians and related operators.
197- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`,
198  {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`.
199- Make public accessors for various attributes to facilitate writing composable code.
200- New backend: `/cpu/self/memcheck/serial`.
201- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA
202  backends.  (Single source is coming soon for OCCA backends.)
203- Fix some missing edge cases in CUDA backend.
204
205### Performance Improvements
206
207- MAGMA backend performance optimization and non-tensor bases.
208- No-copy optimization in {c:func}`CeedOperatorApply`.
209
210### Interface changes
211
212- Replace {code}`CeedElemRestrictionCreateIdentity` and
213  {code}`CeedElemRestrictionCreateBlocked` with more flexible
214  {c:func}`CeedElemRestrictionCreateStrided` and
215  {c:func}`CeedElemRestrictionCreateBlockedStrided`.
216- Add arguments to {c:func}`CeedQFunctionCreateIdentity`.
217- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification
218  with {cpp:enum}`CeedInterlaceMode`.  This is now an attribute of the
219  {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no
220  longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and
221  {c:func}`CeedElemRestrictionApply`.
222
223### Examples
224
225libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`.
226Notable additions include:
227
228- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of
229  a domain in 1, 2, and 3 dimensions by applying a Laplacian.
230
231- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area
232  of domains (like the cube and sphere) by direct integration on a surface mesh;
233  demonstrates geometric dimension different from topological dimension.
234
235- PETSc {ref}`example-petsc-bps`:
236
237  - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support.
238  - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements
239    and transparent CUDA support.
240  - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`):
241    generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric
242    dimension different from topological dimension.
243
244- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid
245  solver with algebraic multigrid coarse solve.
246
247- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly
248  `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`),
249  implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and
250  quasi-2D computational domain support.
251
252- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for
253  linear elasticity, small-strain hyperelasticity, and globalized finite-strain
254  hyperelasticity using p-multigrid with algebraic multigrid coarse solve.
255
256(v0-5)=
257
258## v0.5 (Sep 18, 2019)
259
260For this release, several improvements were made. Two new CUDA backends were added to
261the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art
262performance using single-source {ref}`CeedQFunction`. From this release, users
263can define Q-Functions in a single source code independently of the targeted backend
264with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU
265compilation of the user provided {ref}`CeedQFunction` code. To allow a unified
266declaration, the {ref}`CeedQFunction` API has undergone a slight change:
267the `QFunctionField` parameter `ncomp` has been changed to `size`. This change
268requires setting the previous value of `ncomp` to `ncomp*dim` when adding a
269`QFunctionField` with eval mode `CEED EVAL GRAD`.
270
271Additionally, new CPU backends
272were included in this release, such as the `/cpu/self/opt/*` backends (which are
273written in pure C and use partial **E-vectors** to improve performance) and the
274`/cpu/self/ref/memcheck` backend (which relies upon the
275[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user
276{ref}`CeedQFunction` have no undefined values).
277This release also included various performance improvements, bug fixes, new examples,
278and improved tests. Among these improvements, vectorized instructions for
279{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD`
280instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and
281identity Q-Functions were introduced, and the PETSc benchmark problems were expanded
282to include unstructured meshes handling were. For this expansion, the prior version of
283the PETSc BPs, which only included data associated with structured geometries, were
284renamed `bpsraw`, and the new version of the BPs, which can handle data associated
285with any unstructured geometry, were called `bps`. Additionally, other benchmark
286problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively),
287and BP5 and BP6 (the collocated versions---for which the quadrature points are the same
288as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc
289examples. Furthermoew, another standalone libCEED example, called `ex2`, which
290computes the surface area of a given mesh was added to this release.
291
292Backends available in this release:
293
294| CEED resource (`-ceed`)  | Backend                                             |
295|--------------------------|-----------------------------------------------------|
296| `/cpu/self/ref/serial`   | Serial reference implementation                     |
297| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
298| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks            |
299| `/cpu/self/opt/serial`   | Serial optimized C implementation                   |
300| `/cpu/self/opt/blocked`  | Blocked optimized C implementation                  |
301| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
302| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
303| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
304| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
305| `/cpu/occa`              | Serial OCCA kernels                                 |
306| `/gpu/occa`              | CUDA OCCA kernels                                   |
307| `/omp/occa`              | OpenMP OCCA kernels                                 |
308| `/ocl/occa`              | OpenCL OCCA kernels                                 |
309| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
310| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
311| `/gpu/cuda/shared`       | Optimized pure CUDA kernels using shared memory     |
312| `/gpu/cuda/gen`          | Optimized pure CUDA kernels using code generation   |
313| `/gpu/magma`             | CUDA MAGMA kernels                                  |
314
315Examples available in this release:
316
317:::{list-table}
318:header-rows: 1
319:widths: auto
320* - User code
321  - Example
322* - `ceed`
323  - * ex1 (volume)
324    * ex2 (surface)
325* - `mfem`
326  - * BP1 (scalar mass operator)
327    * BP3 (scalar Laplace operator)
328* - `petsc`
329  - * BP1 (scalar mass operator)
330    * BP2 (vector mass operator)
331    * BP3 (scalar Laplace operator)
332    * BP4 (vector Laplace operator)
333    * BP5 (collocated scalar Laplace operator)
334    * BP6 (collocated vector Laplace operator)
335    * Navier-Stokes
336* - `nek5000`
337  - * BP1 (scalar mass operator)
338    * BP3 (scalar Laplace operator)
339:::
340
341(v0-4)=
342
343## v0.4 (Apr 1, 2019)
344
345libCEED v0.4 was made again publicly available in the second full CEED software
346distribution, release CEED 2.0. This release contained notable features, such as
347four new CPU backends, two new GPU backends, CPU backend optimizations, initial
348support for operator composition, performance benchmarking, and a Navier-Stokes demo.
