xref: /libCEED/doc/sphinx/source/releasenotes.md (revision ca7355308a14805110a7d98dabf1f658d5ec16d5)
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- Update `CeedOperatorContext*` functions to `CeedOperator*Context*` functions for consistency.
12For example, `CeedOperatorContextGetFieldLabel` was renamed to `CeedOperatorGetContextFieldLabel`.
13
14### New features
15
16- Added {c:func}`CeedOperatorGetFieldByName` to access a specific `CeedOperatorField` by its name.
17- Update `/cpu/self/memcheck/*` backends to help verify `CeedVector` array access assumptions and `CeedQFunction` user output assumptions.
18- Update {c:func}`CeedOperatorLinearAssembleDiagonal` to provide default implementation that supports `CeedOperator` with multiple active bases.
19- Added Sycl backends `/gpu/sycl/ref` and `/gpu/sycl/shared`.
20- Added {c:func}`CeedBasisApplyAtPoints` for evalution of values and derivaties at arbitrary points inside elements.
21- Added support for non-tensor $H(\text{curl})$ finite element spaces with {c:func}`CeedBasisCreateHcurl`.
22- 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)).
23
24### Examples
25
26#### {ref}`example-petsc-bps`
27
28- Requires PETSc version 3.19 or later.
29
30(v0-11)=
31
32## v0.11 (Dec 24, 2022)
33
34### Interface changes
35
36- Added {c:func}`CeedOperatorSetName` for more readable {c:func}`CeedOperatorView` output.
37- Added {c:func}`CeedBasisCreateProjection` to facilitate interpolation between nodes for separate `CeedBases`.
38- Rename and move {c:func}`CeedCompositeOperatorGetNumSub` and {c:func}`CeedCompositeOperatorGetSubList` to public interface.
39- Renamed `CEED_BASIS_COLLOCATED` to `CEED_BASIS_NONE` for clarity.
40Some 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.
41
42### New features
43
44- Update `/cpu/self/memcheck/*` backends to help verify `CeedQFunctionContext` data sizes provided by user.
45- Improved support for $H(\text{div})$ bases.
46- Added `CeedInt_FMT` to support potential future use of larger integer sizes.
47- Added `CEED_QFUNCTION_ATTR` for setting compiler attributes/pragmas to `CEED_QFUNCTION_HELPER` and `CEED_QFUNCTION`.
48- OCCA backend updated to latest OCCA release; DPC++ and OMP OCCA modes enabled.
49Due to a limitation of the OCCA parser, typedefs are required to use pointers to arrays in QFunctions with the OCCA backend.
50This issue will be fixed in a future OCCA release.
51
52### Bugfix
53
54- Fix bug in setting device id for GPU backends.
55- Fix storing of indices for `CeedElemRestriction` on the host with GPU backends.
56- Fix `CeedElemRestriction` sizing for {c:func}`CeedOperatorAssemblePointBlockDiagonal`.
57- Fix bugs in CPU implementation of {c:func}`CeedOperatorLinearAssemble` when there are different number of active input modes and active output modes.
58
59### Examples
60
61#### {ref}`example-petsc-navier-stokes`
62
63- Various performance enhancements, analytic matrix-free and assembled Jacobian, and PETSc solver configurations for GPUs.
64- Refactored to improve code reuse and modularity.
65- Support for primitive variables for more accurate boundary layers and all-speed flow.
66- Added $YZ\beta$ shock capturing scheme and Shock Tube example.
67- Added Channel example, with comparison to analytic solutions.
68- Added Flat Plate with boundary layer mesh and compressible Blasius inflow condition based on Chebyshev collocation solution of the Blasius equations.
69- Added strong and weak synthetic turbulence generation (STG) inflow boundary conditions.
70- Added "freestream" boundary conditions based on HLLC Riemann solver.
71- Automated stabilization coefficients for different basis degree.
72
73#### {ref}`example-petsc-bps`
74
75- Support for convergence studies.
76
77### Maintainability
78
79- 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.
