xref: /libCEED/doc/sphinx/source/releasenotes.md (revision 4bd6ffc97dc9a7688ef3a2d802aad5d41776eea1) 
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- Add `bool` field type for `CeedQFunctionContext` and related interfaces to use `bool` fields.
12- `CEED_BASIS_COLLOCATED` removed; users should only use `CEED_BASIS_NONE`.
13- Remove unneeded pointer for `CeedElemRestrictionGetELayout`.
14
15### New features
16
17- Add `CeedOperatorCreateAtPoints` which evaluates the `CeedQFunction` at arbitrary locations in each element, for use in Particle in Cell, Material Point Method, and similar methods.
18- Add `CeedElemRestrictionGetLLayout` to provide L-vector layout for strided `CeedElemRestriction` created with `CEED_BACKEND_STRIDES`.
19
20### Examples
21
22- Add deal.II example with CEED BP suite.
23
24(v0-12)=
25
26## v0.12 (Oct 31, 2023)
27
28### Interface changes
29
30- Update `CeedOperatorContext*` functions to `CeedOperator*Context*` functions for consistency.
31For example, `CeedOperatorContextGetFieldLabel` was renamed to `CeedOperatorGetContextFieldLabel`.
32- Removed `CeedBasisSetNumQuadraturePoints` as redundant and bug-prone interface.
33
34### New features
35
36- Added {c:func}`CeedOperatorGetFieldByName` to access a specific `CeedOperatorField` by its name.
37- Update `/cpu/self/memcheck/*` backends to help verify `CeedVector` array access assumptions and `CeedQFunction` user output assumptions.
38- Update {c:func}`CeedOperatorLinearAssembleDiagonal` to provide default implementation that supports `CeedOperator` with multiple active bases.
39- Added Sycl backends `/gpu/sycl/ref`, `/gpu/sycl/shared`, and `/gpu/sycl/gen`.
40- Added {c:func}`CeedBasisApplyAtPoints` for evaluation of values and derivatives at arbitrary points inside elements.
41- Added support for non-tensor $H(\text{curl})$ finite element spaces with {c:func}`CeedBasisCreateHcurl`.
42- Added {c:func}`CeedElemRestrictionCreateCurlOriented`, similar to {c:func}`CeedElemRestrictionCreateOriented`, for element restrictions requiring more general element transformations such as those for high-order $H(\text{curl})$ spaces on tetrahedra (see [https://dl.acm.org/doi/pdf/10.1145/3524456](https://dl.acm.org/doi/pdf/10.1145/3524456)).
43- Added {c:func}`CeedOperatorLinearAssemblePointBlockDiagonalSymbolic` to create COO mapping for mapping out of {c:func}`CeedOperatorLinearAssemblePointBlockDiagonal`.
44- Added support for application codes which manage multiple {ref}`Ceed` objects, parallelized across OpenMP threads.
45
46### Examples
47
48- Add `DMSwarm` example demonstrating interpolation from background mesh to swarm points and projection from swarm points to background mesh.
49
50#### {ref}`example-petsc-bps`
51
52- Requires PETSc version 3.19 or later.
53
54#### {ref}`example-petsc-navier-stokes`
55
56- Updated restart and checkpointing interface.
57- Add data-driven subgrid-stress model.
58- Add differential filtering of solution.
59- Add turbulence statistics collection over spanwise-symmetric geometries.
60- Add Taylor-Green vortex initial condition.
61- Add Riemann-based outflow boundary conditions.
62- Added vortex shedding and flow past cylinder example, including calculations for lift, drag, and heat transfer.
63- Add Internal Damping Layer (IDL) for helping turbulent simulation stability.
64- Derive `CeedBasis` from `PetscFE`, and various other internal maintainability updates.
65
66(v0-11)=
67
68## v0.11 (Dec 24, 2022)
69
70### Interface changes
71
72- Added {c:func}`CeedOperatorSetName` for more readable {c:func}`CeedOperatorView` output.
73- Added {c:func}`CeedBasisCreateProjection` to facilitate interpolation between nodes for separate `CeedBases`.
74- Rename and move {c:func}`CeedCompositeOperatorGetNumSub` and {c:func}`CeedCompositeOperatorGetSubList` to public interface.
75- Renamed `CEED_BASIS_COLLOCATED` to `CEED_BASIS_NONE` for clarity.
76Some users previously misinterpreted a `CeedOperator` field using `CEED_BASIS_COLLOCATED` as meaning that the entire `CeedOperator` used a quadrature space that is collocated with the nodal space of the active bases.
