xref: /libCEED/doc/sphinx/source/releasenotes.md (revision f07714d9267ae37c0c844e07b7f7e93ea3da4ade)
1# Changes/Release Notes
2
3On this page we provide a summary of the main API changes, new features and examples for each release of libCEED.
4
5(main)=
6
7## Current `main` branch
8
9### Interface changes
10
11- Add `bool` field type for `CeedQFunctionContext` and related interfaces to use `bool` fields.
12- `CEED_BASIS_COLLOCATED` removed; users should only use `CEED_BASIS_NONE`.
13- Remove unneeded pointer for `CeedElemRestrictionGetELayout`.
14
15### New features
16
17- Add `CeedOperatorCreateAtPoints` which evaluates the `CeedQFunction` at arbitrary locations in each element, for use in Particle in Cell, Material Point Method, and similar methods.
18- Add `CeedElemRestrictionGetLLayout` to provide L-vector layout for strided `CeedElemRestriction` created with `CEED_BACKEND_STRIDES`.
19- Add `CeedVectorReturnCeed` and similar when parent `Ceed` context for a libCEED object is only needed once in a calling scope.
20
21### Examples
22
23- Add deal.II example with CEED BP suite.
24
25(v0-12)=
26
27## v0.12 (Oct 31, 2023)
28
29### Interface changes
30
31- Update `CeedOperatorContext*` functions to `CeedOperator*Context*` functions for consistency.
32For example, `CeedOperatorContextGetFieldLabel` was renamed to `CeedOperatorGetContextFieldLabel`.
33- Removed `CeedBasisSetNumQuadraturePoints` as redundant and bug-prone interface.
34
35### New features
36
37- Added {c:func}`CeedOperatorGetFieldByName` to access a specific `CeedOperatorField` by its name.
38- Update `/cpu/self/memcheck/*` backends to help verify `CeedVector` array access assumptions and `CeedQFunction` user output assumptions.
39- Update {c:func}`CeedOperatorLinearAssembleDiagonal` to provide default implementation that supports `CeedOperator` with multiple active bases.
40- Added Sycl backends `/gpu/sycl/ref`, `/gpu/sycl/shared`, and `/gpu/sycl/gen`.
41- Added {c:func}`CeedBasisApplyAtPoints` for evaluation of values and derivatives at arbitrary points inside elements.
42- Added support for non-tensor $H(\text{curl})$ finite element spaces with {c:func}`CeedBasisCreateHcurl`.
43- 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)).
44- Added {c:func}`CeedOperatorLinearAssemblePointBlockDiagonalSymbolic` to create COO mapping for mapping out of {c:func}`CeedOperatorLinearAssemblePointBlockDiagonal`.
45- Added support for application codes which manage multiple {ref}`Ceed` objects, parallelized across OpenMP threads.
46
47### Examples
48
49- Add `DMSwarm` example demonstrating interpolation from background mesh to swarm points and projection from swarm points to background mesh.
50
51#### {ref}`example-petsc-bps`
52
53- Requires PETSc version 3.19 or later.
54
55#### {ref}`example-petsc-navier-stokes`
56
57- Updated restart and checkpointing interface.
58- Add data-driven subgrid-stress model.
59- Add differential filtering of solution.
60- Add turbulence statistics collection over spanwise-symmetric geometries.
61- Add Taylor-Green vortex initial condition.
62- Add Riemann-based outflow boundary conditions.
63- Added vortex shedding and flow past cylinder example, including calculations for lift, drag, and heat transfer.
64- Add Internal Damping Layer (IDL) for helping turbulent simulation stability.
65- Derive `CeedBasis` from `PetscFE`, and various other internal maintainability updates.
66
67(v0-11)=
68
69## v0.11 (Dec 24, 2022)
70
71### Interface changes
72
73- Added {c:func}`CeedOperatorSetName` for more readable {c:func}`CeedOperatorView` output.
74- Added {c:func}`CeedBasisCreateProjection` to facilitate interpolation between nodes for separate `CeedBases`.
75- Rename and move {c:func}`CeedCompositeOperatorGetNumSub` and {c:func}`CeedCompositeOperatorGetSubList` to public interface.
76- Renamed `CEED_BASIS_COLLOCATED` to `CEED_BASIS_NONE` for clarity.
77Some 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.
78
79### New features
80
81- Update `/cpu/self/memcheck/*` backends to help verify `CeedQFunctionContext` data sizes provided by user.
