xref: /libCEED/doc/sphinx/source/releasenotes.md (revision 694f24b00a5ec783a2e8f60ab930eea7b6a5ea1d)
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
14### New features
15
16- Add `CeedOperatorCreateAtPoints` which evaluates the `CeedQFunction` at arbitrary locations in each element, for use in Particle in Cell, Material Point Method, and similar methods.
17
18### Examples
19
20(v0-12)=
21
22## v0.12 (Oct 31, 2023)
23
24### Interface changes
25
26- Update `CeedOperatorContext*` functions to `CeedOperator*Context*` functions for consistency.
27For example, `CeedOperatorContextGetFieldLabel` was renamed to `CeedOperatorGetContextFieldLabel`.
28- Removed `CeedBasisSetNumQuadraturePoints` as redundant and bug-prone interface.
29
30### New features
31
32- Added {c:func}`CeedOperatorGetFieldByName` to access a specific `CeedOperatorField` by its name.
33- Update `/cpu/self/memcheck/*` backends to help verify `CeedVector` array access assumptions and `CeedQFunction` user output assumptions.
34- Update {c:func}`CeedOperatorLinearAssembleDiagonal` to provide default implementation that supports `CeedOperator` with multiple active bases.
35- Added Sycl backends `/gpu/sycl/ref`, `/gpu/sycl/shared`, and `/gpu/sycl/gen`.
36- Added {c:func}`CeedBasisApplyAtPoints` for evaluation of values and derivatives at arbitrary points inside elements.
37- Added support for non-tensor $H(\text{curl})$ finite element spaces with {c:func}`CeedBasisCreateHcurl`.
38- 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)).
39- Added {c:func}`CeedOperatorLinearAssemblePointBlockDiagonalSymbolic` to create COO mapping for mapping out of {c:func}`CeedOperatorLinearAssemblePointBlockDiagonal`.
40- Added support for application codes which manage multiple {ref}`Ceed` objects, parallelized across OpenMP threads.
41
42### Examples
43
44- Add `DMSwarm` example demonstrating interpolation from background mesh to swarm points and projection from swarm points to background mesh.
45
46#### {ref}`example-petsc-bps`
47
48- Requires PETSc version 3.19 or later.
49
50#### {ref}`example-petsc-navier-stokes`
51
52- Updated restart and checkpointing interface.
53- Add data-driven subgrid-stress model.
54- Add differential filtering of solution.
55- Add turbulence statistics collection over spanwise-symmetric geometries.
56- Add Taylor-Green vortex initial condition.
57- Add Riemann-based outflow boundary conditions.
58- Added vortex shedding and flow past cylinder example, including calculations for lift, drag, and heat transfer.
59- Add Internal Damping Layer (IDL) for helping turbulent simulation stability.
60- Derive `CeedBasis` from `PetscFE`, and various other internal maintainability updates.
61
62(v0-11)=
63
64## v0.11 (Dec 24, 2022)
65
66### Interface changes
67
68- Added {c:func}`CeedOperatorSetName` for more readable {c:func}`CeedOperatorView` output.
69- Added {c:func}`CeedBasisCreateProjection` to facilitate interpolation between nodes for separate `CeedBases`.
70- Rename and move {c:func}`CeedCompositeOperatorGetNumSub` and {c:func}`CeedCompositeOperatorGetSubList` to public interface.
71- Renamed `CEED_BASIS_COLLOCATED` to `CEED_BASIS_NONE` for clarity.
72Some 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.
73
74### New features
75
76- Update `/cpu/self/memcheck/*` backends to help verify `CeedQFunctionContext` data sizes provided by user.
77- Improved support for $H(\text{div})$ bases.
78- Added `CeedInt_FMT` to support potential future use of larger integer sizes.
79- Added `CEED_QFUNCTION_ATTR` for setting compiler attributes/pragmas to `CEED_QFUNCTION_HELPER` and `CEED_QFUNCTION`.
80- OCCA backend updated to latest OCCA release; DPC++ and OMP OCCA modes enabled.
81Due to a limitation of the OCCA parser, typedefs are required to use pointers to arrays in QFunctions with the OCCA backend.
82This issue will be fixed in a future OCCA release.
83
84### Bugfix
85
86- Fix bug in setting device id for GPU backends.