77
78### New features
79
80- Update `/cpu/self/memcheck/*` backends to help verify `CeedQFunctionContext` data sizes provided by user.
81- Improved support for $H(\text{div})$ bases.
82- Added `CeedInt_FMT` to support potential future use of larger integer sizes.
83- Added `CEED_QFUNCTION_ATTR` for setting compiler attributes/pragmas to `CEED_QFUNCTION_HELPER` and `CEED_QFUNCTION`.
84- OCCA backend updated to latest OCCA release; DPC++ and OMP OCCA modes enabled.
85Due to a limitation of the OCCA parser, typedefs are required to use pointers to arrays in QFunctions with the OCCA backend.
86This issue will be fixed in a future OCCA release.
87
88### Bugfix
89
90- Fix bug in setting device id for GPU backends.
91- Fix storing of indices for `CeedElemRestriction` on the host with GPU backends.
92- Fix `CeedElemRestriction` sizing for {c:func}`CeedOperatorAssemblePointBlockDiagonal`.
93- Fix bugs in CPU implementation of {c:func}`CeedOperatorLinearAssemble` when there are different number of active input modes and active output modes.
94
95### Examples
96
97#### {ref}`example-petsc-navier-stokes`
98
99- Various performance enhancements, analytic matrix-free and assembled Jacobian, and PETSc solver configurations for GPUs.
100- Refactored to improve code reuse and modularity.
101- Support for primitive variables for more accurate boundary layers and all-speed flow.
102- Added $YZ\beta$ shock capturing scheme and Shock Tube example.
103- Added Channel example, with comparison to analytic solutions.
104- Added Flat Plate with boundary layer mesh and compressible Blasius inflow condition based on Chebyshev collocation solution of the Blasius equations.
105- Added strong and weak synthetic turbulence generation (STG) inflow boundary conditions.
106- Added "freestream" boundary conditions based on HLLC Riemann solver.
107- Automated stabilization coefficients for different basis degree.
108
109#### {ref}`example-petsc-bps`
110
111- Support for convergence studies.
112
113### Maintainability
114
115- Refactored `/gpu/cuda/shared` and `/gpu/cuda/gen` as well as `/gpu/hip/shared` and `/gpu/hip/gen` backend to improve maintainablity and reduce duplicated code.
116- Enabled support for `p > 8` for `/gpu/*/shared` backends.
117- Switch to `clang-format` over `astyle` for automatic formatting; Makefile command changed to `make format` from `make style`.
118- Improved test harness.
119
120(v0-10-1)=
121
122## v0.10.1 (Apr 11, 2022)
123
124### Interface changes
125
126- Added {c:func}`CeedQFunctionSetUserFlopsEstimate` and {c:func}`CeedOperatorGetFlopsEstimate` to facilitate estimating FLOPs in operator application.
127
128### New features
129
130- Switched MAGMA backends to use runtime compilation for tensor basis kernels (and element restriction kernels, in non-deterministic `/gpu/*/magma` backends).
131This reduces time to compile the library and increases the range of parameters for which the MAGMA tensor basis kernels will work.
132
133### Bugfix
134
135- Install JiT source files in install directory to fix GPU functionality for installed libCEED.
136
137(v0-10)=
138
139## v0.10 (Mar 21, 2022)
140
141### Interface changes
142
143- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields.
144- 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`.
145- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added.
146- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients.
147- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces.
148- Warning added when compiling OCCA backend to alert users that this backend is experimental.
149- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`.
150- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name.
151- 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`.
152- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitialized arrays; require initialized data for {c:func}`CeedVectorGetArray`.
153- 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.
154- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retrieve user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`.
155- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums.
156- 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.
157- Added {c:func}`CeedOperatorSetQFunctionUpdated` to facilitate QFunction data re-use between operators sharing the same quadrature space, such as in a multigrid hierarchy.
158- Added {c:func}`CeedOperatorGetActiveVectorLengths` to get shape of CeedOperator.
159
160### New features
161
162- `CeedScalar` can now be set as `float` or `double` at compile time.
163- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends.
164- 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.
165- Remove need to guard library headers in QFunction source for code generation backends.
166- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present.
167- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally.
168- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file.
169- Added support for non-tensor $H(\text{div})$ elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor.
170- 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.
171- Added support for element matrix assembly in GPU backends.
172
173### Maintainability
174
175- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage.