80- Enabled support for `p > 8` for `/gpu/*/shared` backends.
81- Switch to `clang-format` over `astyle` for automatic formatting; Makefile command changed to `make format` from `make style`.
82- Improved test harness.
83
84(v0-10-1)=
85
86## v0.10.1 (Apr 11, 2022)
87
88### Interface changes
89
90- Added {c:func}`CeedQFunctionSetUserFlopsEstimate` and {c:func}`CeedOperatorGetFlopsEstimate` to facilitate estimating FLOPs in operator application.
91
92### New features
93
94- Switched MAGMA backends to use runtime compilation for tensor basis kernels (and element restriction kernels, in non-deterministic `/gpu/*/magma` backends).
95This reduces time to compile the library and increases the range of parameters for which the MAGMA tensor basis kernels will work.
96
97### Bugfix
98
99- Install JiT source files in install directory to fix GPU functionality for installed libCEED.
100
101(v0-10)=
102
103## v0.10 (Mar 21, 2022)
104
105### Interface changes
106
107- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields.
108- 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`.
109- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added.
110- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients.
111- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces.
112- Warning added when compiling OCCA backend to alert users that this backend is experimental.
113- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`.
114- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name.
115- 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`.
116- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitalized arrays; require initalized data for {c:func}`CeedVectorGetArray`.
117- 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.
118- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retreive user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`.
119- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums.
120- 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.
121- Added {c:func}`CeedOperatorSetQFunctionUpdated` to facilitate QFunction data re-use between operators sharing the same quadrature space, such as in a multigrid hierarchy.
122- Added {c:func}`CeedOperatorGetActiveVectorLengths` to get shape of CeedOperator.
123
124### New features
125
126- `CeedScalar` can now be set as `float` or `double` at compile time.
127- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends.
128- 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.
129- Remove need to guard library headers in QFunction source for code generation backends.
130- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present.
131- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally.
132- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file.
133- Added support for non-tensor $H(\text{div})$ elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor.
134- 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.
135- Added support for element matrix assembly in GPU backends.
136
137### Maintainability
138
139- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage.
140- `Include-what-you-use` makefile target added as `make iwyu`.
141- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase.
142- Put GPU JiTed kernel source code into separate files.
143- 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.
144
145(v0-9)=
146
147## v0.9 (Jul 6, 2021)
148
149### Interface changes
150
151- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact.
152
153### New features
154
155- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use.
156- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends.
157- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations.
158- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed).
159
160### Performance improvements
161
162### Examples
163
164- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations.
165- Fluid mechanics example adds GPU support and improves modularity.
166
167### Deprecated backends
168
169- 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.
170
171(v0-8)=
172
173## v0.8 (Mar 31, 2021)
174
175### Interface changes
176
177- Error handling improved to include enumerated error codes for C interface return values.
178- 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.
179
180### New features
181
182- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features.
183- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly.
184- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators.
185
186### Performance improvements
187
188- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`.
189- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`.
190
191### Examples
192
193- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions.
194- {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.
195- {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.
196- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree.
197
198(v0-7)=
199
200## v0.7 (Sep 29, 2020)
201
202### Interface changes
203
204- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors.
205  As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`.
206  These changes improve support for mixed finite element methods.
207- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature.
208- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity.
209- 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.
210- 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.
211- 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.
212- 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.
213  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`.
214- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory.
215- 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.
216
217### New features
218
219- New HIP backend: `/gpu/hip/ref`.
220- CeedQFunction support for user `CUfunction`s in some backends
221
222### Performance improvements
223
224- OCCA backend rebuilt to facilitate future performance enhancements.
225- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`.
226
227### Examples
228
229- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions.
230
231### Deprecated backends
232
233- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`.
234
235(v0-6)=
236
237## v0.6 (Mar 29, 2020)
238
239libCEED v0.6 contains numerous new features and examples, as well as expanded
240documentation in [this new website](https://libceed.org).
