xref: /libCEED/doc/sphinx/source/releasenotes.md (revision a2f6637ed9f7b99e1cdac198632ca3cb6dbe41d4)
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### Examples
16
17(v0-12)=
18
19## v0.12 (Oct 31, 2023)
20
21### Interface changes
22
23- Update `CeedOperatorContext*` functions to `CeedOperator*Context*` functions for consistency.
24For example, `CeedOperatorContextGetFieldLabel` was renamed to `CeedOperatorGetContextFieldLabel`.
25- Removed `CeedBasisSetNumQuadraturePoints` as redundant and bug-prone interface.
26
27### New features
28
29- Added {c:func}`CeedOperatorGetFieldByName` to access a specific `CeedOperatorField` by its name.
30- Update `/cpu/self/memcheck/*` backends to help verify `CeedVector` array access assumptions and `CeedQFunction` user output assumptions.
31- Update {c:func}`CeedOperatorLinearAssembleDiagonal` to provide default implementation that supports `CeedOperator` with multiple active bases.
32- Added Sycl backends `/gpu/sycl/ref`, `/gpu/sycl/shared`, and `/gpu/sycl/gen`.
33- Added {c:func}`CeedBasisApplyAtPoints` for evaluation of values and derivatives at arbitrary points inside elements.
34- Added support for non-tensor $H(\text{curl})$ finite element spaces with {c:func}`CeedBasisCreateHcurl`.
35- 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)).
36- Added {c:func}`CeedOperatorLinearAssemblePointBlockDiagonalSymbolic` to create COO mapping for mapping out of {c:func}`CeedOperatorLinearAssemblePointBlockDiagonal`.
37- Added support for application codes which manage multiple {ref}`Ceed` objects, parallelized across OpenMP threads.
38
39### Examples
40
41- Add `DMSwarm` example demonstrating interpolation from background mesh to swarm points and projection from swarm points to background mesh.
42
43#### {ref}`example-petsc-bps`
44
45- Requires PETSc version 3.19 or later.
46
47#### {ref}`example-petsc-navier-stokes`
48
49- Updated restart and checkpointing interface.
50- Add data-driven subgrid-stress model.
51- Add differential filtering of solution.
52- Add turbulence statistics collection over spanwise-symmetric geometries.
53- Add Taylor-Green vortex initial condition.
54- Add Riemann-based outflow boundary conditions.
55- Added vortex shedding and flow past cylinder example, including calculations for lift, drag, and heat transfer.
56- Add Internal Damping Layer (IDL) for helping turbulent simulation stability.
57- Derive `CeedBasis` from `PetscFE`, and various other internal maintainability updates.
58
59(v0-11)=
60
61## v0.11 (Dec 24, 2022)
62
63### Interface changes
64
65- Added {c:func}`CeedOperatorSetName` for more readable {c:func}`CeedOperatorView` output.
66- Added {c:func}`CeedBasisCreateProjection` to facilitate interpolation between nodes for separate `CeedBases`.
67- Rename and move {c:func}`CeedCompositeOperatorGetNumSub` and {c:func}`CeedCompositeOperatorGetSubList` to public interface.
68- Renamed `CEED_BASIS_COLLOCATED` to `CEED_BASIS_NONE` for clarity.
69Some 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.
70
71### New features
72
73- Update `/cpu/self/memcheck/*` backends to help verify `CeedQFunctionContext` data sizes provided by user.
74- Improved support for $H(\text{div})$ bases.
75- Added `CeedInt_FMT` to support potential future use of larger integer sizes.
76- Added `CEED_QFUNCTION_ATTR` for setting compiler attributes/pragmas to `CEED_QFUNCTION_HELPER` and `CEED_QFUNCTION`.
77- OCCA backend updated to latest OCCA release; DPC++ and OMP OCCA modes enabled.
78Due to a limitation of the OCCA parser, typedefs are required to use pointers to arrays in QFunctions with the OCCA backend.
79This issue will be fixed in a future OCCA release.
80
81### Bugfix
82
83- Fix bug in setting device id for GPU backends.
84- Fix storing of indices for `CeedElemRestriction` on the host with GPU backends.
85- Fix `CeedElemRestriction` sizing for {c:func}`CeedOperatorAssemblePointBlockDiagonal`.
