xref: /libCEED/doc/sphinx/source/releasenotes.md (revision ca94c3ddc8f82b7d93a79f9e4812e99b8be840ff)
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
13### New features
14
15- Add `CeedOperatorCreateAtPoints` which evaluates the `CeedQFunction` at arbitrary locations in each element, for use in Particle in Cell, Material Point Method, and similar methods.
16
17### Examples
18
19(v0-12)=
20
21## v0.12 (Oct 31, 2023)
22
23### Interface changes
24
25- Update `CeedOperatorContext*` functions to `CeedOperator*Context*` functions for consistency.
26For example, `CeedOperatorContextGetFieldLabel` was renamed to `CeedOperatorGetContextFieldLabel`.
27- Removed `CeedBasisSetNumQuadraturePoints` as redundant and bug-prone interface.
28
29### New features
30
31- Added {c:func}`CeedOperatorGetFieldByName` to access a specific `CeedOperatorField` by its name.
32- Update `/cpu/self/memcheck/*` backends to help verify `CeedVector` array access assumptions and `CeedQFunction` user output assumptions.
33- Update {c:func}`CeedOperatorLinearAssembleDiagonal` to provide default implementation that supports `CeedOperator` with multiple active bases.
34- Added Sycl backends `/gpu/sycl/ref`, `/gpu/sycl/shared`, and `/gpu/sycl/gen`.
35- Added {c:func}`CeedBasisApplyAtPoints` for evaluation of values and derivatives at arbitrary points inside elements.
36- Added support for non-tensor $H(\text{curl})$ finite element spaces with {c:func}`CeedBasisCreateHcurl`.
37- 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)).
38- Added {c:func}`CeedOperatorLinearAssemblePointBlockDiagonalSymbolic` to create COO mapping for mapping out of {c:func}`CeedOperatorLinearAssemblePointBlockDiagonal`.
39- Added support for application codes which manage multiple {ref}`Ceed` objects, parallelized across OpenMP threads.
40
41### Examples
42
43- Add `DMSwarm` example demonstrating interpolation from background mesh to swarm points and projection from swarm points to background mesh.
44
45#### {ref}`example-petsc-bps`
46
47- Requires PETSc version 3.19 or later.
48
49#### {ref}`example-petsc-navier-stokes`
50
51- Updated restart and checkpointing interface.
52- Add data-driven subgrid-stress model.
53- Add differential filtering of solution.
54- Add turbulence statistics collection over spanwise-symmetric geometries.
55- Add Taylor-Green vortex initial condition.
56- Add Riemann-based outflow boundary conditions.
57- Added vortex shedding and flow past cylinder example, including calculations for lift, drag, and heat transfer.
58- Add Internal Damping Layer (IDL) for helping turbulent simulation stability.
59- Derive `CeedBasis` from `PetscFE`, and various other internal maintainability updates.
60
61(v0-11)=
62
63## v0.11 (Dec 24, 2022)
64
65### Interface changes
66
67- Added {c:func}`CeedOperatorSetName` for more readable {c:func}`CeedOperatorView` output.
68- Added {c:func}`CeedBasisCreateProjection` to facilitate interpolation between nodes for separate `CeedBases`.
69- Rename and move {c:func}`CeedCompositeOperatorGetNumSub` and {c:func}`CeedCompositeOperatorGetSubList` to public interface.
70- Renamed `CEED_BASIS_COLLOCATED` to `CEED_BASIS_NONE` for clarity.
71Some 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.
72
73### New features
74
75- Update `/cpu/self/memcheck/*` backends to help verify `CeedQFunctionContext` data sizes provided by user.
76- Improved support for $H(\text{div})$ bases.
77- Added `CeedInt_FMT` to support potential future use of larger integer sizes.
78- Added `CEED_QFUNCTION_ATTR` for setting compiler attributes/pragmas to `CEED_QFUNCTION_HELPER` and `CEED_QFUNCTION`.
79- OCCA backend updated to latest OCCA release; DPC++ and OMP OCCA modes enabled.
80Due to a limitation of the OCCA parser, typedefs are required to use pointers to arrays in QFunctions with the OCCA backend.
81This issue will be fixed in a future OCCA release.
82
83### Bugfix
84
85- Fix bug in setting device id for GPU backends.
