xref: /libCEED/doc/sphinx/source/releasenotes.md (revision bb229da952f7e9779ba6cb3cd1ca2ebeac5feb1f)
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
40### New features
41
42- Update `/cpu/self/memcheck/*` backends to help verify `CeedQFunctionContext` data sizes provided by user.
43- Improved support for $H(\text{div})$ bases.
44- Added `CeedInt_FMT` to support potential future use of larger integer sizes.
45- Added `CEED_QFUNCTION_ATTR` for setting compiler attributes/pragmas to `CEED_QFUNCTION_HELPER` and `CEED_QFUNCTION`.
46- OCCA backend updated to latest OCCA release; DPC++ and OMP OCCA modes enabled.
47Due to a limitation of the OCCA parser, typedefs are required to use pointers to arrays in QFunctions with the OCCA backend.
48This issue will be fixed in a future OCCA release.
49
50### Bugfix
51
52- Fix bug in setting device id for GPU backends.
53- Fix storing of indices for `CeedElemRestriction` on the host with GPU backends.
54- Fix `CeedElemRestriction` sizing for {c:func}`CeedOperatorAssemblePointBlockDiagonal`.
55- Fix bugs in CPU implementation of {c:func}`CeedOperatorLinearAssemble` when there are different number of active input modes and active output modes.
56
57### Examples
58
59#### {ref}`example-petsc-navier-stokes`
60
61- Various performance enhancements, analytic matrix-free and assembled Jacobian, and PETSc solver configurations for GPUs.
62- Refactored to improve code reuse and modularity.
63- Support for primitive variables for more accurate boundary layers and all-speed flow.
64- Added $YZ\beta$ shock capturing scheme and Shock Tube example.
65- Added Channel example, with comparison to analytic solutions.
66- Added Flat Plate with boundary layer mesh and compressible Blasius inflow condition based on Chebyshev collocation solution of the Blasius equations.
67- Added strong and weak synthetic turbulence generation (STG) inflow boundary conditions.
68- Added "freestream" boundary conditions based on HLLC Riemann solver.
69- Automated stabilization coefficients for different basis degree.
70
71#### {ref}`example-petsc-bps`
72
73- Support for convergence studies.
74
75### Maintainability
76
77- 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.
78- Enabled support for `p > 8` for `/gpu/*/shared` backends.
79- Switch to `clang-format` over `astyle` for automatic formatting; Makefile command changed to `make format` from `make style`.
80- Improved test harness.
81
82(v0-10-1)=
83
84## v0.10.1 (Apr 11, 2022)
85
86### Interface changes
87
88- Added {c:func}`CeedQFunctionSetUserFlopsEstimate` and {c:func}`CeedOperatorGetFlopsEstimate` to facilitate estimating FLOPs in operator application.
89
90### New features
91
92- Switched MAGMA backends to use runtime compilation for tensor basis kernels (and element restriction kernels, in non-deterministic `/gpu/*/magma` backends).
93This reduces time to compile the library and increases the range of parameters for which the MAGMA tensor basis kernels will work.
94
95### Bugfix
96
97- Install JiT source files in install directory to fix GPU functionality for installed libCEED.
98
99(v0-10)=
100
101## v0.10 (Mar 21, 2022)
102
103### Interface changes
104
105- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields.
106- 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`.
107- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added.
108- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients.
109- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces.
110- Warning added when compiling OCCA backend to alert users that this backend is experimental.
111- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`.
112- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name.
113- 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`.
114- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitalized arrays; require initalized data for {c:func}`CeedVectorGetArray`.
115- 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.
116- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retreive user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`.
117- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums.
118- 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.
119- Added {c:func}`CeedOperatorSetQFunctionUpdated` to facilitate QFunction data re-use between operators sharing the same quadrature space, such as in a multigrid hierarchy.
120- Added {c:func}`CeedOperatorGetActiveVectorLengths` to get shape of CeedOperator.
121
122### New features
123
124- `CeedScalar` can now be set as `float` or `double` at compile time.
125- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends.
126- 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.
127- Remove need to guard library headers in QFunction source for code generation backends.
128- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present.
129- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally.
130- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file.
131- Added support for non-tensor $H(\text{div})$ elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor.
132- 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.
133- Added support for element matrix assembly in GPU backends.
134
135### Maintainability
136
137- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage.
138- `Include-what-you-use` makefile target added as `make iwyu`.
139- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase.
140- Put GPU JiTed kernel source code into separate files.
141- 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.
142
143(v0-9)=
144
145## v0.9 (Jul 6, 2021)
146
147### Interface changes
148
149- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact.
150
151### New features
152
153- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use.
154- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends.
155- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations.
156- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed).
157
158### Performance improvements
159
160### Examples
161
162- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations.
163- Fluid mechanics example adds GPU support and improves modularity.
164
165### Deprecated backends
166
167- 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.
168
169(v0-8)=
170
171## v0.8 (Mar 31, 2021)
172
173### Interface changes
174
175- Error handling improved to include enumerated error codes for C interface return values.
176- 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.
177
178### New features
179
180- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features.
181- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly.
182- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators.
183
184### Performance improvements
185
186- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`.
187- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`.
188
189### Examples
190
191- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions.
192- {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.
193- {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.
194- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree.
195
196(v0-7)=
197
198## v0.7 (Sep 29, 2020)
199
200### Interface changes
201
202- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors.
203  As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`.
204  These changes improve support for mixed finite element methods.
205- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature.
206- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity.
207- 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.
208- 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.
209- 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.
210- 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.
211  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`.
212- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory.
213- 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.
214
215### New features
216
217- New HIP backend: `/gpu/hip/ref`.
218- CeedQFunction support for user `CUfunction`s in some backends
219
220### Performance improvements
221
222- OCCA backend rebuilt to facilitate future performance enhancements.
223- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`.
224
225### Examples
226
227- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions.
228
229### Deprecated backends
230
231- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`.
232
233(v0-6)=
234
235## v0.6 (Mar 29, 2020)
236
237libCEED v0.6 contains numerous new features and examples, as well as expanded
238documentation in [this new website](https://libceed.org).
239
240### New features
241
242- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly
243  1-1 correspondence with the C interface, plus some convenience features.  For instance,
244  data stored in the {cpp:type}`CeedVector` structure are available without copy as
245  {py:class}`numpy.ndarray`.  Short tutorials are provided in
246  [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/).
247- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point,
248  {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal
249  ({c:func}`CeedOperatorAssembleLinearDiagonal`).  These operations are useful for
250  preconditioning ingredients and are used in the libCEED's multigrid examples.
251- The inverse of separable operators can be obtained using
252  {c:func}`CeedOperatorCreateFDMElementInverse` and applied with
253  {c:func}`CeedOperatorApply`.  This is a useful preconditioning ingredient,
254  especially for Laplacians and related operators.
255- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`,
256  {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`.
257- Make public accessors for various attributes to facilitate writing composable code.
258- New backend: `/cpu/self/memcheck/serial`.
259- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA
260  backends.  (Single source is coming soon for OCCA backends.)
261- Fix some missing edge cases in CUDA backend.
262
263### Performance Improvements
264
265- MAGMA backend performance optimization and non-tensor bases.
266- No-copy optimization in {c:func}`CeedOperatorApply`.
267
268### Interface changes
269
270- Replace {code}`CeedElemRestrictionCreateIdentity` and
271  {code}`CeedElemRestrictionCreateBlocked` with more flexible
272  {c:func}`CeedElemRestrictionCreateStrided` and
273  {c:func}`CeedElemRestrictionCreateBlockedStrided`.
274- Add arguments to {c:func}`CeedQFunctionCreateIdentity`.
275- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification
276  with {cpp:enum}`CeedInterlaceMode`.  This is now an attribute of the
277  {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no
278  longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and
279  {c:func}`CeedElemRestrictionApply`.
280
281### Examples
282
283libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`.
284Notable additions include:
285
286- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of
287  a domain in 1, 2, and 3 dimensions by applying a Laplacian.
288
289- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area
290  of domains (like the cube and sphere) by direct integration on a surface mesh;
291  demonstrates geometric dimension different from topological dimension.
292
293- PETSc {ref}`example-petsc-bps`:
294
295  - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support.
296  - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements
297    and transparent CUDA support.
298  - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`):
299    generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric
300    dimension different from topological dimension.
301
302- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid
303  solver with algebraic multigrid coarse solve.
304
305- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly
306  `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`),
307  implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and
308  quasi-2D computational domain support.
309
310- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for
311  linear elasticity, small-strain hyperelasticity, and globalized finite-strain
312  hyperelasticity using p-multigrid with algebraic multigrid coarse solve.