86- Fix bugs in CPU implementation of {c:func}`CeedOperatorLinearAssemble` when there are different number of active input modes and active output modes.
87
88### Examples
89
90#### {ref}`example-petsc-navier-stokes`
91
92- Various performance enhancements, analytic matrix-free and assembled Jacobian, and PETSc solver configurations for GPUs.
93- Refactored to improve code reuse and modularity.
94- Support for primitive variables for more accurate boundary layers and all-speed flow.
95- Added $YZ\beta$ shock capturing scheme and Shock Tube example.
96- Added Channel example, with comparison to analytic solutions.
97- Added Flat Plate with boundary layer mesh and compressible Blasius inflow condition based on Chebyshev collocation solution of the Blasius equations.
98- Added strong and weak synthetic turbulence generation (STG) inflow boundary conditions.
99- Added "freestream" boundary conditions based on HLLC Riemann solver.
100- Automated stabilization coefficients for different basis degree.
101
102#### {ref}`example-petsc-bps`
103
104- Support for convergence studies.
105
106### Maintainability
107
108- 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.
109- Enabled support for `p > 8` for `/gpu/*/shared` backends.
110- Switch to `clang-format` over `astyle` for automatic formatting; Makefile command changed to `make format` from `make style`.
111- Improved test harness.
112
113(v0-10-1)=
114
115## v0.10.1 (Apr 11, 2022)
116
117### Interface changes
118
119- Added {c:func}`CeedQFunctionSetUserFlopsEstimate` and {c:func}`CeedOperatorGetFlopsEstimate` to facilitate estimating FLOPs in operator application.
120
121### New features
122
123- Switched MAGMA backends to use runtime compilation for tensor basis kernels (and element restriction kernels, in non-deterministic `/gpu/*/magma` backends).
124This reduces time to compile the library and increases the range of parameters for which the MAGMA tensor basis kernels will work.
125
126### Bugfix
127
128- Install JiT source files in install directory to fix GPU functionality for installed libCEED.
129
130(v0-10)=
131
132## v0.10 (Mar 21, 2022)
133
134### Interface changes
135
136- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields.
137- 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`.
138- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added.
139- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients.
140- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces.
141- Warning added when compiling OCCA backend to alert users that this backend is experimental.
142- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`.
143- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name.
144- 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`.
145- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitialized arrays; require initialized data for {c:func}`CeedVectorGetArray`.
146- 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.
147- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retrieve user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`.
148- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums.
149- 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.
150- Added {c:func}`CeedOperatorSetQFunctionUpdated` to facilitate QFunction data re-use between operators sharing the same quadrature space, such as in a multigrid hierarchy.
151- Added {c:func}`CeedOperatorGetActiveVectorLengths` to get shape of CeedOperator.
152
153### New features
154
155- `CeedScalar` can now be set as `float` or `double` at compile time.
156- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends.
157- 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.
158- Remove need to guard library headers in QFunction source for code generation backends.
159- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present.
160- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally.
161- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file.
162- Added support for non-tensor $H(\text{div})$ elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor.
163- 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.
164- Added support for element matrix assembly in GPU backends.
165
166### Maintainability
167
168- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage.
169- `Include-what-you-use` makefile target added as `make iwyu`.
170- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase.
171- Put GPU JiTed kernel source code into separate files.
172- 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.
173
174(v0-9)=
175
176## v0.9 (Jul 6, 2021)
177
178### Interface changes
179
180- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact.
181
182### New features
183
184- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use.
185- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends.
186- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations.
187- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed).
188
189### Performance improvements
190
191### Examples
192
193- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations.
194- Fluid mechanics example adds GPU support and improves modularity.
195
196### Deprecated backends
197
198- 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.
199
200(v0-8)=
201
202## v0.8 (Mar 31, 2021)
203
204### Interface changes
205
206- Error handling improved to include enumerated error codes for C interface return values.
207- 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.
208
209### New features
210
211- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features.
212- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly.
213- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators.
214
215### Performance improvements
216
217- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`.
218- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`.
219
220### Examples
221
222- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions.
223- {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.
224- {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.
225- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree.