82- Improved support for $H(\text{div})$ bases.
83- Added `CeedInt_FMT` to support potential future use of larger integer sizes.
84- Added `CEED_QFUNCTION_ATTR` for setting compiler attributes/pragmas to `CEED_QFUNCTION_HELPER` and `CEED_QFUNCTION`.
85- OCCA backend updated to latest OCCA release; DPC++ and OMP OCCA modes enabled.
86Due to a limitation of the OCCA parser, typedefs are required to use pointers to arrays in QFunctions with the OCCA backend.
87This issue will be fixed in a future OCCA release.
88
89### Bugfix
90
91- Fix bug in setting device id for GPU backends.
92- Fix storing of indices for `CeedElemRestriction` on the host with GPU backends.
93- Fix `CeedElemRestriction` sizing for {c:func}`CeedOperatorAssemblePointBlockDiagonal`.
94- Fix bugs in CPU implementation of {c:func}`CeedOperatorLinearAssemble` when there are different number of active input modes and active output modes.
95
96### Examples
97
98#### {ref}`example-petsc-navier-stokes`
99
100- Various performance enhancements, analytic matrix-free and assembled Jacobian, and PETSc solver configurations for GPUs.
101- Refactored to improve code reuse and modularity.
102- Support for primitive variables for more accurate boundary layers and all-speed flow.
103- Added $YZ\beta$ shock capturing scheme and Shock Tube example.
104- Added Channel example, with comparison to analytic solutions.
105- Added Flat Plate with boundary layer mesh and compressible Blasius inflow condition based on Chebyshev collocation solution of the Blasius equations.
106- Added strong and weak synthetic turbulence generation (STG) inflow boundary conditions.
107- Added "freestream" boundary conditions based on HLLC Riemann solver.
108- Automated stabilization coefficients for different basis degree.
109
110#### {ref}`example-petsc-bps`
111
112- Support for convergence studies.
113
114### Maintainability
115
116- 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.
117- Enabled support for `p > 8` for `/gpu/*/shared` backends.
118- Switch to `clang-format` over `astyle` for automatic formatting; Makefile command changed to `make format` from `make style`.
119- Improved test harness.
120
121(v0-10-1)=
122
123## v0.10.1 (Apr 11, 2022)
124
125### Interface changes
126
127- Added {c:func}`CeedQFunctionSetUserFlopsEstimate` and {c:func}`CeedOperatorGetFlopsEstimate` to facilitate estimating FLOPs in operator application.
128
129### New features
130
131- Switched MAGMA backends to use runtime compilation for tensor basis kernels (and element restriction kernels, in non-deterministic `/gpu/*/magma` backends).
132This reduces time to compile the library and increases the range of parameters for which the MAGMA tensor basis kernels will work.
133
134### Bugfix
135
136- Install JiT source files in install directory to fix GPU functionality for installed libCEED.
137
138(v0-10)=
139
140## v0.10 (Mar 21, 2022)
141
142### Interface changes
143
144- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields.
145- 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`.
146- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added.
147- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients.
148- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces.
149- Warning added when compiling OCCA backend to alert users that this backend is experimental.
150- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`.
151- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name.
152- 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`.
153- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitialized arrays; require initialized data for {c:func}`CeedVectorGetArray`.
154- 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.
155- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retrieve user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`.
156- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums.
157- 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.
158- Added {c:func}`CeedOperatorSetQFunctionUpdated` to facilitate QFunction data re-use between operators sharing the same quadrature space, such as in a multigrid hierarchy.
159- Added {c:func}`CeedOperatorGetActiveVectorLengths` to get shape of CeedOperator.
160
161### New features
162
163- `CeedScalar` can now be set as `float` or `double` at compile time.
164- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends.
165- 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.
166- Remove need to guard library headers in QFunction source for code generation backends.
167- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present.
168- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally.
169- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file.
170- Added support for non-tensor $H(\text{div})$ elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor.
171- 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.
172- Added support for element matrix assembly in GPU backends.
173
174### Maintainability
175
176- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage.
177- `Include-what-you-use` makefile target added as `make iwyu`.
178- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase.
179- Put GPU JiTed kernel source code into separate files.
180- 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.
181
182(v0-9)=
183
184## v0.9 (Jul 6, 2021)
185
186### Interface changes
187
188- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact.