313
314(v0-5)=
315
316## v0.5 (Sep 18, 2019)
317
318For this release, several improvements were made. Two new CUDA backends were added to
319the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art
320performance using single-source {ref}`CeedQFunction`. From this release, users
321can define Q-Functions in a single source code independently of the targeted backend
322with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU
323compilation of the user provided {ref}`CeedQFunction` code. To allow a unified
324declaration, the {ref}`CeedQFunction` API has undergone a slight change:
325the `QFunctionField` parameter `ncomp` has been changed to `size`. This change
326requires setting the previous value of `ncomp` to `ncomp*dim` when adding a
327`QFunctionField` with eval mode `CEED EVAL GRAD`.
328
329Additionally, new CPU backends
330were included in this release, such as the `/cpu/self/opt/*` backends (which are
331written in pure C and use partial **E-vectors** to improve performance) and the
332`/cpu/self/ref/memcheck` backend (which relies upon the
333[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user
334{ref}`CeedQFunction` have no undefined values).
335This release also included various performance improvements, bug fixes, new examples,
336and improved tests. Among these improvements, vectorized instructions for
337{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD`
338instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and
339identity Q-Functions were introduced, and the PETSc benchmark problems were expanded
340to include unstructured meshes handling were. For this expansion, the prior version of
341the PETSc BPs, which only included data associated with structured geometries, were
342renamed `bpsraw`, and the new version of the BPs, which can handle data associated
343with any unstructured geometry, were called `bps`. Additionally, other benchmark
344problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively),
345and BP5 and BP6 (the collocated versions---for which the quadrature points are the same
346as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc
347examples. Furthermoew, another standalone libCEED example, called `ex2`, which
348computes the surface area of a given mesh was added to this release.
349
350Backends available in this release:
351
352| CEED resource (`-ceed`)  | Backend                                             |
353|--------------------------|-----------------------------------------------------|
354| `/cpu/self/ref/serial`   | Serial reference implementation                     |
355| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
356| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks            |
357| `/cpu/self/opt/serial`   | Serial optimized C implementation                   |
358| `/cpu/self/opt/blocked`  | Blocked optimized C implementation                  |
359| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
360| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
361| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
362| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
363| `/cpu/occa`              | Serial OCCA kernels                                 |
364| `/gpu/occa`              | CUDA OCCA kernels                                   |
365| `/omp/occa`              | OpenMP OCCA kernels                                 |
366| `/ocl/occa`              | OpenCL OCCA kernels                                 |
367| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
368| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
369| `/gpu/cuda/shared`       | Optimized pure CUDA kernels using shared memory     |
370| `/gpu/cuda/gen`          | Optimized pure CUDA kernels using code generation   |
371| `/gpu/magma`             | CUDA MAGMA kernels                                  |
372
373Examples available in this release:
374
375:::{list-table}
376:header-rows: 1
377:widths: auto
378* - User code
379  - Example
380* - `ceed`
381  - * ex1 (volume)
382    * ex2 (surface)
383* - `mfem`
384  - * BP1 (scalar mass operator)
385    * BP3 (scalar Laplace operator)
386* - `petsc`
387  - * BP1 (scalar mass operator)
388    * BP2 (vector mass operator)
389    * BP3 (scalar Laplace operator)
390    * BP4 (vector Laplace operator)
391    * BP5 (collocated scalar Laplace operator)
392    * BP6 (collocated vector Laplace operator)
393    * Navier-Stokes
394* - `nek5000`
395  - * BP1 (scalar mass operator)
396    * BP3 (scalar Laplace operator)
397:::
398
399(v0-4)=
400
401## v0.4 (Apr 1, 2019)
402
403libCEED v0.4 was made again publicly available in the second full CEED software
404distribution, release CEED 2.0. This release contained notable features, such as
405four new CPU backends, two new GPU backends, CPU backend optimizations, initial
406support for operator composition, performance benchmarking, and a Navier-Stokes demo.
407The new CPU backends in this release came in two families. The `/cpu/self/*/serial`
408backends process one element at a time and are intended for meshes with a smaller number
409of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of
410eight interlaced elements and are intended for meshes with higher numbers of elements.
411The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU
412performance. The `/cpu/self/xsmm/*` backends rely upon the
413[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU
414performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA.