87- Fix storing of indices for `CeedElemRestriction` on the host with GPU backends.
88- Fix `CeedElemRestriction` sizing for {c:func}`CeedOperatorAssemblePointBlockDiagonal`.
89- Fix bugs in CPU implementation of {c:func}`CeedOperatorLinearAssemble` when there are different number of active input modes and active output modes.
90
91### Examples
92
93#### {ref}`example-petsc-navier-stokes`
94
95- Various performance enhancements, analytic matrix-free and assembled Jacobian, and PETSc solver configurations for GPUs.
96- Refactored to improve code reuse and modularity.
97- Support for primitive variables for more accurate boundary layers and all-speed flow.
98- Added $YZ\beta$ shock capturing scheme and Shock Tube example.
99- Added Channel example, with comparison to analytic solutions.
100- Added Flat Plate with boundary layer mesh and compressible Blasius inflow condition based on Chebyshev collocation solution of the Blasius equations.
101- Added strong and weak synthetic turbulence generation (STG) inflow boundary conditions.
102- Added "freestream" boundary conditions based on HLLC Riemann solver.
103- Automated stabilization coefficients for different basis degree.
104
105#### {ref}`example-petsc-bps`
106
107- Support for convergence studies.
108
109### Maintainability
110
111- 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.
112- Enabled support for `p > 8` for `/gpu/*/shared` backends.
113- Switch to `clang-format` over `astyle` for automatic formatting; Makefile command changed to `make format` from `make style`.
114- Improved test harness.
115
116(v0-10-1)=
117
118## v0.10.1 (Apr 11, 2022)
119
120### Interface changes
121
122- Added {c:func}`CeedQFunctionSetUserFlopsEstimate` and {c:func}`CeedOperatorGetFlopsEstimate` to facilitate estimating FLOPs in operator application.
123
124### New features
125
126- Switched MAGMA backends to use runtime compilation for tensor basis kernels (and element restriction kernels, in non-deterministic `/gpu/*/magma` backends).
127This reduces time to compile the library and increases the range of parameters for which the MAGMA tensor basis kernels will work.
128
129### Bugfix
130
131- Install JiT source files in install directory to fix GPU functionality for installed libCEED.
132
133(v0-10)=
134
135## v0.10 (Mar 21, 2022)
136
137### Interface changes
138
139- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields.
140- 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`.
141- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added.
142- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients.
143- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces.
144- Warning added when compiling OCCA backend to alert users that this backend is experimental.
145- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`.
146- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name.
147- 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`.
148- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitialized arrays; require initialized data for {c:func}`CeedVectorGetArray`.
149- 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.
150- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retrieve user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`.
151- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums.
152- 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.
153- Added {c:func}`CeedOperatorSetQFunctionUpdated` to facilitate QFunction data re-use between operators sharing the same quadrature space, such as in a multigrid hierarchy.
154- Added {c:func}`CeedOperatorGetActiveVectorLengths` to get shape of CeedOperator.
155
156### New features
157
158- `CeedScalar` can now be set as `float` or `double` at compile time.
159- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends.
160- 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.
161- Remove need to guard library headers in QFunction source for code generation backends.
162- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present.
163- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally.
164- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file.
165- Added support for non-tensor $H(\text{div})$ elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor.
166- 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.
167- Added support for element matrix assembly in GPU backends.
168
169### Maintainability
170
171- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage.
172- `Include-what-you-use` makefile target added as `make iwyu`.
173- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase.
174- Put GPU JiTed kernel source code into separate files.
175- 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.
176
177(v0-9)=
178
179## v0.9 (Jul 6, 2021)
180
181### Interface changes
182
183- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact.
184
185### New features
186
187- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use.
188- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends.
189- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations.
190- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed).
191
192### Performance improvements
193
194### Examples
195
196- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations.
197- Fluid mechanics example adds GPU support and improves modularity.
198
199### Deprecated backends
200
201- 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.
202
203(v0-8)=
204
205## v0.8 (Mar 31, 2021)
206
207### Interface changes
208
209- Error handling improved to include enumerated error codes for C interface return values.
210- 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.
211
212### New features
213
214- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features.
215- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly.
216- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators.
217
218### Performance improvements
219
220- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`.
221- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`.
222
223### Examples
224
225- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions.
226- {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.
227- {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.
228- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree.
229
230(v0-7)=
231
232## v0.7 (Sep 29, 2020)
233
234### Interface changes
235
236- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors.
237  As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`.
238  These changes improve support for mixed finite element methods.
239- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature.
240- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity.
241- 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.
242- 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.
243- 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.
244- 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.
245  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`.
246- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory.
247- 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.
248
249### New features
250
251- New HIP backend: `/gpu/hip/ref`.
252- CeedQFunction support for user `CUfunction`s in some backends
253
254### Performance improvements
255
256- OCCA backend rebuilt to facilitate future performance enhancements.
257- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`.
258
259### Examples
260
261- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions.
262
263### Deprecated backends
264
265- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`.
266
267(v0-6)=
268
269## v0.6 (Mar 29, 2020)
270
271libCEED v0.6 contains numerous new features and examples, as well as expanded
272documentation in [this new website](https://libceed.org).
273
274### New features
275
276- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly
277  1-1 correspondence with the C interface, plus some convenience features.  For instance,
278  data stored in the {cpp:type}`CeedVector` structure are available without copy as
279  {py:class}`numpy.ndarray`.  Short tutorials are provided in
280  [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/).
281- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point,
282  {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal
283  ({c:func}`CeedOperatorAssembleLinearDiagonal`).  These operations are useful for
284  preconditioning ingredients and are used in the libCEED's multigrid examples.
285- The inverse of separable operators can be obtained using
286  {c:func}`CeedOperatorCreateFDMElementInverse` and applied with
287  {c:func}`CeedOperatorApply`.  This is a useful preconditioning ingredient,
288  especially for Laplacians and related operators.
289- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`,
290  {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`.
291- Make public accessors for various attributes to facilitate writing composable code.
292- New backend: `/cpu/self/memcheck/serial`.
293- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA
294  backends.  (Single source is coming soon for OCCA backends.)
295- Fix some missing edge cases in CUDA backend.
296
297### Performance Improvements
298
299- MAGMA backend performance optimization and non-tensor bases.
300- No-copy optimization in {c:func}`CeedOperatorApply`.
301
302### Interface changes
303
304- Replace {code}`CeedElemRestrictionCreateIdentity` and
305  {code}`CeedElemRestrictionCreateBlocked` with more flexible
306  {c:func}`CeedElemRestrictionCreateStrided` and
307  {c:func}`CeedElemRestrictionCreateBlockedStrided`.
308- Add arguments to {c:func}`CeedQFunctionCreateIdentity`.
309- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification
310  with {cpp:enum}`CeedInterlaceMode`.  This is now an attribute of the
311  {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no
312  longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and
313  {c:func}`CeedElemRestrictionApply`.
314
315### Examples
316
317libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`.
318Notable additions include:
319
320- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of
321  a domain in 1, 2, and 3 dimensions by applying a Laplacian.
322
323- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area
324  of domains (like the cube and sphere) by direct integration on a surface mesh;
325  demonstrates geometric dimension different from topological dimension.
326
327- PETSc {ref}`example-petsc-bps`:
328
329  - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support.
330  - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements
331    and transparent CUDA support.
332  - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`):
333    generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric
334    dimension different from topological dimension.
335
336- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid
337  solver with algebraic multigrid coarse solve.
338
339- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly
340  `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`),
341  implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and
342  quasi-2D computational domain support.
343
344- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for
345  linear elasticity, small-strain hyperelasticity, and globalized finite-strain
346  hyperelasticity using p-multigrid with algebraic multigrid coarse solve.
347
348(v0-5)=
349
350## v0.5 (Sep 18, 2019)
351
352For this release, several improvements were made. Two new CUDA backends were added to
353the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art
354performance using single-source {ref}`CeedQFunction`. From this release, users
355can define Q-Functions in a single source code independently of the targeted backend
356with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU
357compilation of the user provided {ref}`CeedQFunction` code. To allow a unified
358declaration, the {ref}`CeedQFunction` API has undergone a slight change:
359the `QFunctionField` parameter `ncomp` has been changed to `size`. This change
360requires setting the previous value of `ncomp` to `ncomp*dim` when adding a
361`QFunctionField` with eval mode `CEED EVAL GRAD`.