189
190### New features
191
192- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use.
193- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends.
194- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations.
195- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed).
196
197### Performance improvements
198
199### Examples
200
201- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations.
202- Fluid mechanics example adds GPU support and improves modularity.
203
204### Deprecated backends
205
206- 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.
207
208(v0-8)=
209
210## v0.8 (Mar 31, 2021)
211
212### Interface changes
213
214- Error handling improved to include enumerated error codes for C interface return values.
215- 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.
216
217### New features
218
219- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features.
220- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly.
221- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators.
222
223### Performance improvements
224
225- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`.
226- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`.
227
228### Examples
229
230- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions.
231- {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.
232- {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.
233- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree.
234
235(v0-7)=
236
237## v0.7 (Sep 29, 2020)
238
239### Interface changes
240
241- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors.
242  As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`.
243  These changes improve support for mixed finite element methods.
244- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature.
245- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity.
246- 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.
247- 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.
248- 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.
249- 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.
250  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`.
251- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory.
252- 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.
253
254### New features
255
256- New HIP backend: `/gpu/hip/ref`.
257- CeedQFunction support for user `CUfunction`s in some backends
258
259### Performance improvements
260
261- OCCA backend rebuilt to facilitate future performance enhancements.
262- PETSc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`.
263
264### Examples
265
266- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions.
267
268### Deprecated backends
269
270- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`.
271
272(v0-6)=
273
274## v0.6 (Mar 29, 2020)
275
276libCEED v0.6 contains numerous new features and examples, as well as expanded
277documentation in [this new website](https://libceed.org).
278
279### New features
280
281- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly
282  1-1 correspondence with the C interface, plus some convenience features.  For instance,
283  data stored in the {cpp:type}`CeedVector` structure are available without copy as
284  {py:class}`numpy.ndarray`.  Short tutorials are provided in
285  [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/).
286- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point,
287  {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal
288  ({c:func}`CeedOperatorAssembleLinearDiagonal`).  These operations are useful for
289  preconditioning ingredients and are used in the libCEED's multigrid examples.
290- The inverse of separable operators can be obtained using
291  {c:func}`CeedOperatorCreateFDMElementInverse` and applied with
292  {c:func}`CeedOperatorApply`.  This is a useful preconditioning ingredient,
293  especially for Laplacians and related operators.
294- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`,
295  {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`.
296- Make public accessors for various attributes to facilitate writing composable code.
297- New backend: `/cpu/self/memcheck/serial`.
298- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA
299  backends.  (Single source is coming soon for OCCA backends.)
300- Fix some missing edge cases in CUDA backend.
301
302### Performance Improvements
303
304- MAGMA backend performance optimization and non-tensor bases.
305- No-copy optimization in {c:func}`CeedOperatorApply`.
306
307### Interface changes
308
309- Replace {code}`CeedElemRestrictionCreateIdentity` and
310  {code}`CeedElemRestrictionCreateBlocked` with more flexible
311  {c:func}`CeedElemRestrictionCreateStrided` and
312  {c:func}`CeedElemRestrictionCreateBlockedStrided`.
313- Add arguments to {c:func}`CeedQFunctionCreateIdentity`.
314- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification
315  with {cpp:enum}`CeedInterlaceMode`.  This is now an attribute of the
316  {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no
317  longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and
318  {c:func}`CeedElemRestrictionApply`.
319
320### Examples
321
322libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`.
323Notable additions include:
324
325- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of
326  a domain in 1, 2, and 3 dimensions by applying a Laplacian.
327
328- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area
329  of domains (like the cube and sphere) by direct integration on a surface mesh;
330  demonstrates geometric dimension different from topological dimension.
331
332- PETSc {ref}`example-petsc-bps`:
333
334  - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support.
335  - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements
336    and transparent CUDA support.
337  - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`):
338    generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric
339    dimension different from topological dimension.
340
341- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid
342  solver with algebraic multigrid coarse solve.
343
344- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly
345  `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`),
346  implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and
347  quasi-2D computational domain support.
348
349- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for
350  linear elasticity, small-strain hyperelasticity, and globalized finite-strain
351  hyperelasticity using p-multigrid with algebraic multigrid coarse solve.