176- `Include-what-you-use` makefile target added as `make iwyu`.
177- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase.
178- Put GPU JiTed kernel source code into separate files.
179- 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.
180
181(v0-9)=
182
183## v0.9 (Jul 6, 2021)
184
185### Interface changes
186
187- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact.
188
189### New features
190
191- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use.
192- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends.
193- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations.
194- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed).
195
196### Performance improvements
197
198### Examples
199
200- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations.
201- Fluid mechanics example adds GPU support and improves modularity.
202
203### Deprecated backends
204
205- 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.
206
207(v0-8)=
208
209## v0.8 (Mar 31, 2021)
210
211### Interface changes
212
213- Error handling improved to include enumerated error codes for C interface return values.
214- 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.
215
216### New features
217
218- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features.
219- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly.
220- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators.
221
222### Performance improvements
223
224- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`.
225- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`.
226
227### Examples
228
229- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions.
230- {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.
231- {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.
232- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree.
233
234(v0-7)=
235
236## v0.7 (Sep 29, 2020)
237
238### Interface changes
239
240- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors.
241  As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`.
242  These changes improve support for mixed finite element methods.
243- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature.
244- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity.
245- 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.
246- 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.
247- 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.
248- 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.
249  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`.
250- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory.
251- 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.
252
253### New features
254
255- New HIP backend: `/gpu/hip/ref`.
256- CeedQFunction support for user `CUfunction`s in some backends
257
258### Performance improvements
259
260- OCCA backend rebuilt to facilitate future performance enhancements.
261- PETSc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`.
262
263### Examples
264
265- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions.
266
267### Deprecated backends
268
269- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`.
270
271(v0-6)=
272
273## v0.6 (Mar 29, 2020)
274
275libCEED v0.6 contains numerous new features and examples, as well as expanded
276documentation in [this new website](https://libceed.org).
277
278### New features
279
280- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly
281  1-1 correspondence with the C interface, plus some convenience features.  For instance,
282  data stored in the {cpp:type}`CeedVector` structure are available without copy as
283  {py:class}`numpy.ndarray`.  Short tutorials are provided in
284  [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/).
285- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point,
286  {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal
287  ({c:func}`CeedOperatorAssembleLinearDiagonal`).  These operations are useful for
288  preconditioning ingredients and are used in the libCEED's multigrid examples.
289- The inverse of separable operators can be obtained using
290  {c:func}`CeedOperatorCreateFDMElementInverse` and applied with
291  {c:func}`CeedOperatorApply`.  This is a useful preconditioning ingredient,
292  especially for Laplacians and related operators.
293- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`,
294  {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`.
295- Make public accessors for various attributes to facilitate writing composable code.
296- New backend: `/cpu/self/memcheck/serial`.
297- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA
298  backends.  (Single source is coming soon for OCCA backends.)
299- Fix some missing edge cases in CUDA backend.
300
301### Performance Improvements
302
303- MAGMA backend performance optimization and non-tensor bases.
304- No-copy optimization in {c:func}`CeedOperatorApply`.
305
306### Interface changes
307
308- Replace {code}`CeedElemRestrictionCreateIdentity` and
309  {code}`CeedElemRestrictionCreateBlocked` with more flexible
310  {c:func}`CeedElemRestrictionCreateStrided` and
311  {c:func}`CeedElemRestrictionCreateBlockedStrided`.
312- Add arguments to {c:func}`CeedQFunctionCreateIdentity`.
313- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification
314  with {cpp:enum}`CeedInterlaceMode`.  This is now an attribute of the
315  {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no
316  longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and
317  {c:func}`CeedElemRestrictionApply`.
318
319### Examples
320
321libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`.
322Notable additions include:
323
324- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of
325  a domain in 1, 2, and 3 dimensions by applying a Laplacian.
326
327- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area
328  of domains (like the cube and sphere) by direct integration on a surface mesh;
329  demonstrates geometric dimension different from topological dimension.
330
331- PETSc {ref}`example-petsc-bps`:
332
333  - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support.
334  - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements
335    and transparent CUDA support.
336  - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`):
337    generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric
338    dimension different from topological dimension.
339
340- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid
341  solver with algebraic multigrid coarse solve.
342
343- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly
344  `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`),
345  implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and
346  quasi-2D computational domain support.
347
348- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for
349  linear elasticity, small-strain hyperelasticity, and globalized finite-strain
350  hyperelasticity using p-multigrid with algebraic multigrid coarse solve.