226
227(v0-7)=
228
229## v0.7 (Sep 29, 2020)
230
231### Interface changes
232
233- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors.
234  As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`.
235  These changes improve support for mixed finite element methods.
236- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature.
237- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity.
238- 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.
239- 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.
240- 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.
241- 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.
242  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`.
243- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory.
244- 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.
245
246### New features
247
248- New HIP backend: `/gpu/hip/ref`.
249- CeedQFunction support for user `CUfunction`s in some backends
250
251### Performance improvements
252
253- OCCA backend rebuilt to facilitate future performance enhancements.
254- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`.
255
256### Examples
257
258- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions.
259
260### Deprecated backends
261
262- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`.
263
264(v0-6)=
265
266## v0.6 (Mar 29, 2020)
267
268libCEED v0.6 contains numerous new features and examples, as well as expanded
269documentation in [this new website](https://libceed.org).
270
271### New features
272
273- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly
274  1-1 correspondence with the C interface, plus some convenience features.  For instance,
275  data stored in the {cpp:type}`CeedVector` structure are available without copy as
276  {py:class}`numpy.ndarray`.  Short tutorials are provided in
277  [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/).
278- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point,
279  {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal
280  ({c:func}`CeedOperatorAssembleLinearDiagonal`).  These operations are useful for
281  preconditioning ingredients and are used in the libCEED's multigrid examples.
282- The inverse of separable operators can be obtained using
283  {c:func}`CeedOperatorCreateFDMElementInverse` and applied with
284  {c:func}`CeedOperatorApply`.  This is a useful preconditioning ingredient,
285  especially for Laplacians and related operators.
286- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`,
287  {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`.
288- Make public accessors for various attributes to facilitate writing composable code.
289- New backend: `/cpu/self/memcheck/serial`.
290- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA
291  backends.  (Single source is coming soon for OCCA backends.)
292- Fix some missing edge cases in CUDA backend.
293
294### Performance Improvements
295
296- MAGMA backend performance optimization and non-tensor bases.
297- No-copy optimization in {c:func}`CeedOperatorApply`.
298
299### Interface changes
300
301- Replace {code}`CeedElemRestrictionCreateIdentity` and
302  {code}`CeedElemRestrictionCreateBlocked` with more flexible
303  {c:func}`CeedElemRestrictionCreateStrided` and
304  {c:func}`CeedElemRestrictionCreateBlockedStrided`.
305- Add arguments to {c:func}`CeedQFunctionCreateIdentity`.
306- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification
307  with {cpp:enum}`CeedInterlaceMode`.  This is now an attribute of the
308  {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no
309  longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and
310  {c:func}`CeedElemRestrictionApply`.
311
312### Examples
313
314libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`.
315Notable additions include:
316
317- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of
318  a domain in 1, 2, and 3 dimensions by applying a Laplacian.
319
320- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area
321  of domains (like the cube and sphere) by direct integration on a surface mesh;
322  demonstrates geometric dimension different from topological dimension.
323
324- PETSc {ref}`example-petsc-bps`:
325
326  - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support.
327  - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements
328    and transparent CUDA support.
329  - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`):
330    generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric
331    dimension different from topological dimension.
332
333- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid
334  solver with algebraic multigrid coarse solve.
335
336- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly
337  `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`),
338  implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and
339  quasi-2D computational domain support.
340
341- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for
342  linear elasticity, small-strain hyperelasticity, and globalized finite-strain
343  hyperelasticity using p-multigrid with algebraic multigrid coarse solve.
344
345(v0-5)=
346
347## v0.5 (Sep 18, 2019)
348
349For this release, several improvements were made. Two new CUDA backends were added to
350the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art
351performance using single-source {ref}`CeedQFunction`. From this release, users
352can define Q-Functions in a single source code independently of the targeted backend
353with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU
354compilation of the user provided {ref}`CeedQFunction` code. To allow a unified
355declaration, the {ref}`CeedQFunction` API has undergone a slight change:
356the `QFunctionField` parameter `ncomp` has been changed to `size`. This change
357requires setting the previous value of `ncomp` to `ncomp*dim` when adding a
358`QFunctionField` with eval mode `CEED EVAL GRAD`.