352
353(v0-5)=
354
355## v0.5 (Sep 18, 2019)
356
357For this release, several improvements were made. Two new CUDA backends were added to
358the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art
359performance using single-source {ref}`CeedQFunction`. From this release, users
360can define Q-Functions in a single source code independently of the targeted backend
361with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU
362compilation of the user provided {ref}`CeedQFunction` code. To allow a unified
363declaration, the {ref}`CeedQFunction` API has undergone a slight change:
364the `QFunctionField` parameter `ncomp` has been changed to `size`. This change
365requires setting the previous value of `ncomp` to `ncomp*dim` when adding a
366`QFunctionField` with eval mode `CEED EVAL GRAD`.
367
368Additionally, new CPU backends
369were included in this release, such as the `/cpu/self/opt/*` backends (which are
370written in pure C and use partial **E-vectors** to improve performance) and the
371`/cpu/self/ref/memcheck` backend (which relies upon the
372[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user
373{ref}`CeedQFunction` have no undefined values).
374This release also included various performance improvements, bug fixes, new examples,
375and improved tests. Among these improvements, vectorized instructions for
376{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD`
377instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and
378identity Q-Functions were introduced, and the PETSc benchmark problems were expanded
379to include unstructured meshes handling were. For this expansion, the prior version of
380the PETSc BPs, which only included data associated with structured geometries, were
381renamed `bpsraw`, and the new version of the BPs, which can handle data associated
382with any unstructured geometry, were called `bps`. Additionally, other benchmark
383problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively),
384and BP5 and BP6 (the collocated versions---for which the quadrature points are the same
385as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc
386examples. Furthermoew, another standalone libCEED example, called `ex2`, which
387computes the surface area of a given mesh was added to this release.
388
389Backends available in this release:
390
391| CEED resource (`-ceed`)  | Backend                                             |
392|--------------------------|-----------------------------------------------------|
393| `/cpu/self/ref/serial`   | Serial reference implementation                     |
394| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
395| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks            |
396| `/cpu/self/opt/serial`   | Serial optimized C implementation                   |
397| `/cpu/self/opt/blocked`  | Blocked optimized C implementation                  |
398| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
399| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
400| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
401| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
402| `/cpu/occa`              | Serial OCCA kernels                                 |
403| `/gpu/occa`              | CUDA OCCA kernels                                   |
404| `/omp/occa`              | OpenMP OCCA kernels                                 |
405| `/ocl/occa`              | OpenCL OCCA kernels                                 |
406| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
407| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
408| `/gpu/cuda/shared`       | Optimized pure CUDA kernels using shared memory     |
409| `/gpu/cuda/gen`          | Optimized pure CUDA kernels using code generation   |
410| `/gpu/magma`             | CUDA MAGMA kernels                                  |
411
412Examples available in this release:
413
414:::{list-table}
415:header-rows: 1
416:widths: auto
417* - User code
418  - Example
419* - `ceed`
420  - * ex1 (volume)
421    * ex2 (surface)
422* - `mfem`
423  - * BP1 (scalar mass operator)
424    * BP3 (scalar Laplace operator)
425* - `petsc`
426  - * BP1 (scalar mass operator)
427    * BP2 (vector mass operator)
428    * BP3 (scalar Laplace operator)
429    * BP4 (vector Laplace operator)
430    * BP5 (collocated scalar Laplace operator)
431    * BP6 (collocated vector Laplace operator)
432    * Navier-Stokes
433* - `nek5000`
434  - * BP1 (scalar mass operator)
435    * BP3 (scalar Laplace operator)
436:::
437
438(v0-4)=
439
440## v0.4 (Apr 1, 2019)
441
442libCEED v0.4 was made again publicly available in the second full CEED software
443distribution, release CEED 2.0. This release contained notable features, such as
444four new CPU backends, two new GPU backends, CPU backend optimizations, initial
445support for operator composition, performance benchmarking, and a Navier-Stokes demo.
446The new CPU backends in this release came in two families. The `/cpu/self/*/serial`
447backends process one element at a time and are intended for meshes with a smaller number
448of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of
449eight interlaced elements and are intended for meshes with higher numbers of elements.
450The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU
451performance. The `/cpu/self/xsmm/*` backends rely upon the
452[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU
453performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA.
454The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable
455performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple
456parallelization approach, where each thread treats a finite element. Using just in time
457compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this
458backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend
459achieve good peak performance for 1D, 2D, and low order 3D problems, but performance
460deteriorates very quickly when threads run out of registers.
461
462A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED
463examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`).