86- Fix storing of indices for `CeedElemRestriction` on the host with GPU backends.
87- Fix `CeedElemRestriction` sizing for {c:func}`CeedOperatorAssemblePointBlockDiagonal`.
88- Fix bugs in CPU implementation of {c:func}`CeedOperatorLinearAssemble` when there are different number of active input modes and active output modes.
89
90### Examples
91
92#### {ref}`example-petsc-navier-stokes`
93
94- Various performance enhancements, analytic matrix-free and assembled Jacobian, and PETSc solver configurations for GPUs.
95- Refactored to improve code reuse and modularity.
96- Support for primitive variables for more accurate boundary layers and all-speed flow.
97- Added $YZ\beta$ shock capturing scheme and Shock Tube example.
98- Added Channel example, with comparison to analytic solutions.
99- Added Flat Plate with boundary layer mesh and compressible Blasius inflow condition based on Chebyshev collocation solution of the Blasius equations.
100- Added strong and weak synthetic turbulence generation (STG) inflow boundary conditions.
101- Added "freestream" boundary conditions based on HLLC Riemann solver.
102- Automated stabilization coefficients for different basis degree.
103
104#### {ref}`example-petsc-bps`
105
106- Support for convergence studies.
107
108### Maintainability
109
110- 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.
111- Enabled support for `p > 8` for `/gpu/*/shared` backends.
112- Switch to `clang-format` over `astyle` for automatic formatting; Makefile command changed to `make format` from `make style`.
113- Improved test harness.
114
115(v0-10-1)=
116
117## v0.10.1 (Apr 11, 2022)
118
119### Interface changes
120
121- Added {c:func}`CeedQFunctionSetUserFlopsEstimate` and {c:func}`CeedOperatorGetFlopsEstimate` to facilitate estimating FLOPs in operator application.
122
123### New features
124
125- Switched MAGMA backends to use runtime compilation for tensor basis kernels (and element restriction kernels, in non-deterministic `/gpu/*/magma` backends).
126This reduces time to compile the library and increases the range of parameters for which the MAGMA tensor basis kernels will work.
127
128### Bugfix
129
130- Install JiT source files in install directory to fix GPU functionality for installed libCEED.
131
132(v0-10)=
133
134## v0.10 (Mar 21, 2022)
135
136### Interface changes
137
138- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields.
139- 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`.
140- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added.
141- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients.
142- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces.
143- Warning added when compiling OCCA backend to alert users that this backend is experimental.
144- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`.
145- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name.
146- 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`.
147- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitialized arrays; require initialized data for {c:func}`CeedVectorGetArray`.
148- 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.
149- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retrieve user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`.
150- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums.
151- 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.
152- Added {c:func}`CeedOperatorSetQFunctionUpdated` to facilitate QFunction data re-use between operators sharing the same quadrature space, such as in a multigrid hierarchy.
153- Added {c:func}`CeedOperatorGetActiveVectorLengths` to get shape of CeedOperator.
154
155### New features
156
157- `CeedScalar` can now be set as `float` or `double` at compile time.
158- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends.
159- 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.
160- Remove need to guard library headers in QFunction source for code generation backends.
161- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present.
162- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally.
163- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file.
164- Added support for non-tensor $H(\text{div})$ elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor.
165- 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.
166- Added support for element matrix assembly in GPU backends.
167
168### Maintainability
169
170- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage.
171- `Include-what-you-use` makefile target added as `make iwyu`.
172- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase.
173- Put GPU JiTed kernel source code into separate files.
174- 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.
175
176(v0-9)=
177
178## v0.9 (Jul 6, 2021)
179
180### Interface changes
181
182- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact.
183
184### New features
185
186- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use.
187- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends.
188- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations.
189- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed).
190
191### Performance improvements
192
193### Examples
194
195- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations.
196- Fluid mechanics example adds GPU support and improves modularity.
197
198### Deprecated backends
199
200- 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.
201
202(v0-8)=
203
204## v0.8 (Mar 31, 2021)
205
206### Interface changes
207
208- Error handling improved to include enumerated error codes for C interface return values.
209- 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.
210
211### New features
212
213- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features.
214- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly.
215- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators.
216
217### Performance improvements
218
219- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`.
220- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`.
221
222### Examples
223
224- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions.
225- {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.
226- {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.
227- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree.
228
229(v0-7)=
230
231## v0.7 (Sep 29, 2020)
232
233### Interface changes
234
235- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors.
236  As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`.
237  These changes improve support for mixed finite element methods.
238- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature.
239- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity.
240- 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.
241- 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.
242- 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.
243- 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.
244  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`.
245- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory.
246- 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.
247
248### New features
249
250- New HIP backend: `/gpu/hip/ref`.
251- CeedQFunction support for user `CUfunction`s in some backends
252
253### Performance improvements
254
255- OCCA backend rebuilt to facilitate future performance enhancements.
256- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`.
257
258### Examples
259
260- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions.
261
262### Deprecated backends
263
264- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`.
265
266(v0-6)=
267
268## v0.6 (Mar 29, 2020)
269
270libCEED v0.6 contains numerous new features and examples, as well as expanded
271documentation in [this new website](https://libceed.org).
272
273### New features
274
275- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly
276  1-1 correspondence with the C interface, plus some convenience features.  For instance,
277  data stored in the {cpp:type}`CeedVector` structure are available without copy as
278  {py:class}`numpy.ndarray`.  Short tutorials are provided in
279  [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/).
280- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point,
281  {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal
282  ({c:func}`CeedOperatorAssembleLinearDiagonal`).  These operations are useful for
283  preconditioning ingredients and are used in the libCEED's multigrid examples.
284- The inverse of separable operators can be obtained using
285  {c:func}`CeedOperatorCreateFDMElementInverse` and applied with
286  {c:func}`CeedOperatorApply`.  This is a useful preconditioning ingredient,
287  especially for Laplacians and related operators.
288- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`,
289  {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`.
290- Make public accessors for various attributes to facilitate writing composable code.
291- New backend: `/cpu/self/memcheck/serial`.
292- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA
293  backends.  (Single source is coming soon for OCCA backends.)
294- Fix some missing edge cases in CUDA backend.
295
296### Performance Improvements
297
298- MAGMA backend performance optimization and non-tensor bases.
299- No-copy optimization in {c:func}`CeedOperatorApply`.
300
301### Interface changes
302
303- Replace {code}`CeedElemRestrictionCreateIdentity` and
304  {code}`CeedElemRestrictionCreateBlocked` with more flexible
305  {c:func}`CeedElemRestrictionCreateStrided` and
306  {c:func}`CeedElemRestrictionCreateBlockedStrided`.
307- Add arguments to {c:func}`CeedQFunctionCreateIdentity`.
308- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification
309  with {cpp:enum}`CeedInterlaceMode`.  This is now an attribute of the
310  {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no
311  longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and
312  {c:func}`CeedElemRestrictionApply`.
313
314### Examples
315
316libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`.
317Notable additions include:
318
319- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of
320  a domain in 1, 2, and 3 dimensions by applying a Laplacian.
321
322- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area
323  of domains (like the cube and sphere) by direct integration on a surface mesh;
324  demonstrates geometric dimension different from topological dimension.
325
326- PETSc {ref}`example-petsc-bps`:
327
328  - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support.
329  - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements
330    and transparent CUDA support.
331  - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`):
332    generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric
333    dimension different from topological dimension.
334
335- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid
336  solver with algebraic multigrid coarse solve.
337
338- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly
339  `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`),
340  implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and
341  quasi-2D computational domain support.
342
343- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for
344  linear elasticity, small-strain hyperelasticity, and globalized finite-strain
345  hyperelasticity using p-multigrid with algebraic multigrid coarse solve.
346
347(v0-5)=
348
349## v0.5 (Sep 18, 2019)
350
351For this release, several improvements were made. Two new CUDA backends were added to
352the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art
353performance using single-source {ref}`CeedQFunction`. From this release, users
354can define Q-Functions in a single source code independently of the targeted backend
355with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU
356compilation of the user provided {ref}`CeedQFunction` code. To allow a unified
357declaration, the {ref}`CeedQFunction` API has undergone a slight change:
358the `QFunctionField` parameter `ncomp` has been changed to `size`. This change
359requires setting the previous value of `ncomp` to `ncomp*dim` when adding a
360`QFunctionField` with eval mode `CEED EVAL GRAD`.