359
360Additionally, new CPU backends
361were included in this release, such as the `/cpu/self/opt/*` backends (which are
362written in pure C and use partial **E-vectors** to improve performance) and the
363`/cpu/self/ref/memcheck` backend (which relies upon the
364[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user
365{ref}`CeedQFunction` have no undefined values).
366This release also included various performance improvements, bug fixes, new examples,
367and improved tests. Among these improvements, vectorized instructions for
368{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD`
369instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and
370identity Q-Functions were introduced, and the PETSc benchmark problems were expanded
371to include unstructured meshes handling were. For this expansion, the prior version of
372the PETSc BPs, which only included data associated with structured geometries, were
373renamed `bpsraw`, and the new version of the BPs, which can handle data associated
374with any unstructured geometry, were called `bps`. Additionally, other benchmark
375problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively),
376and BP5 and BP6 (the collocated versions---for which the quadrature points are the same
377as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc
378examples. Furthermoew, another standalone libCEED example, called `ex2`, which
379computes the surface area of a given mesh was added to this release.
380
381Backends available in this release:
382
383| CEED resource (`-ceed`)  | Backend                                             |
384|--------------------------|-----------------------------------------------------|
385| `/cpu/self/ref/serial`   | Serial reference implementation                     |
386| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
387| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks            |
388| `/cpu/self/opt/serial`   | Serial optimized C implementation                   |
389| `/cpu/self/opt/blocked`  | Blocked optimized C implementation                  |
390| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
391| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
392| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
393| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
394| `/cpu/occa`              | Serial OCCA kernels                                 |
395| `/gpu/occa`              | CUDA OCCA kernels                                   |
396| `/omp/occa`              | OpenMP OCCA kernels                                 |
397| `/ocl/occa`              | OpenCL OCCA kernels                                 |
398| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
399| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
400| `/gpu/cuda/shared`       | Optimized pure CUDA kernels using shared memory     |
401| `/gpu/cuda/gen`          | Optimized pure CUDA kernels using code generation   |
402| `/gpu/magma`             | CUDA MAGMA kernels                                  |
403
404Examples available in this release:
405
406:::{list-table}
407:header-rows: 1
408:widths: auto
409* - User code
410  - Example
411* - `ceed`
412  - * ex1 (volume)
413    * ex2 (surface)
414* - `mfem`
415  - * BP1 (scalar mass operator)
416    * BP3 (scalar Laplace operator)
417* - `petsc`
418  - * BP1 (scalar mass operator)
419    * BP2 (vector mass operator)
420    * BP3 (scalar Laplace operator)
421    * BP4 (vector Laplace operator)
422    * BP5 (collocated scalar Laplace operator)
423    * BP6 (collocated vector Laplace operator)
424    * Navier-Stokes
425* - `nek5000`
426  - * BP1 (scalar mass operator)
427    * BP3 (scalar Laplace operator)
428:::
429
430(v0-4)=
431
432## v0.4 (Apr 1, 2019)
433
434libCEED v0.4 was made again publicly available in the second full CEED software
435distribution, release CEED 2.0. This release contained notable features, such as
436four new CPU backends, two new GPU backends, CPU backend optimizations, initial
437support for operator composition, performance benchmarking, and a Navier-Stokes demo.
438The new CPU backends in this release came in two families. The `/cpu/self/*/serial`
439backends process one element at a time and are intended for meshes with a smaller number
440of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of
441eight interlaced elements and are intended for meshes with higher numbers of elements.
442The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU
443performance. The `/cpu/self/xsmm/*` backends rely upon the
444[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU
445performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA.
446The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable
447performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple
448parallelization approach, where each thread treats a finite element. Using just in time
449compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this
450backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend
451achieve good peak performance for 1D, 2D, and low order 3D problems, but performance
452deteriorates very quickly when threads run out of registers.
453
454A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED
455examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`).
456This example solves the time-dependent Navier-Stokes equations of compressible gas
457dynamics in a static Eulerian three-dimensional frame, using structured high-order
458finite/spectral element spatial discretizations and explicit high-order time-stepping
459(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc,
460so that the pointwise physics (defined at quadrature points) is separated from the
461parallelization and meshing concerns.