362
363Additionally, new CPU backends
364were included in this release, such as the `/cpu/self/opt/*` backends (which are
365written in pure C and use partial **E-vectors** to improve performance) and the
366`/cpu/self/ref/memcheck` backend (which relies upon the
367[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user
368{ref}`CeedQFunction` have no undefined values).
369This release also included various performance improvements, bug fixes, new examples,
370and improved tests. Among these improvements, vectorized instructions for
371{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD`
372instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and
373identity Q-Functions were introduced, and the PETSc benchmark problems were expanded
374to include unstructured meshes handling were. For this expansion, the prior version of
375the PETSc BPs, which only included data associated with structured geometries, were
376renamed `bpsraw`, and the new version of the BPs, which can handle data associated
377with any unstructured geometry, were called `bps`. Additionally, other benchmark
378problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively),
379and BP5 and BP6 (the collocated versions---for which the quadrature points are the same
380as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc
381examples. Furthermoew, another standalone libCEED example, called `ex2`, which
382computes the surface area of a given mesh was added to this release.
383
384Backends available in this release:
385
386| CEED resource (`-ceed`)  | Backend                                             |
387|--------------------------|-----------------------------------------------------|
388| `/cpu/self/ref/serial`   | Serial reference implementation                     |
389| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
390| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks            |
391| `/cpu/self/opt/serial`   | Serial optimized C implementation                   |
392| `/cpu/self/opt/blocked`  | Blocked optimized C implementation                  |
393| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
394| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
395| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
396| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
397| `/cpu/occa`              | Serial OCCA kernels                                 |
398| `/gpu/occa`              | CUDA OCCA kernels                                   |
399| `/omp/occa`              | OpenMP OCCA kernels                                 |
400| `/ocl/occa`              | OpenCL OCCA kernels                                 |
401| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
402| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
403| `/gpu/cuda/shared`       | Optimized pure CUDA kernels using shared memory     |
404| `/gpu/cuda/gen`          | Optimized pure CUDA kernels using code generation   |
405| `/gpu/magma`             | CUDA MAGMA kernels                                  |
406
407Examples available in this release:
408
409:::{list-table}
410:header-rows: 1
411:widths: auto
412* - User code
413  - Example
414* - `ceed`
415  - * ex1 (volume)
416    * ex2 (surface)
417* - `mfem`
418  - * BP1 (scalar mass operator)
419    * BP3 (scalar Laplace operator)
420* - `petsc`
421  - * BP1 (scalar mass operator)
422    * BP2 (vector mass operator)
423    * BP3 (scalar Laplace operator)
424    * BP4 (vector Laplace operator)
425    * BP5 (collocated scalar Laplace operator)
426    * BP6 (collocated vector Laplace operator)
427    * Navier-Stokes
428* - `nek5000`
429  - * BP1 (scalar mass operator)
430    * BP3 (scalar Laplace operator)
431:::
432
433(v0-4)=
434
435## v0.4 (Apr 1, 2019)
436
437libCEED v0.4 was made again publicly available in the second full CEED software
438distribution, release CEED 2.0. This release contained notable features, such as
439four new CPU backends, two new GPU backends, CPU backend optimizations, initial
440support for operator composition, performance benchmarking, and a Navier-Stokes demo.
441The new CPU backends in this release came in two families. The `/cpu/self/*/serial`
442backends process one element at a time and are intended for meshes with a smaller number
443of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of
444eight interlaced elements and are intended for meshes with higher numbers of elements.
445The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU
446performance. The `/cpu/self/xsmm/*` backends rely upon the
447[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU
448performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA.
449The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable
450performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple
451parallelization approach, where each thread treats a finite element. Using just in time
452compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this
453backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend
454achieve good peak performance for 1D, 2D, and low order 3D problems, but performance
455deteriorates very quickly when threads run out of registers.
456
457A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED
458examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`).
459This example solves the time-dependent Navier-Stokes equations of compressible gas
460dynamics in a static Eulerian three-dimensional frame, using structured high-order
461finite/spectral element spatial discretizations and explicit high-order time-stepping
462(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc,
463so that the pointwise physics (defined at quadrature points) is separated from the
464parallelization and meshing concerns.
