xref: /libCEED/doc/sphinx/source/releasenotes.md (revision 121d4b7fb417fe929bc878bd436cd4820a8e3053)
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- Added {c:func}`CeedOperatorSetName` for more readable {c:func}`CeedOperatorView` output.
12- Added {c:func}`CeedBasisCreateProjection` to facilitate interpolation between nodes for separate `CeedBases`.
13
14### New features
15
16- Update `/cpu/self/memcheck/*` backends to help verify `CeedQFunctionContext` data sizes provided by user.
17
18### Bugfix
19
20- Fix bug in setting device id for GPU backends.
21- Fix storing of indices for `CeedElemRestriction` on the host with GPU backends.
22- Fix `CeedElemRestriction` sizing for {c:func}`CeedOperatorAssemblePointBlockDiagonal`.
23- Fix bugs in CPU implementation of {c:func}`CeedOperatorLinearAssemble` when there are different number of active input modes and active output modes.
24
25### Examples
26
27- Added various performance enhancements for {ref}`example-petsc-navier-stokes`.
28- Refactored {ref}`example-petsc-navier-stokes` to improve code reuse.
29- Added Shock Tube, Channel, and Flat Plate boundary layer problems to {ref}`example-petsc-navier-stokes`.
30
31(v0-10-1)=
32
33## v0.10.1 (Apr 11, 2022)
34
35### Interface changes
36
37- Added {c:func}`CeedQFunctionSetUserFlopsEstimate` and {c:func}`CeedOperatorGetFlopsEstimate` to facilitate estimating FLOPs in operator application.
38
39### New features
40
41- Switched MAGMA backends to use runtime compilation for tensor basis kernels (and element restriction kernels, in non-deterministic `/gpu/*/magma` backends).
42This reduces time to compile the library and increases the range of parameters for which the MAGMA tensor basis kernels will work.
43
44### Bugfix
45
46- Install JiT source files in install directory to fix GPU functionality for installed libCEED.
47
48(v0-10)=
49
50## v0.10 (Mar 21, 2022)
51
52### Interface changes
53
54- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields.
55- 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`.
56- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added.
57- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients.
58- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces.
59- Warning added when compiling OCCA backend to alert users that this backend is experimental.
60- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`.
61- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name.
62- 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`.
63- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitalized arrays; require initalized data for {c:func}`CeedVectorGetArray`.
64- 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.
65- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retreive user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`.
66- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums.
67- 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.
68- Added {c:func}`CeedOperatorSetQFunctionUpdated` to facilitate QFunction data re-use between operators sharing the same quadrature space, such as in a multigrid hierarchy.
69- Added {c:func}`CeedOperatorGetActiveVectorLengths` to get shape of CeedOperator.
70
71### New features
72
73- `CeedScalar` can now be set as `float` or `double` at compile time.
74- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends.
75- 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.
76- Remove need to guard library headers in QFunction source for code generation backends.
77- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present.
78- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally.
79- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file.
80- Added support for non-tensor H(div) elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor.
81- 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.
82- Added support for element matrix assembly in GPU backends.
83
84### Maintainability
85
86- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage.
87- `Include-what-you-use` makefile target added as `make iwyu`.
88- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase.
89- Put GPU JiTed kernel source code into separate files.
90- 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.
91
92(v0-9)=
93
94## v0.9 (Jul 6, 2021)
95
96### Interface changes
97
98- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact.
99
100### New features
101
102- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use.
103- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends.
104- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations.
105- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed).
106
107### Performance improvements
108
109### Examples
110
111- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations.
112- Fluid mechanics example adds GPU support and improves modularity.
113
114### Deprecated backends
115
116- 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.
117
118(v0-8)=
119
120## v0.8 (Mar 31, 2021)
121
122### Interface changes
123
124- Error handling improved to include enumerated error codes for C interface return values.
125- 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.
126
127### New features
128
129- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features.
130- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly.
131- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators.
132
133### Performance improvements
134
135- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`.
136- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`.
137
138### Examples
139
140- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions.
141- {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.
142- {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.
143- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree.
144
145(v0-7)=
146
147## v0.7 (Sep 29, 2020)
148
149### Interface changes
150
151- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors.
152  As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`.
153  These changes improve support for mixed finite element methods.
154- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature.
155- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity.
156- 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.
157- 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.
158- 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.
159- 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.
160  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`.
161- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory.
162- 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.
163
164### New features
165
166- New HIP backend: `/gpu/hip/ref`.
167- CeedQFunction support for user `CUfunction`s in some backends
168
169### Performance improvements
170
171- OCCA backend rebuilt to facilitate future performance enhancements.
172- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`.
173
174### Examples
175
176- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions.
177
178### Deprecated backends
179
180- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`.
181
182(v0-6)=
183
184## v0.6 (Mar 29, 2020)
185
186libCEED v0.6 contains numerous new features and examples, as well as expanded
187documentation in [this new website](https://libceed.org).
188
189### New features
190
191- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly
192  1-1 correspondence with the C interface, plus some convenience features.  For instance,
193  data stored in the {cpp:type}`CeedVector` structure are available without copy as
194  {py:class}`numpy.ndarray`.  Short tutorials are provided in
195  [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/).
196- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point,
197  {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal
198  ({c:func}`CeedOperatorAssembleLinearDiagonal`).  These operations are useful for
199  preconditioning ingredients and are used in the libCEED's multigrid examples.
200- The inverse of separable operators can be obtained using
201  {c:func}`CeedOperatorCreateFDMElementInverse` and applied with
202  {c:func}`CeedOperatorApply`.  This is a useful preconditioning ingredient,
203  especially for Laplacians and related operators.
204- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`,
205  {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`.
206- Make public accessors for various attributes to facilitate writing composable code.
207- New backend: `/cpu/self/memcheck/serial`.
208- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA
209  backends.  (Single source is coming soon for OCCA backends.)
210- Fix some missing edge cases in CUDA backend.
211
212### Performance Improvements
213
214- MAGMA backend performance optimization and non-tensor bases.
215- No-copy optimization in {c:func}`CeedOperatorApply`.
216
217### Interface changes
218
219- Replace {code}`CeedElemRestrictionCreateIdentity` and
220  {code}`CeedElemRestrictionCreateBlocked` with more flexible
221  {c:func}`CeedElemRestrictionCreateStrided` and
222  {c:func}`CeedElemRestrictionCreateBlockedStrided`.
223- Add arguments to {c:func}`CeedQFunctionCreateIdentity`.
224- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification
225  with {cpp:enum}`CeedInterlaceMode`.  This is now an attribute of the
226  {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no
227  longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and
228  {c:func}`CeedElemRestrictionApply`.
229
230### Examples
231
232libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`.
233Notable additions include:
234
235- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of
236  a domain in 1, 2, and 3 dimensions by applying a Laplacian.
237
238- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area
239  of domains (like the cube and sphere) by direct integration on a surface mesh;
240  demonstrates geometric dimension different from topological dimension.
241
242- PETSc {ref}`example-petsc-bps`:
243
244  - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support.
245  - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements
246    and transparent CUDA support.
247  - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`):
248    generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric
249    dimension different from topological dimension.
250
251- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid
252  solver with algebraic multigrid coarse solve.
253
254- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly
255  `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`),
256  implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and
257  quasi-2D computational domain support.
258
259- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for
260  linear elasticity, small-strain hyperelasticity, and globalized finite-strain
261  hyperelasticity using p-multigrid with algebraic multigrid coarse solve.
262
263(v0-5)=
264
265## v0.5 (Sep 18, 2019)
266
267For this release, several improvements were made. Two new CUDA backends were added to
268the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art
269performance using single-source {ref}`CeedQFunction`. From this release, users
270can define Q-Functions in a single source code independently of the targeted backend
271with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU
272compilation of the user provided {ref}`CeedQFunction` code. To allow a unified
273declaration, the {ref}`CeedQFunction` API has undergone a slight change:
274the `QFunctionField` parameter `ncomp` has been changed to `size`. This change
275requires setting the previous value of `ncomp` to `ncomp*dim` when adding a
276`QFunctionField` with eval mode `CEED EVAL GRAD`.