464This example solves the time-dependent Navier-Stokes equations of compressible gas
465dynamics in a static Eulerian three-dimensional frame, using structured high-order
466finite/spectral element spatial discretizations and explicit high-order time-stepping
467(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc,
468so that the pointwise physics (defined at quadrature points) is separated from the
469parallelization and meshing concerns.
470
471Backends available in this release:
472
473| CEED resource (`-ceed`)  | Backend                                             |
474|--------------------------|-----------------------------------------------------|
475| `/cpu/self/ref/serial`   | Serial reference implementation                     |
476| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
477| `/cpu/self/tmpl`         | Backend template, defaults to `/cpu/self/blocked`   |
478| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
479| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
480| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
481| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
482| `/cpu/occa`              | Serial OCCA kernels                                 |
483| `/gpu/occa`              | CUDA OCCA kernels                                   |
484| `/omp/occa`              | OpenMP OCCA kernels                                 |
485| `/ocl/occa`              | OpenCL OCCA kernels                                 |
486| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
487| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
488| `/gpu/magma`             | CUDA MAGMA kernels                                  |
489
490Examples available in this release:
491
492:::{list-table}
493:header-rows: 1
494:widths: auto
495* - User code
496  - Example
497* - `ceed`
498  - * ex1 (volume)
499* - `mfem`
500  - * BP1 (scalar mass operator)
501    * BP3 (scalar Laplace operator)
502* - `petsc`
503  - * BP1 (scalar mass operator)
504    * BP3 (scalar Laplace operator)
505    * Navier-Stokes
506* - `nek5000`
507  - * BP1 (scalar mass operator)
508    * BP3 (scalar Laplace operator)
509:::
510
511(v0-3)=
512
513## v0.3 (Sep 30, 2018)
514
515Notable features in this release include active/passive field interface, support for
516non-tensor bases, backend optimization, and improved Fortran interface. This release
517also focused on providing improved continuous integration, and many new tests with code
518coverage reports of about 90%. This release also provided a significant change to the
519public interface: a {ref}`CeedQFunction` can take any number of named input and output
520arguments while {ref}`CeedOperator` connects them to the actual data, which may be
521supplied explicitly to `CeedOperatorApply()` (active) or separately via
522`CeedOperatorSetField()` (passive). This interface change enables reusable libraries
523of CeedQFunctions and composition of block solvers constructed using
524{ref}`CeedOperator`. A concept of blocked restriction was added to this release and
525used in an optimized CPU backend. Although this is typically not visible to the user,
526it enables effective use of arbitrary-length SIMD while maintaining cache locality.
527This CPU backend also implements an algebraic factorization of tensor product gradients
528to perform fewer operations than standard application of interpolation and
529differentiation from nodes to quadrature points. This algebraic formulation
530automatically supports non-polynomial and non-interpolatory bases, thus is more general
531than the more common derivation in terms of Lagrange polynomials on the quadrature points.
532
533Backends available in this release:
534
535| CEED resource (`-ceed`) | Backend                                             |
536|-------------------------|-----------------------------------------------------|
537| `/cpu/self/blocked`     | Blocked reference implementation                    |
538| `/cpu/self/ref`         | Serial reference implementation                     |
539| `/cpu/self/tmpl`        | Backend template, defaults to `/cpu/self/blocked`   |
540| `/cpu/occa`             | Serial OCCA kernels                                 |
541| `/gpu/occa`             | CUDA OCCA kernels                                   |
542| `/omp/occa`             | OpenMP OCCA kernels                                 |
543| `/ocl/occa`             | OpenCL OCCA kernels                                 |
544| `/gpu/magma`            | CUDA MAGMA kernels                                  |
545
546Examples available in this release:
547
548:::{list-table}
549:header-rows: 1
550:widths: auto
551* - User code
552  - Example
553* - `ceed`
554  - * ex1 (volume)
555* - `mfem`
556  - * BP1 (scalar mass operator)
557    * BP3 (scalar Laplace operator)
558* - `petsc`
559  - * BP1 (scalar mass operator)
560    * BP3 (scalar Laplace operator)
561* - `nek5000`
562  - * BP1 (scalar mass operator)
563    * BP3 (scalar Laplace operator)
564:::
565
566(v0-21)=
567
568## v0.21 (Sep 30, 2018)
569
570A MAGMA backend (which relies upon the
571[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this
572release. This initial integration set up the framework of using MAGMA and provided the
573libCEED functionality through MAGMA kernels as one of libCEED’s computational backends.