349The new CPU backends in this release came in two families. The `/cpu/self/*/serial`
350backends process one element at a time and are intended for meshes with a smaller number
351of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of
352eight interlaced elements and are intended for meshes with higher numbers of elements.
353The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU
354performance. The `/cpu/self/xsmm/*` backends rely upon the
355[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU
356performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA.
357The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable
358performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple
359parallelization approach, where each thread treats a finite element. Using just in time
360compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this
361backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend
362achieve good peak performance for 1D, 2D, and low order 3D problems, but performance
363deteriorates very quickly when threads run out of registers.
364
365A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED
366examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`).
367This example solves the time-dependent Navier-Stokes equations of compressible gas
368dynamics in a static Eulerian three-dimensional frame, using structured high-order
369finite/spectral element spatial discretizations and explicit high-order time-stepping
370(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc,
371so that the pointwise physics (defined at quadrature points) is separated from the
372parallelization and meshing concerns.
373
374Backends available in this release:
375
376| CEED resource (`-ceed`)  | Backend                                             |
377|--------------------------|-----------------------------------------------------|
378| `/cpu/self/ref/serial`   | Serial reference implementation                     |
379| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
380| `/cpu/self/tmpl`         | Backend template, defaults to `/cpu/self/blocked`   |
381| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
382| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
383| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
384| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
385| `/cpu/occa`              | Serial OCCA kernels                                 |
386| `/gpu/occa`              | CUDA OCCA kernels                                   |
387| `/omp/occa`              | OpenMP OCCA kernels                                 |
388| `/ocl/occa`              | OpenCL OCCA kernels                                 |
389| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
390| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
391| `/gpu/magma`             | CUDA MAGMA kernels                                  |
392
393Examples available in this release:
394
395:::{list-table}
396:header-rows: 1
397:widths: auto
398* - User code
399  - Example
400* - `ceed`
401  - * ex1 (volume)
402* - `mfem`
403  - * BP1 (scalar mass operator)
404    * BP3 (scalar Laplace operator)
405* - `petsc`
406  - * BP1 (scalar mass operator)
407    * BP3 (scalar Laplace operator)
408    * Navier-Stokes
409* - `nek5000`
410  - * BP1 (scalar mass operator)
411    * BP3 (scalar Laplace operator)
412:::
413
414(v0-3)=
415
416## v0.3 (Sep 30, 2018)
417
418Notable features in this release include active/passive field interface, support for
419non-tensor bases, backend optimization, and improved Fortran interface. This release
420also focused on providing improved continuous integration, and many new tests with code
421coverage reports of about 90%. This release also provided a significant change to the
422public interface: a {ref}`CeedQFunction` can take any number of named input and output
423arguments while {ref}`CeedOperator` connects them to the actual data, which may be
424supplied explicitly to `CeedOperatorApply()` (active) or separately via
425`CeedOperatorSetField()` (passive). This interface change enables reusable libraries
426of CeedQFunctions and composition of block solvers constructed using
427{ref}`CeedOperator`. A concept of blocked restriction was added to this release and
428used in an optimized CPU backend. Although this is typically not visible to the user,
429it enables effective use of arbitrary-length SIMD while maintaining cache locality.
430This CPU backend also implements an algebraic factorization of tensor product gradients
431to perform fewer operations than standard application of interpolation and
432differentiation from nodes to quadrature points. This algebraic formulation
433automatically supports non-polynomial and non-interpolatory bases, thus is more general
434than the more common derivation in terms of Lagrange polynomials on the quadrature points.