277
278Additionally, new CPU backends
279were included in this release, such as the `/cpu/self/opt/*` backends (which are
280written in pure C and use partial **E-vectors** to improve performance) and the
281`/cpu/self/ref/memcheck` backend (which relies upon the
282[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user
283{ref}`CeedQFunction` have no undefined values).
284This release also included various performance improvements, bug fixes, new examples,
285and improved tests. Among these improvements, vectorized instructions for
286{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD`
287instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and
288identity Q-Functions were introduced, and the PETSc benchmark problems were expanded
289to include unstructured meshes handling were. For this expansion, the prior version of
290the PETSc BPs, which only included data associated with structured geometries, were
291renamed `bpsraw`, and the new version of the BPs, which can handle data associated
292with any unstructured geometry, were called `bps`. Additionally, other benchmark
293problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively),
294and BP5 and BP6 (the collocated versions---for which the quadrature points are the same
295as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc
296examples. Furthermoew, another standalone libCEED example, called `ex2`, which
297computes the surface area of a given mesh was added to this release.
298
299Backends available in this release:
300
301| CEED resource (`-ceed`)  | Backend                                             |
302|--------------------------|-----------------------------------------------------|
303| `/cpu/self/ref/serial`   | Serial reference implementation                     |
304| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
305| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks            |
306| `/cpu/self/opt/serial`   | Serial optimized C implementation                   |
307| `/cpu/self/opt/blocked`  | Blocked optimized C implementation                  |
308| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
309| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
310| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
311| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
312| `/cpu/occa`              | Serial OCCA kernels                                 |
313| `/gpu/occa`              | CUDA OCCA kernels                                   |
314| `/omp/occa`              | OpenMP OCCA kernels                                 |
315| `/ocl/occa`              | OpenCL OCCA kernels                                 |
316| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
317| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
318| `/gpu/cuda/shared`       | Optimized pure CUDA kernels using shared memory     |
319| `/gpu/cuda/gen`          | Optimized pure CUDA kernels using code generation   |
320| `/gpu/magma`             | CUDA MAGMA kernels                                  |
321
322Examples available in this release:
323
324:::{list-table}
325:header-rows: 1
326:widths: auto
327* - User code
328  - Example
329* - `ceed`
330  - * ex1 (volume)
331    * ex2 (surface)
332* - `mfem`
333  - * BP1 (scalar mass operator)
334    * BP3 (scalar Laplace operator)
335* - `petsc`
336  - * BP1 (scalar mass operator)
337    * BP2 (vector mass operator)
338    * BP3 (scalar Laplace operator)
339    * BP4 (vector Laplace operator)
340    * BP5 (collocated scalar Laplace operator)
341    * BP6 (collocated vector Laplace operator)
342    * Navier-Stokes
343* - `nek5000`
344  - * BP1 (scalar mass operator)
345    * BP3 (scalar Laplace operator)
346:::
347
348(v0-4)=
349
350## v0.4 (Apr 1, 2019)
351
352libCEED v0.4 was made again publicly available in the second full CEED software
353distribution, release CEED 2.0. This release contained notable features, such as
354four new CPU backends, two new GPU backends, CPU backend optimizations, initial
355support for operator composition, performance benchmarking, and a Navier-Stokes demo.
356The new CPU backends in this release came in two families. The `/cpu/self/*/serial`
357backends process one element at a time and are intended for meshes with a smaller number
358of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of
359eight interlaced elements and are intended for meshes with higher numbers of elements.
360The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU
361performance. The `/cpu/self/xsmm/*` backends rely upon the
362[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU
363performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA.
364The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable
365performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple
366parallelization approach, where each thread treats a finite element. Using just in time
367compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this
368backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend
369achieve good peak performance for 1D, 2D, and low order 3D problems, but performance
370deteriorates very quickly when threads run out of registers.
371
372A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED
373examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`).