574As any other backend, the MAGMA backend provides extended basic data structures for
575{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements
576the fundamental CEED building blocks to work with the new data structures.
577In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data
578but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled
579internally, and thus seamlessly to the user, through the functions/methods that are
580provided to support them.
581
582Backends available in this release:
583
584| CEED resource (`-ceed`) | Backend                         |
585|-------------------------|---------------------------------|
586| `/cpu/self`             | Serial reference implementation |
587| `/cpu/occa`             | Serial OCCA kernels             |
588| `/gpu/occa`             | CUDA OCCA kernels               |
589| `/omp/occa`             | OpenMP OCCA kernels             |
590| `/ocl/occa`             | OpenCL OCCA kernels             |
591| `/gpu/magma`            | CUDA MAGMA kernels              |
592
593Examples available in this release:
594
595:::{list-table}
596:header-rows: 1
597:widths: auto
598* - User code
599  - Example
600* - `ceed`
601  - * ex1 (volume)
602* - `mfem`
603  - * BP1 (scalar mass operator)
604    * BP3 (scalar Laplace operator)
605* - `petsc`
606  - * BP1 (scalar mass operator)
607* - `nek5000`
608  - * BP1 (scalar mass operator)
609:::
610
611(v0-2)=
612
613## v0.2 (Mar 30, 2018)
614
615libCEED was made publicly available the first full CEED software distribution, release
616CEED 1.0. The distribution was made available using the Spack package manager to provide
617a common, easy-to-use build environment, where the user can build the CEED distribution
618with all dependencies. This release included a new Fortran interface for the library.
619This release also contained major improvements in the OCCA backend (including a new
620`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to
621compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an
622`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application
623of the Laplace operator) was also added to this release.
624
625Backends available in this release:
626
627| CEED resource (`-ceed`) | Backend                         |
628|-------------------------|---------------------------------|
629| `/cpu/self`             | Serial reference implementation |
630| `/cpu/occa`             | Serial OCCA kernels             |
631| `/gpu/occa`             | CUDA OCCA kernels               |
632| `/omp/occa`             | OpenMP OCCA kernels             |
633| `/ocl/occa`             | OpenCL OCCA kernels             |
634
635Examples available in this release:
636
637:::{list-table}
638:header-rows: 1
639:widths: auto
640* - User code
641  - Example
642* - `ceed`
643  - * ex1 (volume)
644* - `mfem`
645  - * BP1 (scalar mass operator)
646    * BP3 (scalar Laplace operator)
647* - `petsc`
648  - * BP1 (scalar mass operator)
649* - `nek5000`
650  - * BP1 (scalar mass operator)
651:::
652
653(v0-1)=
654
655## v0.1 (Jan 3, 2018)
656
657Initial low-level API of the CEED project. The low-level API provides a set of Finite
658Elements kernels and components for writing new low-level kernels. Examples include:
659vector and sparse linear algebra, element matrix assembly over a batch of elements,
660partial assembly and action for efficient high-order operators like mass, diffusion,
661advection, etc. The main goal of the low-level API is to establish the basis for the
662high-level API. Also, identifying such low-level kernels and providing a reference
663implementation for them serves as the basis for specialized backend implementations.
664This release contained several backends: `/cpu/self`, and backends which rely upon the
665[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`,
666`/gpu/occa`, and `/omp/occa`.
667It also included several examples, in the `examples` folder:
668A standalone code that shows the usage of libCEED (with no external
669dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1
670(with the application of the mass operator); and a `petsc` example to perform BP1
671(with the application of the mass operator).
672
673Backends available in this release:
674
675| CEED resource (`-ceed`) | Backend                         |
676|-------------------------|---------------------------------|
677| `/cpu/self`             | Serial reference implementation |
678| `/cpu/occa`             | Serial OCCA kernels             |
679| `/gpu/occa`             | CUDA OCCA kernels               |
680| `/omp/occa`             | OpenMP OCCA kernels             |
681
682Examples available in this release:
683
684| User code             | Example                           |
685|-----------------------|-----------------------------------|
686| `ceed`                | ex1 (scalar Laplace operator)     |
687| `mfem`                | BP1 (scalar mass operator)        |
688| `petsc`               | BP1 (scalar mass operator)        |
689```
690