361
362Additionally, new CPU backends
363were included in this release, such as the `/cpu/self/opt/*` backends (which are
364written in pure C and use partial **E-vectors** to improve performance) and the
365`/cpu/self/ref/memcheck` backend (which relies upon the
366[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user
367{ref}`CeedQFunction` have no undefined values).
368This release also included various performance improvements, bug fixes, new examples,
369and improved tests. Among these improvements, vectorized instructions for
370{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD`
371instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and
372identity Q-Functions were introduced, and the PETSc benchmark problems were expanded
373to include unstructured meshes handling were. For this expansion, the prior version of
374the PETSc BPs, which only included data associated with structured geometries, were
375renamed `bpsraw`, and the new version of the BPs, which can handle data associated
376with any unstructured geometry, were called `bps`. Additionally, other benchmark
377problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively),
378and BP5 and BP6 (the collocated versions---for which the quadrature points are the same
379as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc
380examples. Furthermoew, another standalone libCEED example, called `ex2`, which
381computes the surface area of a given mesh was added to this release.
382
383Backends available in this release:
384
385| CEED resource (`-ceed`)  | Backend                                             |
386|--------------------------|-----------------------------------------------------|
387| `/cpu/self/ref/serial`   | Serial reference implementation                     |
388| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
389| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks            |
390| `/cpu/self/opt/serial`   | Serial optimized C implementation                   |
391| `/cpu/self/opt/blocked`  | Blocked optimized C implementation                  |
392| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
393| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
394| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
395| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
396| `/cpu/occa`              | Serial OCCA kernels                                 |
397| `/gpu/occa`              | CUDA OCCA kernels                                   |
398| `/omp/occa`              | OpenMP OCCA kernels                                 |
399| `/ocl/occa`              | OpenCL OCCA kernels                                 |
400| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
401| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
402| `/gpu/cuda/shared`       | Optimized pure CUDA kernels using shared memory     |
403| `/gpu/cuda/gen`          | Optimized pure CUDA kernels using code generation   |
404| `/gpu/magma`             | CUDA MAGMA kernels                                  |
405
406Examples available in this release:
407
408:::{list-table}
409:header-rows: 1
410:widths: auto
411* - User code
412  - Example
413* - `ceed`
414  - * ex1 (volume)
415    * ex2 (surface)
416* - `mfem`
417  - * BP1 (scalar mass operator)
418    * BP3 (scalar Laplace operator)
419* - `petsc`
420  - * BP1 (scalar mass operator)
421    * BP2 (vector mass operator)
422    * BP3 (scalar Laplace operator)
423    * BP4 (vector Laplace operator)
424    * BP5 (collocated scalar Laplace operator)
425    * BP6 (collocated vector Laplace operator)
426    * Navier-Stokes
427* - `nek5000`
428  - * BP1 (scalar mass operator)
429    * BP3 (scalar Laplace operator)
430:::
431
432(v0-4)=
433
434## v0.4 (Apr 1, 2019)
435
436libCEED v0.4 was made again publicly available in the second full CEED software
437distribution, release CEED 2.0. This release contained notable features, such as
438four new CPU backends, two new GPU backends, CPU backend optimizations, initial
439support for operator composition, performance benchmarking, and a Navier-Stokes demo.
440The new CPU backends in this release came in two families. The `/cpu/self/*/serial`
441backends process one element at a time and are intended for meshes with a smaller number
442of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of
443eight interlaced elements and are intended for meshes with higher numbers of elements.
444The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU
445performance. The `/cpu/self/xsmm/*` backends rely upon the
446[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU
447performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA.
448The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable
449performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple
450parallelization approach, where each thread treats a finite element. Using just in time
451compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this
452backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend
453achieve good peak performance for 1D, 2D, and low order 3D problems, but performance
454deteriorates very quickly when threads run out of registers.
455
456A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED
457examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`).
458This example solves the time-dependent Navier-Stokes equations of compressible gas
459dynamics in a static Eulerian three-dimensional frame, using structured high-order
460finite/spectral element spatial discretizations and explicit high-order time-stepping
461(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc,
462so that the pointwise physics (defined at quadrature points) is separated from the
463parallelization and meshing concerns.