462
463Backends available in this release:
464
465| CEED resource (`-ceed`)  | Backend                                             |
466|--------------------------|-----------------------------------------------------|
467| `/cpu/self/ref/serial`   | Serial reference implementation                     |
468| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
469| `/cpu/self/tmpl`         | Backend template, defaults to `/cpu/self/blocked`   |
470| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
471| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
472| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
473| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
474| `/cpu/occa`              | Serial OCCA kernels                                 |
475| `/gpu/occa`              | CUDA OCCA kernels                                   |
476| `/omp/occa`              | OpenMP OCCA kernels                                 |
477| `/ocl/occa`              | OpenCL OCCA kernels                                 |
478| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
479| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
480| `/gpu/magma`             | CUDA MAGMA kernels                                  |
481
482Examples available in this release:
483
484:::{list-table}
485:header-rows: 1
486:widths: auto
487* - User code
488  - Example
489* - `ceed`
490  - * ex1 (volume)
491* - `mfem`
492  - * BP1 (scalar mass operator)
493    * BP3 (scalar Laplace operator)
494* - `petsc`
495  - * BP1 (scalar mass operator)
496    * BP3 (scalar Laplace operator)
497    * Navier-Stokes
498* - `nek5000`
499  - * BP1 (scalar mass operator)
500    * BP3 (scalar Laplace operator)
501:::
502
503(v0-3)=
504
505## v0.3 (Sep 30, 2018)
506
507Notable features in this release include active/passive field interface, support for
508non-tensor bases, backend optimization, and improved Fortran interface. This release
509also focused on providing improved continuous integration, and many new tests with code
510coverage reports of about 90%. This release also provided a significant change to the
511public interface: a {ref}`CeedQFunction` can take any number of named input and output
512arguments while {ref}`CeedOperator` connects them to the actual data, which may be
513supplied explicitly to `CeedOperatorApply()` (active) or separately via
514`CeedOperatorSetField()` (passive). This interface change enables reusable libraries
515of CeedQFunctions and composition of block solvers constructed using
516{ref}`CeedOperator`. A concept of blocked restriction was added to this release and
517used in an optimized CPU backend. Although this is typically not visible to the user,
518it enables effective use of arbitrary-length SIMD while maintaining cache locality.
519This CPU backend also implements an algebraic factorization of tensor product gradients
520to perform fewer operations than standard application of interpolation and
521differentiation from nodes to quadrature points. This algebraic formulation
522automatically supports non-polynomial and non-interpolatory bases, thus is more general
523than the more common derivation in terms of Lagrange polynomials on the quadrature points.
524
525Backends available in this release:
526
527| CEED resource (`-ceed`) | Backend                                             |
528|-------------------------|-----------------------------------------------------|
529| `/cpu/self/blocked`     | Blocked reference implementation                    |
530| `/cpu/self/ref`         | Serial reference implementation                     |
531| `/cpu/self/tmpl`        | Backend template, defaults to `/cpu/self/blocked`   |
532| `/cpu/occa`             | Serial OCCA kernels                                 |
533| `/gpu/occa`             | CUDA OCCA kernels                                   |
534| `/omp/occa`             | OpenMP OCCA kernels                                 |
535| `/ocl/occa`             | OpenCL OCCA kernels                                 |
536| `/gpu/magma`            | CUDA MAGMA kernels                                  |
537
538Examples available in this release:
539
540:::{list-table}
541:header-rows: 1
542:widths: auto
543* - User code
544  - Example
545* - `ceed`
546  - * ex1 (volume)
547* - `mfem`
548  - * BP1 (scalar mass operator)
549    * BP3 (scalar Laplace operator)
550* - `petsc`
551  - * BP1 (scalar mass operator)
552    * BP3 (scalar Laplace operator)
553* - `nek5000`
554  - * BP1 (scalar mass operator)
555    * BP3 (scalar Laplace operator)
556:::
557
558(v0-21)=
559
560## v0.21 (Sep 30, 2018)
561
562A MAGMA backend (which relies upon the
563[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this
564release. This initial integration set up the framework of using MAGMA and provided the
565libCEED functionality through MAGMA kernels as one of libCEED’s computational backends.