241
242### New features
243
244- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly
245  1-1 correspondence with the C interface, plus some convenience features.  For instance,
246  data stored in the {cpp:type}`CeedVector` structure are available without copy as
247  {py:class}`numpy.ndarray`.  Short tutorials are provided in
248  [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/).
249- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point,
250  {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal
251  ({c:func}`CeedOperatorAssembleLinearDiagonal`).  These operations are useful for
252  preconditioning ingredients and are used in the libCEED's multigrid examples.
253- The inverse of separable operators can be obtained using
254  {c:func}`CeedOperatorCreateFDMElementInverse` and applied with
255  {c:func}`CeedOperatorApply`.  This is a useful preconditioning ingredient,
256  especially for Laplacians and related operators.
257- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`,
258  {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`.
259- Make public accessors for various attributes to facilitate writing composable code.
260- New backend: `/cpu/self/memcheck/serial`.
261- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA
262  backends.  (Single source is coming soon for OCCA backends.)
263- Fix some missing edge cases in CUDA backend.
264
265### Performance Improvements
266
267- MAGMA backend performance optimization and non-tensor bases.
268- No-copy optimization in {c:func}`CeedOperatorApply`.
269
270### Interface changes
271
272- Replace {code}`CeedElemRestrictionCreateIdentity` and
273  {code}`CeedElemRestrictionCreateBlocked` with more flexible
274  {c:func}`CeedElemRestrictionCreateStrided` and
275  {c:func}`CeedElemRestrictionCreateBlockedStrided`.
276- Add arguments to {c:func}`CeedQFunctionCreateIdentity`.
277- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification
278  with {cpp:enum}`CeedInterlaceMode`.  This is now an attribute of the
279  {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no
280  longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and
281  {c:func}`CeedElemRestrictionApply`.
282
283### Examples
284
285libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`.
286Notable additions include:
287
288- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of
289  a domain in 1, 2, and 3 dimensions by applying a Laplacian.
290
291- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area
292  of domains (like the cube and sphere) by direct integration on a surface mesh;
293  demonstrates geometric dimension different from topological dimension.
294
295- PETSc {ref}`example-petsc-bps`:
296
297  - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support.
298  - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements
299    and transparent CUDA support.
300  - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`):
301    generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric
302    dimension different from topological dimension.
303
304- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid
305  solver with algebraic multigrid coarse solve.
306
307- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly
308  `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`),
309  implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and
310  quasi-2D computational domain support.
311
312- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for
313  linear elasticity, small-strain hyperelasticity, and globalized finite-strain
314  hyperelasticity using p-multigrid with algebraic multigrid coarse solve.
315
316(v0-5)=
317
318## v0.5 (Sep 18, 2019)
319
320For this release, several improvements were made. Two new CUDA backends were added to
321the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art
322performance using single-source {ref}`CeedQFunction`. From this release, users
323can define Q-Functions in a single source code independently of the targeted backend
324with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU
325compilation of the user provided {ref}`CeedQFunction` code. To allow a unified
326declaration, the {ref}`CeedQFunction` API has undergone a slight change:
327the `QFunctionField` parameter `ncomp` has been changed to `size`. This change
328requires setting the previous value of `ncomp` to `ncomp*dim` when adding a
329`QFunctionField` with eval mode `CEED EVAL GRAD`.
330
331Additionally, new CPU backends
332were included in this release, such as the `/cpu/self/opt/*` backends (which are
333written in pure C and use partial **E-vectors** to improve performance) and the
334`/cpu/self/ref/memcheck` backend (which relies upon the
335[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user
336{ref}`CeedQFunction` have no undefined values).
337This release also included various performance improvements, bug fixes, new examples,
338and improved tests. Among these improvements, vectorized instructions for
339{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD`
340instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and
341identity Q-Functions were introduced, and the PETSc benchmark problems were expanded
342to include unstructured meshes handling were. For this expansion, the prior version of
343the PETSc BPs, which only included data associated with structured geometries, were
344renamed `bpsraw`, and the new version of the BPs, which can handle data associated
345with any unstructured geometry, were called `bps`. Additionally, other benchmark
346problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively),
347and BP5 and BP6 (the collocated versions---for which the quadrature points are the same
348as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc
349examples. Furthermoew, another standalone libCEED example, called `ex2`, which
350computes the surface area of a given mesh was added to this release.