464
465Backends available in this release:
466
467| CEED resource (`-ceed`)  | Backend                                             |
468|--------------------------|-----------------------------------------------------|
469| `/cpu/self/ref/serial`   | Serial reference implementation                     |
470| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
471| `/cpu/self/tmpl`         | Backend template, defaults to `/cpu/self/blocked`   |
472| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
473| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
474| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
475| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
476| `/cpu/occa`              | Serial OCCA kernels                                 |
477| `/gpu/occa`              | CUDA OCCA kernels                                   |
478| `/omp/occa`              | OpenMP OCCA kernels                                 |
479| `/ocl/occa`              | OpenCL OCCA kernels                                 |
480| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
481| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
482| `/gpu/magma`             | CUDA MAGMA kernels                                  |
483
484Examples available in this release:
485
486:::{list-table}
487:header-rows: 1
488:widths: auto
489* - User code
490  - Example
491* - `ceed`
492  - * ex1 (volume)
493* - `mfem`
494  - * BP1 (scalar mass operator)
495    * BP3 (scalar Laplace operator)
496* - `petsc`
497  - * BP1 (scalar mass operator)
498    * BP3 (scalar Laplace operator)
499    * Navier-Stokes
500* - `nek5000`
501  - * BP1 (scalar mass operator)
502    * BP3 (scalar Laplace operator)
503:::
504
505(v0-3)=
506
507## v0.3 (Sep 30, 2018)
508
509Notable features in this release include active/passive field interface, support for
510non-tensor bases, backend optimization, and improved Fortran interface. This release
511also focused on providing improved continuous integration, and many new tests with code
512coverage reports of about 90%. This release also provided a significant change to the
513public interface: a {ref}`CeedQFunction` can take any number of named input and output
514arguments while {ref}`CeedOperator` connects them to the actual data, which may be
515supplied explicitly to `CeedOperatorApply()` (active) or separately via
516`CeedOperatorSetField()` (passive). This interface change enables reusable libraries
517of CeedQFunctions and composition of block solvers constructed using
518{ref}`CeedOperator`. A concept of blocked restriction was added to this release and
519used in an optimized CPU backend. Although this is typically not visible to the user,
520it enables effective use of arbitrary-length SIMD while maintaining cache locality.
521This CPU backend also implements an algebraic factorization of tensor product gradients
522to perform fewer operations than standard application of interpolation and
523differentiation from nodes to quadrature points. This algebraic formulation
524automatically supports non-polynomial and non-interpolatory bases, thus is more general
525than the more common derivation in terms of Lagrange polynomials on the quadrature points.
526
527Backends available in this release:
528
529| CEED resource (`-ceed`) | Backend                                             |
530|-------------------------|-----------------------------------------------------|
531| `/cpu/self/blocked`     | Blocked reference implementation                    |
532| `/cpu/self/ref`         | Serial reference implementation                     |
533| `/cpu/self/tmpl`        | Backend template, defaults to `/cpu/self/blocked`   |
534| `/cpu/occa`             | Serial OCCA kernels                                 |
535| `/gpu/occa`             | CUDA OCCA kernels                                   |
536| `/omp/occa`             | OpenMP OCCA kernels                                 |
537| `/ocl/occa`             | OpenCL OCCA kernels                                 |
538| `/gpu/magma`            | CUDA MAGMA kernels                                  |
539
540Examples available in this release:
541
542:::{list-table}
543:header-rows: 1
544:widths: auto
545* - User code
546  - Example
547* - `ceed`
548  - * ex1 (volume)
549* - `mfem`
550  - * BP1 (scalar mass operator)
551    * BP3 (scalar Laplace operator)
552* - `petsc`
553  - * BP1 (scalar mass operator)
554    * BP3 (scalar Laplace operator)
555* - `nek5000`
556  - * BP1 (scalar mass operator)
557    * BP3 (scalar Laplace operator)
558:::
559
560(v0-21)=
561
562## v0.21 (Sep 30, 2018)
563
564A MAGMA backend (which relies upon the
565[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this
566release. This initial integration set up the framework of using MAGMA and provided the
567libCEED functionality through MAGMA kernels as one of libCEED’s computational backends.