351
352(v0-5)=
353
354## v0.5 (Sep 18, 2019)
355
356For this release, several improvements were made. Two new CUDA backends were added to
357the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art
358performance using single-source {ref}`CeedQFunction`. From this release, users
359can define Q-Functions in a single source code independently of the targeted backend
360with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU
361compilation of the user provided {ref}`CeedQFunction` code. To allow a unified
362declaration, the {ref}`CeedQFunction` API has undergone a slight change:
363the `QFunctionField` parameter `ncomp` has been changed to `size`. This change
364requires setting the previous value of `ncomp` to `ncomp*dim` when adding a
365`QFunctionField` with eval mode `CEED EVAL GRAD`.
366
367Additionally, new CPU backends
368were included in this release, such as the `/cpu/self/opt/*` backends (which are
369written in pure C and use partial **E-vectors** to improve performance) and the
370`/cpu/self/ref/memcheck` backend (which relies upon the
371[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user
372{ref}`CeedQFunction` have no undefined values).
373This release also included various performance improvements, bug fixes, new examples,
374and improved tests. Among these improvements, vectorized instructions for
375{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD`
376instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and
377identity Q-Functions were introduced, and the PETSc benchmark problems were expanded
378to include unstructured meshes handling were. For this expansion, the prior version of
379the PETSc BPs, which only included data associated with structured geometries, were
380renamed `bpsraw`, and the new version of the BPs, which can handle data associated
381with any unstructured geometry, were called `bps`. Additionally, other benchmark
382problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively),
383and BP5 and BP6 (the collocated versions---for which the quadrature points are the same
384as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc
385examples. Furthermoew, another standalone libCEED example, called `ex2`, which
386computes the surface area of a given mesh was added to this release.
387
388Backends available in this release:
389
390| CEED resource (`-ceed`)  | Backend                                             |
391|--------------------------|-----------------------------------------------------|
392| `/cpu/self/ref/serial`   | Serial reference implementation                     |
393| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
394| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks            |
395| `/cpu/self/opt/serial`   | Serial optimized C implementation                   |
396| `/cpu/self/opt/blocked`  | Blocked optimized C implementation                  |
397| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
398| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
399| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
400| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
401| `/cpu/occa`              | Serial OCCA kernels                                 |
402| `/gpu/occa`              | CUDA OCCA kernels                                   |
403| `/omp/occa`              | OpenMP OCCA kernels                                 |
404| `/ocl/occa`              | OpenCL OCCA kernels                                 |
405| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
406| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
407| `/gpu/cuda/shared`       | Optimized pure CUDA kernels using shared memory     |
408| `/gpu/cuda/gen`          | Optimized pure CUDA kernels using code generation   |
409| `/gpu/magma`             | CUDA MAGMA kernels                                  |
410
411Examples available in this release:
412
413:::{list-table}
414:header-rows: 1
415:widths: auto
416* - User code
417  - Example
418* - `ceed`
419  - * ex1 (volume)
420    * ex2 (surface)
421* - `mfem`
422  - * BP1 (scalar mass operator)
423    * BP3 (scalar Laplace operator)
424* - `petsc`
425  - * BP1 (scalar mass operator)
426    * BP2 (vector mass operator)
427    * BP3 (scalar Laplace operator)
428    * BP4 (vector Laplace operator)
429    * BP5 (collocated scalar Laplace operator)
430    * BP6 (collocated vector Laplace operator)
431    * Navier-Stokes
432* - `nek5000`
433  - * BP1 (scalar mass operator)
434    * BP3 (scalar Laplace operator)
435:::
436
437(v0-4)=
438
439## v0.4 (Apr 1, 2019)
440
441libCEED v0.4 was made again publicly available in the second full CEED software
442distribution, release CEED 2.0. This release contained notable features, such as
443four new CPU backends, two new GPU backends, CPU backend optimizations, initial
444support for operator composition, performance benchmarking, and a Navier-Stokes demo.
445The new CPU backends in this release came in two families. The `/cpu/self/*/serial`
446backends process one element at a time and are intended for meshes with a smaller number
447of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of
448eight interlaced elements and are intended for meshes with higher numbers of elements.
449The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU
450performance. The `/cpu/self/xsmm/*` backends rely upon the
451[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU
452performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA.