351
352Backends available in this release:
353
354| CEED resource (`-ceed`)  | Backend                                             |
355|--------------------------|-----------------------------------------------------|
356| `/cpu/self/ref/serial`   | Serial reference implementation                     |
357| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
358| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks            |
359| `/cpu/self/opt/serial`   | Serial optimized C implementation                   |
360| `/cpu/self/opt/blocked`  | Blocked optimized C implementation                  |
361| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
362| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
363| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
364| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
365| `/cpu/occa`              | Serial OCCA kernels                                 |
366| `/gpu/occa`              | CUDA OCCA kernels                                   |
367| `/omp/occa`              | OpenMP OCCA kernels                                 |
368| `/ocl/occa`              | OpenCL OCCA kernels                                 |
369| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
370| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
371| `/gpu/cuda/shared`       | Optimized pure CUDA kernels using shared memory     |
372| `/gpu/cuda/gen`          | Optimized pure CUDA kernels using code generation   |
373| `/gpu/magma`             | CUDA MAGMA kernels                                  |
374
375Examples available in this release:
376
377:::{list-table}
378:header-rows: 1
379:widths: auto
380* - User code
381  - Example
382* - `ceed`
383  - * ex1 (volume)
384    * ex2 (surface)
385* - `mfem`
386  - * BP1 (scalar mass operator)
387    * BP3 (scalar Laplace operator)
388* - `petsc`
389  - * BP1 (scalar mass operator)
390    * BP2 (vector mass operator)
391    * BP3 (scalar Laplace operator)
392    * BP4 (vector Laplace operator)
393    * BP5 (collocated scalar Laplace operator)
394    * BP6 (collocated vector Laplace operator)
395    * Navier-Stokes
396* - `nek5000`
397  - * BP1 (scalar mass operator)
398    * BP3 (scalar Laplace operator)
399:::
400
401(v0-4)=
402
403## v0.4 (Apr 1, 2019)
404
405libCEED v0.4 was made again publicly available in the second full CEED software
406distribution, release CEED 2.0. This release contained notable features, such as
407four new CPU backends, two new GPU backends, CPU backend optimizations, initial
408support for operator composition, performance benchmarking, and a Navier-Stokes demo.
409The new CPU backends in this release came in two families. The `/cpu/self/*/serial`
410backends process one element at a time and are intended for meshes with a smaller number
411of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of
412eight interlaced elements and are intended for meshes with higher numbers of elements.
413The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU
414performance. The `/cpu/self/xsmm/*` backends rely upon the
415[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU
416performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA.
417The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable
418performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple
419parallelization approach, where each thread treats a finite element. Using just in time
420compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this
421backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend
422achieve good peak performance for 1D, 2D, and low order 3D problems, but performance
423deteriorates very quickly when threads run out of registers.
424
425A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED
426examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`).
427This example solves the time-dependent Navier-Stokes equations of compressible gas
428dynamics in a static Eulerian three-dimensional frame, using structured high-order
429finite/spectral element spatial discretizations and explicit high-order time-stepping
430(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc,
431so that the pointwise physics (defined at quadrature points) is separated from the
432parallelization and meshing concerns.