374This example solves the time-dependent Navier-Stokes equations of compressible gas
375dynamics in a static Eulerian three-dimensional frame, using structured high-order
376finite/spectral element spatial discretizations and explicit high-order time-stepping
377(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc,
378so that the pointwise physics (defined at quadrature points) is separated from the
379parallelization and meshing concerns.
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/tmpl`         | Backend template, defaults to `/cpu/self/blocked`   |
388| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
389| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
390| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
391| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
392| `/cpu/occa`              | Serial OCCA kernels                                 |
393| `/gpu/occa`              | CUDA OCCA kernels                                   |
394| `/omp/occa`              | OpenMP OCCA kernels                                 |
395| `/ocl/occa`              | OpenCL OCCA kernels                                 |
396| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
397| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
398| `/gpu/magma`             | CUDA MAGMA kernels                                  |
399
400Examples available in this release:
401
402:::{list-table}
403:header-rows: 1
404:widths: auto
405* - User code
406  - Example
407* - `ceed`
408  - * ex1 (volume)
409* - `mfem`
410  - * BP1 (scalar mass operator)
411    * BP3 (scalar Laplace operator)
412* - `petsc`
413  - * BP1 (scalar mass operator)
414    * BP3 (scalar Laplace operator)
415    * Navier-Stokes
416* - `nek5000`
417  - * BP1 (scalar mass operator)
418    * BP3 (scalar Laplace operator)
419:::
420
421(v0-3)=
422
423## v0.3 (Sep 30, 2018)
424
425Notable features in this release include active/passive field interface, support for
426non-tensor bases, backend optimization, and improved Fortran interface. This release
427also focused on providing improved continuous integration, and many new tests with code
428coverage reports of about 90%. This release also provided a significant change to the
429public interface: a {ref}`CeedQFunction` can take any number of named input and output
430arguments while {ref}`CeedOperator` connects them to the actual data, which may be
431supplied explicitly to `CeedOperatorApply()` (active) or separately via
432`CeedOperatorSetField()` (passive). This interface change enables reusable libraries
433of CeedQFunctions and composition of block solvers constructed using
434{ref}`CeedOperator`. A concept of blocked restriction was added to this release and
435used in an optimized CPU backend. Although this is typically not visible to the user,
436it enables effective use of arbitrary-length SIMD while maintaining cache locality.
437This CPU backend also implements an algebraic factorization of tensor product gradients
438to perform fewer operations than standard application of interpolation and
439differentiation from nodes to quadrature points. This algebraic formulation
440automatically supports non-polynomial and non-interpolatory bases, thus is more general
441than the more common derivation in terms of Lagrange polynomials on the quadrature points.
442
443Backends available in this release:
444
445| CEED resource (`-ceed`) | Backend                                             |
446|-------------------------|-----------------------------------------------------|
447| `/cpu/self/blocked`     | Blocked reference implementation                    |
448| `/cpu/self/ref`         | Serial reference implementation                     |
449| `/cpu/self/tmpl`        | Backend template, defaults to `/cpu/self/blocked`   |
450| `/cpu/occa`             | Serial OCCA kernels                                 |
451| `/gpu/occa`             | CUDA OCCA kernels                                   |
452| `/omp/occa`             | OpenMP OCCA kernels                                 |
453| `/ocl/occa`             | OpenCL OCCA kernels                                 |
454| `/gpu/magma`            | CUDA MAGMA kernels                                  |
455
456Examples available in this release:
457
458:::{list-table}
459:header-rows: 1
460:widths: auto
461* - User code
462  - Example
463* - `ceed`
464  - * ex1 (volume)
465* - `mfem`
466  - * BP1 (scalar mass operator)
467    * BP3 (scalar Laplace operator)
468* - `petsc`
469  - * BP1 (scalar mass operator)
470    * BP3 (scalar Laplace operator)
471* - `nek5000`
472  - * BP1 (scalar mass operator)
473    * BP3 (scalar Laplace operator)
474:::
475
476(v0-21)=
477
478## v0.21 (Sep 30, 2018)
479
480A MAGMA backend (which relies upon the
481[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this
482release. This initial integration set up the framework of using MAGMA and provided the
483libCEED functionality through MAGMA kernels as one of libCEED’s computational backends.