453The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable
454performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple
455parallelization approach, where each thread treats a finite element. Using just in time
456compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this
457backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend
458achieve good peak performance for 1D, 2D, and low order 3D problems, but performance
459deteriorates very quickly when threads run out of registers.
460
461A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED
462examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`).
463This example solves the time-dependent Navier-Stokes equations of compressible gas
464dynamics in a static Eulerian three-dimensional frame, using structured high-order
465finite/spectral element spatial discretizations and explicit high-order time-stepping
466(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc,
467so that the pointwise physics (defined at quadrature points) is separated from the
468parallelization and meshing concerns.
469
470Backends available in this release:
471
472| CEED resource (`-ceed`)  | Backend                                             |
473|--------------------------|-----------------------------------------------------|
474| `/cpu/self/ref/serial`   | Serial reference implementation                     |
475| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
476| `/cpu/self/tmpl`         | Backend template, defaults to `/cpu/self/blocked`   |
477| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
478| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
479| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
480| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
481| `/cpu/occa`              | Serial OCCA kernels                                 |
482| `/gpu/occa`              | CUDA OCCA kernels                                   |
483| `/omp/occa`              | OpenMP OCCA kernels                                 |
484| `/ocl/occa`              | OpenCL OCCA kernels                                 |
485| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
486| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
487| `/gpu/magma`             | CUDA MAGMA kernels                                  |
488
489Examples available in this release:
490
491:::{list-table}
492:header-rows: 1
493:widths: auto
494* - User code
495  - Example
496* - `ceed`
497  - * ex1 (volume)
498* - `mfem`
499  - * BP1 (scalar mass operator)
500    * BP3 (scalar Laplace operator)
501* - `petsc`
502  - * BP1 (scalar mass operator)
503    * BP3 (scalar Laplace operator)
504    * Navier-Stokes
505* - `nek5000`
506  - * BP1 (scalar mass operator)
507    * BP3 (scalar Laplace operator)
508:::
509
510(v0-3)=
511
512## v0.3 (Sep 30, 2018)
513
514Notable features in this release include active/passive field interface, support for
515non-tensor bases, backend optimization, and improved Fortran interface. This release
516also focused on providing improved continuous integration, and many new tests with code
517coverage reports of about 90%. This release also provided a significant change to the
518public interface: a {ref}`CeedQFunction` can take any number of named input and output
519arguments while {ref}`CeedOperator` connects them to the actual data, which may be
520supplied explicitly to `CeedOperatorApply()` (active) or separately via
521`CeedOperatorSetField()` (passive). This interface change enables reusable libraries
522of CeedQFunctions and composition of block solvers constructed using
523{ref}`CeedOperator`. A concept of blocked restriction was added to this release and
524used in an optimized CPU backend. Although this is typically not visible to the user,
525it enables effective use of arbitrary-length SIMD while maintaining cache locality.
526This CPU backend also implements an algebraic factorization of tensor product gradients
527to perform fewer operations than standard application of interpolation and
528differentiation from nodes to quadrature points. This algebraic formulation
529automatically supports non-polynomial and non-interpolatory bases, thus is more general
530than the more common derivation in terms of Lagrange polynomials on the quadrature points.
531
532Backends available in this release:
533
534| CEED resource (`-ceed`) | Backend                                             |
535|-------------------------|-----------------------------------------------------|
536| `/cpu/self/blocked`     | Blocked reference implementation                    |
537| `/cpu/self/ref`         | Serial reference implementation                     |
538| `/cpu/self/tmpl`        | Backend template, defaults to `/cpu/self/blocked`   |
539| `/cpu/occa`             | Serial OCCA kernels                                 |
540| `/gpu/occa`             | CUDA OCCA kernels                                   |
541| `/omp/occa`             | OpenMP OCCA kernels                                 |
542| `/ocl/occa`             | OpenCL OCCA kernels                                 |
543| `/gpu/magma`            | CUDA MAGMA kernels                                  |
544
545Examples available in this release:
546
547:::{list-table}
548:header-rows: 1
549:widths: auto
550* - User code
551  - Example
552* - `ceed`
553  - * ex1 (volume)
554* - `mfem`
555  - * BP1 (scalar mass operator)
556    * BP3 (scalar Laplace operator)
557* - `petsc`
558  - * BP1 (scalar mass operator)
559    * BP3 (scalar Laplace operator)
560* - `nek5000`
561  - * BP1 (scalar mass operator)
562    * BP3 (scalar Laplace operator)
563:::
564
565(v0-21)=
566
567## v0.21 (Sep 30, 2018)
568
569A MAGMA backend (which relies upon the
570[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this
571release. This initial integration set up the framework of using MAGMA and provided the
572libCEED functionality through MAGMA kernels as one of libCEED’s computational backends.