433
434Backends available in this release:
435
436| CEED resource (`-ceed`)  | Backend                                             |
437|--------------------------|-----------------------------------------------------|
438| `/cpu/self/ref/serial`   | Serial reference implementation                     |
439| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
440| `/cpu/self/tmpl`         | Backend template, defaults to `/cpu/self/blocked`   |
441| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
442| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
443| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
444| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
445| `/cpu/occa`              | Serial OCCA kernels                                 |
446| `/gpu/occa`              | CUDA OCCA kernels                                   |
447| `/omp/occa`              | OpenMP OCCA kernels                                 |
448| `/ocl/occa`              | OpenCL OCCA kernels                                 |
449| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
450| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
451| `/gpu/magma`             | CUDA MAGMA kernels                                  |
452
453Examples available in this release:
454
455:::{list-table}
456:header-rows: 1
457:widths: auto
458* - User code
459  - Example
460* - `ceed`
461  - * ex1 (volume)
462* - `mfem`
463  - * BP1 (scalar mass operator)
464    * BP3 (scalar Laplace operator)
465* - `petsc`
466  - * BP1 (scalar mass operator)
467    * BP3 (scalar Laplace operator)
468    * Navier-Stokes
469* - `nek5000`
470  - * BP1 (scalar mass operator)
471    * BP3 (scalar Laplace operator)
472:::
473
474(v0-3)=
475
476## v0.3 (Sep 30, 2018)
477
478Notable features in this release include active/passive field interface, support for
479non-tensor bases, backend optimization, and improved Fortran interface. This release
480also focused on providing improved continuous integration, and many new tests with code
481coverage reports of about 90%. This release also provided a significant change to the
482public interface: a {ref}`CeedQFunction` can take any number of named input and output
483arguments while {ref}`CeedOperator` connects them to the actual data, which may be
484supplied explicitly to `CeedOperatorApply()` (active) or separately via
485`CeedOperatorSetField()` (passive). This interface change enables reusable libraries
486of CeedQFunctions and composition of block solvers constructed using
487{ref}`CeedOperator`. A concept of blocked restriction was added to this release and
488used in an optimized CPU backend. Although this is typically not visible to the user,
489it enables effective use of arbitrary-length SIMD while maintaining cache locality.
490This CPU backend also implements an algebraic factorization of tensor product gradients
491to perform fewer operations than standard application of interpolation and
492differentiation from nodes to quadrature points. This algebraic formulation
493automatically supports non-polynomial and non-interpolatory bases, thus is more general
494than the more common derivation in terms of Lagrange polynomials on the quadrature points.
495
496Backends available in this release:
497
498| CEED resource (`-ceed`) | Backend                                             |
499|-------------------------|-----------------------------------------------------|
500| `/cpu/self/blocked`     | Blocked reference implementation                    |
501| `/cpu/self/ref`         | Serial reference implementation                     |
502| `/cpu/self/tmpl`        | Backend template, defaults to `/cpu/self/blocked`   |
503| `/cpu/occa`             | Serial OCCA kernels                                 |
504| `/gpu/occa`             | CUDA OCCA kernels                                   |
505| `/omp/occa`             | OpenMP OCCA kernels                                 |
506| `/ocl/occa`             | OpenCL OCCA kernels                                 |
507| `/gpu/magma`            | CUDA MAGMA kernels                                  |
508
509Examples available in this release:
510
511:::{list-table}
512:header-rows: 1
513:widths: auto
514* - User code
515  - Example
516* - `ceed`
517  - * ex1 (volume)
518* - `mfem`
519  - * BP1 (scalar mass operator)
520    * BP3 (scalar Laplace operator)
521* - `petsc`
522  - * BP1 (scalar mass operator)
523    * BP3 (scalar Laplace operator)
524* - `nek5000`
525  - * BP1 (scalar mass operator)
526    * BP3 (scalar Laplace operator)
527:::
528
529(v0-21)=
530
531## v0.21 (Sep 30, 2018)
532
533A MAGMA backend (which relies upon the
534[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this
535release. This initial integration set up the framework of using MAGMA and provided the
536libCEED functionality through MAGMA kernels as one of libCEED’s computational backends.
537As any other backend, the MAGMA backend provides extended basic data structures for
538{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements
539the fundamental CEED building blocks to work with the new data structures.
540In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data
541but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled
542internally, and thus seamlessly to the user, through the functions/methods that are
543provided to support them.