484As any other backend, the MAGMA backend provides extended basic data structures for
485{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements
486the fundamental CEED building blocks to work with the new data structures.
487In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data
488but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled
489internally, and thus seamlessly to the user, through the functions/methods that are
490provided to support them.
491
492Backends available in this release:
493
494| CEED resource (`-ceed`) | Backend                         |
495|-------------------------|---------------------------------|
496| `/cpu/self`             | Serial reference implementation |
497| `/cpu/occa`             | Serial OCCA kernels             |
498| `/gpu/occa`             | CUDA OCCA kernels               |
499| `/omp/occa`             | OpenMP OCCA kernels             |
500| `/ocl/occa`             | OpenCL OCCA kernels             |
501| `/gpu/magma`            | CUDA MAGMA kernels              |
502
503Examples available in this release:
504
505:::{list-table}
506:header-rows: 1
507:widths: auto
508* - User code
509  - Example
510* - `ceed`
511  - * ex1 (volume)
512* - `mfem`
513  - * BP1 (scalar mass operator)
514    * BP3 (scalar Laplace operator)
515* - `petsc`
516  - * BP1 (scalar mass operator)
517* - `nek5000`
518  - * BP1 (scalar mass operator)
519:::
520
521(v0-2)=
522
523## v0.2 (Mar 30, 2018)
524
525libCEED was made publicly available the first full CEED software distribution, release
526CEED 1.0. The distribution was made available using the Spack package manager to provide
527a common, easy-to-use build environment, where the user can build the CEED distribution
528with all dependencies. This release included a new Fortran interface for the library.
529This release also contained major improvements in the OCCA backend (including a new
530`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to
531compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an
532`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application
533of the Laplace operator) was also added to this release.
534
535Backends available in this release:
536
537| CEED resource (`-ceed`) | Backend                         |
538|-------------------------|---------------------------------|
539| `/cpu/self`             | Serial reference implementation |
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
545Examples available in this release:
546
547:::{list-table}
548:header-rows: 1
549:widths: auto
550* - User code
551  - Example
552* - `ceed`
553  - * ex1 (volume)
554* - `mfem`
555  - * BP1 (scalar mass operator)
556    * BP3 (scalar Laplace operator)
557* - `petsc`
558  - * BP1 (scalar mass operator)
559* - `nek5000`
560  - * BP1 (scalar mass operator)
561:::
562
563(v0-1)=
564
565## v0.1 (Jan 3, 2018)
566
567Initial low-level API of the CEED project. The low-level API provides a set of Finite
568Elements kernels and components for writing new low-level kernels. Examples include:
569vector and sparse linear algebra, element matrix assembly over a batch of elements,
570partial assembly and action for efficient high-order operators like mass, diffusion,
571advection, etc. The main goal of the low-level API is to establish the basis for the
572high-level API. Also, identifying such low-level kernels and providing a reference
573implementation for them serves as the basis for specialized backend implementations.
574This release contained several backends: `/cpu/self`, and backends which rely upon the
575[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`,
576`/gpu/occa`, and `/omp/occa`.
577It also included several examples, in the `examples` folder:
578A standalone code that shows the usage of libCEED (with no external
579dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1
580(with the application of the mass operator); and a `petsc` example to perform BP1
581(with the application of the mass operator).
582
583Backends available in this release:
584
585| CEED resource (`-ceed`) | Backend                         |
586|-------------------------|---------------------------------|
587| `/cpu/self`             | Serial reference implementation |
588| `/cpu/occa`             | Serial OCCA kernels             |
589| `/gpu/occa`             | CUDA OCCA kernels               |
590| `/omp/occa`             | OpenMP OCCA kernels             |
591
592Examples available in this release:
593
594| User code             | Example                           |
595|-----------------------|-----------------------------------|
596| `ceed`                | ex1 (scalar Laplace operator)     |
597| `mfem`                | BP1 (scalar mass operator)        |
598| `petsc`               | BP1 (scalar mass operator)        |
599```
600