465
466Backends available in this release:
467
468| CEED resource (`-ceed`)  | Backend                                             |
469|--------------------------|-----------------------------------------------------|
470| `/cpu/self/ref/serial`   | Serial reference implementation                     |
471| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
472| `/cpu/self/tmpl`         | Backend template, defaults to `/cpu/self/blocked`   |
473| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
474| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
475| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
476| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
477| `/cpu/occa`              | Serial OCCA kernels                                 |
478| `/gpu/occa`              | CUDA OCCA kernels                                   |
479| `/omp/occa`              | OpenMP OCCA kernels                                 |
480| `/ocl/occa`              | OpenCL OCCA kernels                                 |
481| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
482| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
483| `/gpu/magma`             | CUDA MAGMA kernels                                  |
484
485Examples available in this release:
486
487:::{list-table}
488:header-rows: 1
489:widths: auto
490* - User code
491  - Example
492* - `ceed`
493  - * ex1 (volume)
494* - `mfem`
495  - * BP1 (scalar mass operator)
496    * BP3 (scalar Laplace operator)
497* - `petsc`
498  - * BP1 (scalar mass operator)
499    * BP3 (scalar Laplace operator)
500    * Navier-Stokes
501* - `nek5000`
502  - * BP1 (scalar mass operator)
503    * BP3 (scalar Laplace operator)
504:::
505
506(v0-3)=
507
508## v0.3 (Sep 30, 2018)
509
510Notable features in this release include active/passive field interface, support for
511non-tensor bases, backend optimization, and improved Fortran interface. This release
512also focused on providing improved continuous integration, and many new tests with code
513coverage reports of about 90%. This release also provided a significant change to the
514public interface: a {ref}`CeedQFunction` can take any number of named input and output
515arguments while {ref}`CeedOperator` connects them to the actual data, which may be
516supplied explicitly to `CeedOperatorApply()` (active) or separately via
517`CeedOperatorSetField()` (passive). This interface change enables reusable libraries
518of CeedQFunctions and composition of block solvers constructed using
519{ref}`CeedOperator`. A concept of blocked restriction was added to this release and
520used in an optimized CPU backend. Although this is typically not visible to the user,
521it enables effective use of arbitrary-length SIMD while maintaining cache locality.
522This CPU backend also implements an algebraic factorization of tensor product gradients
523to perform fewer operations than standard application of interpolation and
524differentiation from nodes to quadrature points. This algebraic formulation
525automatically supports non-polynomial and non-interpolatory bases, thus is more general
526than the more common derivation in terms of Lagrange polynomials on the quadrature points.
527
528Backends available in this release:
529
530| CEED resource (`-ceed`) | Backend                                             |
531|-------------------------|-----------------------------------------------------|
532| `/cpu/self/blocked`     | Blocked reference implementation                    |
533| `/cpu/self/ref`         | Serial reference implementation                     |
534| `/cpu/self/tmpl`        | Backend template, defaults to `/cpu/self/blocked`   |
535| `/cpu/occa`             | Serial OCCA kernels                                 |
536| `/gpu/occa`             | CUDA OCCA kernels                                   |
537| `/omp/occa`             | OpenMP OCCA kernels                                 |
538| `/ocl/occa`             | OpenCL OCCA kernels                                 |
539| `/gpu/magma`            | CUDA MAGMA kernels                                  |
540
541Examples available in this release:
542
543:::{list-table}
544:header-rows: 1
545:widths: auto
546* - User code
547  - Example
548* - `ceed`
549  - * ex1 (volume)
550* - `mfem`
551  - * BP1 (scalar mass operator)
552    * BP3 (scalar Laplace operator)
553* - `petsc`
554  - * BP1 (scalar mass operator)
555    * BP3 (scalar Laplace operator)
556* - `nek5000`
557  - * BP1 (scalar mass operator)
558    * BP3 (scalar Laplace operator)
559:::
560
561(v0-21)=
562
563## v0.21 (Sep 30, 2018)
564
565A MAGMA backend (which relies upon the
566[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this
567release. This initial integration set up the framework of using MAGMA and provided the
568libCEED functionality through MAGMA kernels as one of libCEED’s computational backends.