544
545Backends available in this release:
546
547| CEED resource (`-ceed`) | Backend                         |
548|-------------------------|---------------------------------|
549| `/cpu/self`             | Serial reference implementation |
550| `/cpu/occa`             | Serial OCCA kernels             |
551| `/gpu/occa`             | CUDA OCCA kernels               |
552| `/omp/occa`             | OpenMP OCCA kernels             |
553| `/ocl/occa`             | OpenCL OCCA kernels             |
554| `/gpu/magma`            | CUDA MAGMA kernels              |
555
556Examples available in this release:
557
558:::{list-table}
559:header-rows: 1
560:widths: auto
561* - User code
562  - Example
563* - `ceed`
564  - * ex1 (volume)
565* - `mfem`
566  - * BP1 (scalar mass operator)
567    * BP3 (scalar Laplace operator)
568* - `petsc`
569  - * BP1 (scalar mass operator)
570* - `nek5000`
571  - * BP1 (scalar mass operator)
572:::
573
574(v0-2)=
575
576## v0.2 (Mar 30, 2018)
577
578libCEED was made publicly available the first full CEED software distribution, release
579CEED 1.0. The distribution was made available using the Spack package manager to provide
580a common, easy-to-use build environment, where the user can build the CEED distribution
581with all dependencies. This release included a new Fortran interface for the library.
582This release also contained major improvements in the OCCA backend (including a new
583`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to
584compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an
585`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application
586of the Laplace operator) was also added to this release.
587
588Backends available in this release:
589
590| CEED resource (`-ceed`) | Backend                         |
591|-------------------------|---------------------------------|
592| `/cpu/self`             | Serial reference implementation |
593| `/cpu/occa`             | Serial OCCA kernels             |
594| `/gpu/occa`             | CUDA OCCA kernels               |
595| `/omp/occa`             | OpenMP OCCA kernels             |
596| `/ocl/occa`             | OpenCL OCCA kernels             |
597
598Examples available in this release:
599
600:::{list-table}
601:header-rows: 1
602:widths: auto
603* - User code
604  - Example
605* - `ceed`
606  - * ex1 (volume)
607* - `mfem`
608  - * BP1 (scalar mass operator)
609    * BP3 (scalar Laplace operator)
610* - `petsc`
611  - * BP1 (scalar mass operator)
612* - `nek5000`
613  - * BP1 (scalar mass operator)
614:::
615
616(v0-1)=
617
618## v0.1 (Jan 3, 2018)
619
620Initial low-level API of the CEED project. The low-level API provides a set of Finite
621Elements kernels and components for writing new low-level kernels. Examples include:
622vector and sparse linear algebra, element matrix assembly over a batch of elements,
623partial assembly and action for efficient high-order operators like mass, diffusion,
624advection, etc. The main goal of the low-level API is to establish the basis for the
625high-level API. Also, identifying such low-level kernels and providing a reference
626implementation for them serves as the basis for specialized backend implementations.
627This release contained several backends: `/cpu/self`, and backends which rely upon the
628[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`,
629`/gpu/occa`, and `/omp/occa`.
630It also included several examples, in the `examples` folder:
631A standalone code that shows the usage of libCEED (with no external
632dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1
633(with the application of the mass operator); and a `petsc` example to perform BP1
634(with the application of the mass operator).
635
636Backends available in this release:
637
638| CEED resource (`-ceed`) | Backend                         |
639|-------------------------|---------------------------------|
640| `/cpu/self`             | Serial reference implementation |
641| `/cpu/occa`             | Serial OCCA kernels             |
642| `/gpu/occa`             | CUDA OCCA kernels               |
643| `/omp/occa`             | OpenMP OCCA kernels             |
644
645Examples available in this release:
646
647| User code             | Example                           |
648|-----------------------|-----------------------------------|
649| `ceed`                | ex1 (scalar Laplace operator)     |
650| `mfem`                | BP1 (scalar mass operator)        |
651| `petsc`               | BP1 (scalar mass operator)        |
652```
653