566As any other backend, the MAGMA backend provides extended basic data structures for
567{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements
568the fundamental CEED building blocks to work with the new data structures.
569In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data
570but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled
571internally, and thus seamlessly to the user, through the functions/methods that are
572provided to support them.
573
574Backends available in this release:
575
576| CEED resource (`-ceed`) | Backend                         |
577|-------------------------|---------------------------------|
578| `/cpu/self`             | Serial reference implementation |
579| `/cpu/occa`             | Serial OCCA kernels             |
580| `/gpu/occa`             | CUDA OCCA kernels               |
581| `/omp/occa`             | OpenMP OCCA kernels             |
582| `/ocl/occa`             | OpenCL OCCA kernels             |
583| `/gpu/magma`            | CUDA MAGMA kernels              |
584
585Examples available in this release:
586
587:::{list-table}
588:header-rows: 1
589:widths: auto
590* - User code
591  - Example
592* - `ceed`
593  - * ex1 (volume)
594* - `mfem`
595  - * BP1 (scalar mass operator)
596    * BP3 (scalar Laplace operator)
597* - `petsc`
598  - * BP1 (scalar mass operator)
599* - `nek5000`
600  - * BP1 (scalar mass operator)
601:::
602
603(v0-2)=
604
605## v0.2 (Mar 30, 2018)
606
607libCEED was made publicly available the first full CEED software distribution, release
608CEED 1.0. The distribution was made available using the Spack package manager to provide
609a common, easy-to-use build environment, where the user can build the CEED distribution
610with all dependencies. This release included a new Fortran interface for the library.
611This release also contained major improvements in the OCCA backend (including a new
612`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to
613compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an
614`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application
615of the Laplace operator) was also added to this release.
616
617Backends available in this release:
618
619| CEED resource (`-ceed`) | Backend                         |
620|-------------------------|---------------------------------|
621| `/cpu/self`             | Serial reference implementation |
622| `/cpu/occa`             | Serial OCCA kernels             |
623| `/gpu/occa`             | CUDA OCCA kernels               |
624| `/omp/occa`             | OpenMP OCCA kernels             |
625| `/ocl/occa`             | OpenCL OCCA kernels             |
626
627Examples available in this release:
628
629:::{list-table}
630:header-rows: 1
631:widths: auto
632* - User code
633  - Example
634* - `ceed`
635  - * ex1 (volume)
636* - `mfem`
637  - * BP1 (scalar mass operator)
638    * BP3 (scalar Laplace operator)
639* - `petsc`
640  - * BP1 (scalar mass operator)
641* - `nek5000`
642  - * BP1 (scalar mass operator)
643:::
644
645(v0-1)=
646
647## v0.1 (Jan 3, 2018)
648
649Initial low-level API of the CEED project. The low-level API provides a set of Finite
650Elements kernels and components for writing new low-level kernels. Examples include:
651vector and sparse linear algebra, element matrix assembly over a batch of elements,
652partial assembly and action for efficient high-order operators like mass, diffusion,
653advection, etc. The main goal of the low-level API is to establish the basis for the
654high-level API. Also, identifying such low-level kernels and providing a reference
655implementation for them serves as the basis for specialized backend implementations.
656This release contained several backends: `/cpu/self`, and backends which rely upon the
657[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`,
658`/gpu/occa`, and `/omp/occa`.
659It also included several examples, in the `examples` folder:
660A standalone code that shows the usage of libCEED (with no external
661dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1
662(with the application of the mass operator); and a `petsc` example to perform BP1
663(with the application of the mass operator).
664
665Backends available in this release:
666
667| CEED resource (`-ceed`) | Backend                         |
668|-------------------------|---------------------------------|
669| `/cpu/self`             | Serial reference implementation |
670| `/cpu/occa`             | Serial OCCA kernels             |
671| `/gpu/occa`             | CUDA OCCA kernels               |
672| `/omp/occa`             | OpenMP OCCA kernels             |
673
674Examples available in this release:
675
676| User code             | Example                           |
677|-----------------------|-----------------------------------|
678| `ceed`                | ex1 (scalar Laplace operator)     |
679| `mfem`                | BP1 (scalar mass operator)        |
680| `petsc`               | BP1 (scalar mass operator)        |
681```
682