573As any other backend, the MAGMA backend provides extended basic data structures for
574{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements
575the fundamental CEED building blocks to work with the new data structures.
576In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data
577but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled
578internally, and thus seamlessly to the user, through the functions/methods that are
579provided to support them.
580
581Backends available in this release:
582
583| CEED resource (`-ceed`) | Backend                         |
584|-------------------------|---------------------------------|
585| `/cpu/self`             | Serial reference implementation |
586| `/cpu/occa`             | Serial OCCA kernels             |
587| `/gpu/occa`             | CUDA OCCA kernels               |
588| `/omp/occa`             | OpenMP OCCA kernels             |
589| `/ocl/occa`             | OpenCL OCCA kernels             |
590| `/gpu/magma`            | CUDA MAGMA kernels              |
591
592Examples available in this release:
593
594:::{list-table}
595:header-rows: 1
596:widths: auto
597* - User code
598  - Example
599* - `ceed`
600  - * ex1 (volume)
601* - `mfem`
602  - * BP1 (scalar mass operator)
603    * BP3 (scalar Laplace operator)
604* - `petsc`
605  - * BP1 (scalar mass operator)
606* - `nek5000`
607  - * BP1 (scalar mass operator)
608:::
609
610(v0-2)=
611
612## v0.2 (Mar 30, 2018)
613
614libCEED was made publicly available the first full CEED software distribution, release
615CEED 1.0. The distribution was made available using the Spack package manager to provide
616a common, easy-to-use build environment, where the user can build the CEED distribution
617with all dependencies. This release included a new Fortran interface for the library.
618This release also contained major improvements in the OCCA backend (including a new
619`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to
620compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an
621`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application
622of the Laplace operator) was also added to this release.
623
624Backends available in this release:
625
626| CEED resource (`-ceed`) | Backend                         |
627|-------------------------|---------------------------------|
628| `/cpu/self`             | Serial reference implementation |
629| `/cpu/occa`             | Serial OCCA kernels             |
630| `/gpu/occa`             | CUDA OCCA kernels               |
631| `/omp/occa`             | OpenMP OCCA kernels             |
632| `/ocl/occa`             | OpenCL OCCA kernels             |
633
634Examples available in this release:
635
636:::{list-table}
637:header-rows: 1
638:widths: auto
639* - User code
640  - Example
641* - `ceed`
642  - * ex1 (volume)
643* - `mfem`
644  - * BP1 (scalar mass operator)
645    * BP3 (scalar Laplace operator)
646* - `petsc`
647  - * BP1 (scalar mass operator)
648* - `nek5000`
649  - * BP1 (scalar mass operator)
650:::
651
652(v0-1)=
653
654## v0.1 (Jan 3, 2018)
655
656Initial low-level API of the CEED project. The low-level API provides a set of Finite
657Elements kernels and components for writing new low-level kernels. Examples include:
658vector and sparse linear algebra, element matrix assembly over a batch of elements,
659partial assembly and action for efficient high-order operators like mass, diffusion,
660advection, etc. The main goal of the low-level API is to establish the basis for the
661high-level API. Also, identifying such low-level kernels and providing a reference
662implementation for them serves as the basis for specialized backend implementations.
663This release contained several backends: `/cpu/self`, and backends which rely upon the
664[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`,
665`/gpu/occa`, and `/omp/occa`.
666It also included several examples, in the `examples` folder:
667A standalone code that shows the usage of libCEED (with no external
668dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1
669(with the application of the mass operator); and a `petsc` example to perform BP1
670(with the application of the mass operator).
671
672Backends available in this release:
673
674| CEED resource (`-ceed`) | Backend                         |
675|-------------------------|---------------------------------|
676| `/cpu/self`             | Serial reference implementation |
677| `/cpu/occa`             | Serial OCCA kernels             |
678| `/gpu/occa`             | CUDA OCCA kernels               |
679| `/omp/occa`             | OpenMP OCCA kernels             |
680
681Examples available in this release:
682
683| User code             | Example                           |
684|-----------------------|-----------------------------------|
685| `ceed`                | ex1 (scalar Laplace operator)     |
686| `mfem`                | BP1 (scalar mass operator)        |
687| `petsc`               | BP1 (scalar mass operator)        |
688```
689