569As any other backend, the MAGMA backend provides extended basic data structures for
570{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements
571the fundamental CEED building blocks to work with the new data structures.
572In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data
573but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled
574internally, and thus seamlessly to the user, through the functions/methods that are
575provided to support them.
576
577Backends available in this release:
578
579| CEED resource (`-ceed`) | Backend                         |
580|-------------------------|---------------------------------|
581| `/cpu/self`             | Serial reference implementation |
582| `/cpu/occa`             | Serial OCCA kernels             |
583| `/gpu/occa`             | CUDA OCCA kernels               |
584| `/omp/occa`             | OpenMP OCCA kernels             |
585| `/ocl/occa`             | OpenCL OCCA kernels             |
586| `/gpu/magma`            | CUDA MAGMA kernels              |
587
588Examples available in this release:
589
590:::{list-table}
591:header-rows: 1
592:widths: auto
593* - User code
594  - Example
595* - `ceed`
596  - * ex1 (volume)
597* - `mfem`
598  - * BP1 (scalar mass operator)
599    * BP3 (scalar Laplace operator)
600* - `petsc`
601  - * BP1 (scalar mass operator)
602* - `nek5000`
603  - * BP1 (scalar mass operator)
604:::
605
606(v0-2)=
607
608## v0.2 (Mar 30, 2018)
609
610libCEED was made publicly available the first full CEED software distribution, release
611CEED 1.0. The distribution was made available using the Spack package manager to provide
612a common, easy-to-use build environment, where the user can build the CEED distribution
613with all dependencies. This release included a new Fortran interface for the library.
614This release also contained major improvements in the OCCA backend (including a new
615`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to
616compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an
617`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application
618of the Laplace operator) was also added to this release.
619
620Backends available in this release:
621
622| CEED resource (`-ceed`) | Backend                         |
623|-------------------------|---------------------------------|
624| `/cpu/self`             | Serial reference implementation |
625| `/cpu/occa`             | Serial OCCA kernels             |
626| `/gpu/occa`             | CUDA OCCA kernels               |
627| `/omp/occa`             | OpenMP OCCA kernels             |
628| `/ocl/occa`             | OpenCL OCCA kernels             |
629
630Examples available in this release:
631
632:::{list-table}
633:header-rows: 1
634:widths: auto
635* - User code
636  - Example
637* - `ceed`
638  - * ex1 (volume)
639* - `mfem`
640  - * BP1 (scalar mass operator)
641    * BP3 (scalar Laplace operator)
642* - `petsc`
643  - * BP1 (scalar mass operator)
644* - `nek5000`
645  - * BP1 (scalar mass operator)
646:::
647
648(v0-1)=
649
650## v0.1 (Jan 3, 2018)
651
652Initial low-level API of the CEED project. The low-level API provides a set of Finite
653Elements kernels and components for writing new low-level kernels. Examples include:
654vector and sparse linear algebra, element matrix assembly over a batch of elements,
655partial assembly and action for efficient high-order operators like mass, diffusion,
656advection, etc. The main goal of the low-level API is to establish the basis for the
657high-level API. Also, identifying such low-level kernels and providing a reference
658implementation for them serves as the basis for specialized backend implementations.
659This release contained several backends: `/cpu/self`, and backends which rely upon the
660[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`,
661`/gpu/occa`, and `/omp/occa`.
662It also included several examples, in the `examples` folder:
663A standalone code that shows the usage of libCEED (with no external
664dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1
665(with the application of the mass operator); and a `petsc` example to perform BP1
666(with the application of the mass operator).
667
668Backends available in this release:
669
670| CEED resource (`-ceed`) | Backend                         |
671|-------------------------|---------------------------------|
672| `/cpu/self`             | Serial reference implementation |
673| `/cpu/occa`             | Serial OCCA kernels             |
674| `/gpu/occa`             | CUDA OCCA kernels               |
675| `/omp/occa`             | OpenMP OCCA kernels             |
676
677Examples available in this release:
678
679| User code             | Example                           |
680|-----------------------|-----------------------------------|
681| `ceed`                | ex1 (scalar Laplace operator)     |
682| `mfem`                | BP1 (scalar mass operator)        |
683| `petsc`               | BP1 (scalar mass operator)        |
684```
685