435
436Backends available in this release:
437
438| CEED resource (`-ceed`) | Backend                                             |
439|-------------------------|-----------------------------------------------------|
440| `/cpu/self/blocked`     | Blocked reference implementation                    |
441| `/cpu/self/ref`         | Serial reference implementation                     |
442| `/cpu/self/tmpl`        | Backend template, defaults to `/cpu/self/blocked`   |
443| `/cpu/occa`             | Serial OCCA kernels                                 |
444| `/gpu/occa`             | CUDA OCCA kernels                                   |
445| `/omp/occa`             | OpenMP OCCA kernels                                 |
446| `/ocl/occa`             | OpenCL OCCA kernels                                 |
447| `/gpu/magma`            | CUDA MAGMA kernels                                  |
448
449Examples available in this release:
450
451:::{list-table}
452:header-rows: 1
453:widths: auto
454* - User code
455  - Example
456* - `ceed`
457  - * ex1 (volume)
458* - `mfem`
459  - * BP1 (scalar mass operator)
460    * BP3 (scalar Laplace operator)
461* - `petsc`
462  - * BP1 (scalar mass operator)
463    * BP3 (scalar Laplace operator)
464* - `nek5000`
465  - * BP1 (scalar mass operator)
466    * BP3 (scalar Laplace operator)
467:::
468
469(v0-21)=
470
471## v0.21 (Sep 30, 2018)
472
473A MAGMA backend (which relies upon the
474[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this
475release. This initial integration set up the framework of using MAGMA and provided the
476libCEED functionality through MAGMA kernels as one of libCEED’s computational backends.
477As any other backend, the MAGMA backend provides extended basic data structures for
478{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements
479the fundamental CEED building blocks to work with the new data structures.
480In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data
481but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled
482internally, and thus seamlessly to the user, through the functions/methods that are
483provided to support them.
484
485Backends available in this release:
486
487| CEED resource (`-ceed`) | Backend                         |
488|-------------------------|---------------------------------|
489| `/cpu/self`             | Serial reference implementation |
490| `/cpu/occa`             | Serial OCCA kernels             |
491| `/gpu/occa`             | CUDA OCCA kernels               |
492| `/omp/occa`             | OpenMP OCCA kernels             |
493| `/ocl/occa`             | OpenCL OCCA kernels             |
494| `/gpu/magma`            | CUDA MAGMA kernels              |
495
496Examples available in this release:
497
498:::{list-table}
499:header-rows: 1
500:widths: auto
501* - User code
502  - Example
503* - `ceed`
504  - * ex1 (volume)
505* - `mfem`
506  - * BP1 (scalar mass operator)
507    * BP3 (scalar Laplace operator)
508* - `petsc`
509  - * BP1 (scalar mass operator)
510* - `nek5000`
511  - * BP1 (scalar mass operator)
512:::
513
514(v0-2)=
515
516## v0.2 (Mar 30, 2018)
517
518libCEED was made publicly available the first full CEED software distribution, release
519CEED 1.0. The distribution was made available using the Spack package manager to provide
520a common, easy-to-use build environment, where the user can build the CEED distribution
521with all dependencies. This release included a new Fortran interface for the library.
522This release also contained major improvements in the OCCA backend (including a new
523`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to
524compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an
525`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application
526of the Laplace operator) was also added to this release.
527
528Backends available in this release:
529
530| CEED resource (`-ceed`) | Backend                         |
531|-------------------------|---------------------------------|
532| `/cpu/self`             | Serial reference implementation |
533| `/cpu/occa`             | Serial OCCA kernels             |
534| `/gpu/occa`             | CUDA OCCA kernels               |
535| `/omp/occa`             | OpenMP OCCA kernels             |
536| `/ocl/occa`             | OpenCL OCCA kernels             |
537
538Examples available in this release:
539
540:::{list-table}
541:header-rows: 1
542:widths: auto
543* - User code
544  - Example
545* - `ceed`
546  - * ex1 (volume)
547* - `mfem`
548  - * BP1 (scalar mass operator)
549    * BP3 (scalar Laplace operator)
550* - `petsc`
551  - * BP1 (scalar mass operator)
552* - `nek5000`
553  - * BP1 (scalar mass operator)
554:::
555
556(v0-1)=
557
558## v0.1 (Jan 3, 2018)
559
560Initial low-level API of the CEED project. The low-level API provides a set of Finite
561Elements kernels and components for writing new low-level kernels. Examples include:
562vector and sparse linear algebra, element matrix assembly over a batch of elements,
563partial assembly and action for efficient high-order operators like mass, diffusion,
564advection, etc. The main goal of the low-level API is to establish the basis for the
565high-level API. Also, identifying such low-level kernels and providing a reference
566implementation for them serves as the basis for specialized backend implementations.
567This release contained several backends: `/cpu/self`, and backends which rely upon the
568[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`,
569`/gpu/occa`, and `/omp/occa`.
570It also included several examples, in the `examples` folder:
571A standalone code that shows the usage of libCEED (with no external
572dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1
573(with the application of the mass operator); and a `petsc` example to perform BP1
574(with the application of the mass operator).
575
576Backends available in this release:
577
578| CEED resource (`-ceed`) | Backend                         |
579|-------------------------|---------------------------------|
580| `/cpu/self`             | Serial reference implementation |
581| `/cpu/occa`             | Serial OCCA kernels             |
582| `/gpu/occa`             | CUDA OCCA kernels               |
583| `/omp/occa`             | OpenMP OCCA kernels             |
584
585Examples available in this release:
586
587| User code             | Example                           |
588|-----------------------|-----------------------------------|
589| `ceed`                | ex1 (scalar Laplace operator)     |
590| `mfem`                | BP1 (scalar mass operator)        |
591| `petsc`               | BP1 (scalar mass operator)        |
592```
593