415The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable
416performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple
417parallelization approach, where each thread treats a finite element. Using just in time
418compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this
419backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend
420achieve good peak performance for 1D, 2D, and low order 3D problems, but performance
421deteriorates very quickly when threads run out of registers.
422
423A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED
424examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`).
425This example solves the time-dependent Navier-Stokes equations of compressible gas
426dynamics in a static Eulerian three-dimensional frame, using structured high-order
427finite/spectral element spatial discretizations and explicit high-order time-stepping
428(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc,
429so that the pointwise physics (defined at quadrature points) is separated from the
430parallelization and meshing concerns.
431
432Backends available in this release:
433
434| CEED resource (`-ceed`)  | Backend                                             |
435|--------------------------|-----------------------------------------------------|
436| `/cpu/self/ref/serial`   | Serial reference implementation                     |
437| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
438| `/cpu/self/tmpl`         | Backend template, defaults to `/cpu/self/blocked`   |
439| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
440| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
441| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
442| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
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/cuda/ref`          | Reference pure CUDA kernels                         |
448| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
449| `/gpu/magma`             | CUDA MAGMA kernels                                  |
450
451Examples available in this release:
452
453:::{list-table}
454:header-rows: 1
455:widths: auto
456* - User code
457  - Example
458* - `ceed`
459  - * ex1 (volume)
460* - `mfem`
461  - * BP1 (scalar mass operator)
462    * BP3 (scalar Laplace operator)
463* - `petsc`
464  - * BP1 (scalar mass operator)
465    * BP3 (scalar Laplace operator)
466    * Navier-Stokes
467* - `nek5000`
468  - * BP1 (scalar mass operator)
469    * BP3 (scalar Laplace operator)
470:::
471
472(v0-3)=
473
474## v0.3 (Sep 30, 2018)
475
476Notable features in this release include active/passive field interface, support for
477non-tensor bases, backend optimization, and improved Fortran interface. This release
478also focused on providing improved continuous integration, and many new tests with code
479coverage reports of about 90%. This release also provided a significant change to the
480public interface: a {ref}`CeedQFunction` can take any number of named input and output
481arguments while {ref}`CeedOperator` connects them to the actual data, which may be
482supplied explicitly to `CeedOperatorApply()` (active) or separately via
483`CeedOperatorSetField()` (passive). This interface change enables reusable libraries
484of CeedQFunctions and composition of block solvers constructed using
485{ref}`CeedOperator`. A concept of blocked restriction was added to this release and
486used in an optimized CPU backend. Although this is typically not visible to the user,
487it enables effective use of arbitrary-length SIMD while maintaining cache locality.
488This CPU backend also implements an algebraic factorization of tensor product gradients
489to perform fewer operations than standard application of interpolation and
490differentiation from nodes to quadrature points. This algebraic formulation
491automatically supports non-polynomial and non-interpolatory bases, thus is more general
492than the more common derivation in terms of Lagrange polynomials on the quadrature points.
493
494Backends available in this release:
495
496| CEED resource (`-ceed`) | Backend                                             |
497|-------------------------|-----------------------------------------------------|
498| `/cpu/self/blocked`     | Blocked reference implementation                    |
499| `/cpu/self/ref`         | Serial reference implementation                     |
500| `/cpu/self/tmpl`        | Backend template, defaults to `/cpu/self/blocked`   |
501| `/cpu/occa`             | Serial OCCA kernels                                 |
502| `/gpu/occa`             | CUDA OCCA kernels                                   |
503| `/omp/occa`             | OpenMP OCCA kernels                                 |
504| `/ocl/occa`             | OpenCL OCCA kernels                                 |
505| `/gpu/magma`            | CUDA MAGMA kernels                                  |
506
507Examples available in this release:
508
509:::{list-table}
510:header-rows: 1
511:widths: auto
512* - User code
513  - Example
514* - `ceed`
515  - * ex1 (volume)
516* - `mfem`
517  - * BP1 (scalar mass operator)
518    * BP3 (scalar Laplace operator)
519* - `petsc`
520  - * BP1 (scalar mass operator)
521    * BP3 (scalar Laplace operator)
522* - `nek5000`
523  - * BP1 (scalar mass operator)
524    * BP3 (scalar Laplace operator)
525:::
526
527(v0-21)=
528
529## v0.21 (Sep 30, 2018)
530
531A MAGMA backend (which relies upon the
532[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this
533release. This initial integration set up the framework of using MAGMA and provided the
534libCEED functionality through MAGMA kernels as one of libCEED’s computational backends.