568As any other backend, the MAGMA backend provides extended basic data structures for
569{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements
570the fundamental CEED building blocks to work with the new data structures.
571In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data
572but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled
573internally, and thus seamlessly to the user, through the functions/methods that are
574provided to support them.
575
576Backends available in this release:
577
578| CEED resource (`-ceed`) | Backend                         |
579|-------------------------|---------------------------------|
580| `/cpu/self`             | Serial reference implementation |
581| `/cpu/occa`             | Serial OCCA kernels             |
582| `/gpu/occa`             | CUDA OCCA kernels               |
583| `/omp/occa`             | OpenMP OCCA kernels             |
584| `/ocl/occa`             | OpenCL OCCA kernels             |
585| `/gpu/magma`            | CUDA MAGMA kernels              |
586
587Examples available in this release:
588
589:::{list-table}
590:header-rows: 1
591:widths: auto
592* - User code
593  - Example
594* - `ceed`
595  - * ex1 (volume)
596* - `mfem`
597  - * BP1 (scalar mass operator)
598    * BP3 (scalar Laplace operator)
599* - `petsc`
600  - * BP1 (scalar mass operator)
601* - `nek5000`
602  - * BP1 (scalar mass operator)
603:::
604
605(v0-2)=
606
607## v0.2 (Mar 30, 2018)
608
609libCEED was made publicly available the first full CEED software distribution, release
610CEED 1.0. The distribution was made available using the Spack package manager to provide
611a common, easy-to-use build environment, where the user can build the CEED distribution
612with all dependencies. This release included a new Fortran interface for the library.
613This release also contained major improvements in the OCCA backend (including a new
614`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to
615compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an
616`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application
617of the Laplace operator) was also added to this release.
618
619Backends available in this release:
620
621| CEED resource (`-ceed`) | Backend                         |
622|-------------------------|---------------------------------|
623| `/cpu/self`             | Serial reference implementation |
624| `/cpu/occa`             | Serial OCCA kernels             |
625| `/gpu/occa`             | CUDA OCCA kernels               |
626| `/omp/occa`             | OpenMP OCCA kernels             |
627| `/ocl/occa`             | OpenCL OCCA kernels             |
628
629Examples available in this release:
630
631:::{list-table}
632:header-rows: 1
633:widths: auto
634* - User code
635  - Example
636* - `ceed`
637  - * ex1 (volume)
638* - `mfem`
639  - * BP1 (scalar mass operator)
640    * BP3 (scalar Laplace operator)
641* - `petsc`
642  - * BP1 (scalar mass operator)
643* - `nek5000`
644  - * BP1 (scalar mass operator)
645:::
646
647(v0-1)=
648
649## v0.1 (Jan 3, 2018)
650
651Initial low-level API of the CEED project. The low-level API provides a set of Finite
652Elements kernels and components for writing new low-level kernels. Examples include:
653vector and sparse linear algebra, element matrix assembly over a batch of elements,
654partial assembly and action for efficient high-order operators like mass, diffusion,
655advection, etc. The main goal of the low-level API is to establish the basis for the
656high-level API. Also, identifying such low-level kernels and providing a reference
657implementation for them serves as the basis for specialized backend implementations.
658This release contained several backends: `/cpu/self`, and backends which rely upon the
659[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`,
660`/gpu/occa`, and `/omp/occa`.
661It also included several examples, in the `examples` folder:
662A standalone code that shows the usage of libCEED (with no external
663dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1
664(with the application of the mass operator); and a `petsc` example to perform BP1
665(with the application of the mass operator).
666
667Backends available in this release:
668
669| CEED resource (`-ceed`) | Backend                         |
670|-------------------------|---------------------------------|
671| `/cpu/self`             | Serial reference implementation |
672| `/cpu/occa`             | Serial OCCA kernels             |
673| `/gpu/occa`             | CUDA OCCA kernels               |
674| `/omp/occa`             | OpenMP OCCA kernels             |
675
676Examples available in this release:
677
678| User code             | Example                           |
679|-----------------------|-----------------------------------|
680| `ceed`                | ex1 (scalar Laplace operator)     |
681| `mfem`                | BP1 (scalar mass operator)        |
682| `petsc`               | BP1 (scalar mass operator)        |
683```
684