535As any other backend, the MAGMA backend provides extended basic data structures for
536{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements
537the fundamental CEED building blocks to work with the new data structures.
538In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data
539but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled
540internally, and thus seamlessly to the user, through the functions/methods that are
541provided to support them.
542
543Backends available in this release:
544
545| CEED resource (`-ceed`) | Backend                         |
546|-------------------------|---------------------------------|
547| `/cpu/self`             | Serial reference implementation |
548| `/cpu/occa`             | Serial OCCA kernels             |
549| `/gpu/occa`             | CUDA OCCA kernels               |
550| `/omp/occa`             | OpenMP OCCA kernels             |
551| `/ocl/occa`             | OpenCL OCCA kernels             |
552| `/gpu/magma`            | CUDA MAGMA kernels              |
553
554Examples available in this release:
555
556:::{list-table}
557:header-rows: 1
558:widths: auto
559* - User code
560  - Example
561* - `ceed`
562  - * ex1 (volume)
563* - `mfem`
564  - * BP1 (scalar mass operator)
565    * BP3 (scalar Laplace operator)
566* - `petsc`
567  - * BP1 (scalar mass operator)
568* - `nek5000`
569  - * BP1 (scalar mass operator)
570:::
571
572(v0-2)=
573
574## v0.2 (Mar 30, 2018)
575
576libCEED was made publicly available the first full CEED software distribution, release
577CEED 1.0. The distribution was made available using the Spack package manager to provide
578a common, easy-to-use build environment, where the user can build the CEED distribution
579with all dependencies. This release included a new Fortran interface for the library.
580This release also contained major improvements in the OCCA backend (including a new
581`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to
582compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an
583`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application
584of the Laplace operator) was also added to this release.
585
586Backends available in this release:
587
588| CEED resource (`-ceed`) | Backend                         |
589|-------------------------|---------------------------------|
590| `/cpu/self`             | Serial reference implementation |
591| `/cpu/occa`             | Serial OCCA kernels             |
592| `/gpu/occa`             | CUDA OCCA kernels               |
593| `/omp/occa`             | OpenMP OCCA kernels             |
594| `/ocl/occa`             | OpenCL OCCA kernels             |
595
596Examples available in this release:
597
598:::{list-table}
599:header-rows: 1
600:widths: auto
601* - User code
602  - Example
603* - `ceed`
604  - * ex1 (volume)
605* - `mfem`
606  - * BP1 (scalar mass operator)
607    * BP3 (scalar Laplace operator)
608* - `petsc`
609  - * BP1 (scalar mass operator)
610* - `nek5000`
611  - * BP1 (scalar mass operator)
612:::
613
614(v0-1)=
615
616## v0.1 (Jan 3, 2018)
617
618Initial low-level API of the CEED project. The low-level API provides a set of Finite
619Elements kernels and components for writing new low-level kernels. Examples include:
620vector and sparse linear algebra, element matrix assembly over a batch of elements,
621partial assembly and action for efficient high-order operators like mass, diffusion,
622advection, etc. The main goal of the low-level API is to establish the basis for the
623high-level API. Also, identifying such low-level kernels and providing a reference
624implementation for them serves as the basis for specialized backend implementations.
625This release contained several backends: `/cpu/self`, and backends which rely upon the
626[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`,
627`/gpu/occa`, and `/omp/occa`.
628It also included several examples, in the `examples` folder:
629A standalone code that shows the usage of libCEED (with no external
630dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1
631(with the application of the mass operator); and a `petsc` example to perform BP1
632(with the application of the mass operator).
633
634Backends available in this release:
635
636| CEED resource (`-ceed`) | Backend                         |
637|-------------------------|---------------------------------|
638| `/cpu/self`             | Serial reference implementation |
639| `/cpu/occa`             | Serial OCCA kernels             |
640| `/gpu/occa`             | CUDA OCCA kernels               |
641| `/omp/occa`             | OpenMP OCCA kernels             |
642
643Examples available in this release:
644
645| User code             | Example                           |
646|-----------------------|-----------------------------------|
647| `ceed`                | ex1 (scalar Laplace operator)     |
648| `mfem`                | BP1 (scalar mass operator)        |
649| `petsc`               | BP1 (scalar mass operator)        |
650```
651