xref: /libCEED/doc/sphinx/source/releasenotes.md (revision 5dfaedb85d2aa5da89951bb5d8f41d61be09bbf6)
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- Added `CeedInt_FMT` to support potential future use of larger interger sizes.
18- Added CEED_QFUNCTION_ATTR for setting compiler attributes/pragmas to CEED_QFUNCTION_HELPER and CEED_QFUNCTION
19
20### Bugfix
21
22- Fix bug in setting device id for GPU backends.
23- Fix storing of indices for `CeedElemRestriction` on the host with GPU backends.
24- Fix `CeedElemRestriction` sizing for {c:func}`CeedOperatorAssemblePointBlockDiagonal`.
25- Fix bugs in CPU implementation of {c:func}`CeedOperatorLinearAssemble` when there are different number of active input modes and active output modes.
26
27### Examples
28
29- Added various performance enhancements for {ref}`example-petsc-navier-stokes`.
30- Refactored {ref}`example-petsc-navier-stokes` to improve code reuse.
31- Added Shock Tube, Channel, and Flat Plate boundary layer problems to {ref}`example-petsc-navier-stokes`.
32- Added ability to use QFunctions for strong STG inflow in {ref}`example-petsc-navier-stokes`.
33
34(v0-10-1)=
35
36## v0.10.1 (Apr 11, 2022)
37
38### Interface changes
39
40- Added {c:func}`CeedQFunctionSetUserFlopsEstimate` and {c:func}`CeedOperatorGetFlopsEstimate` to facilitate estimating FLOPs in operator application.
41
42### New features
43
44- Switched MAGMA backends to use runtime compilation for tensor basis kernels (and element restriction kernels, in non-deterministic `/gpu/*/magma` backends).
45This reduces time to compile the library and increases the range of parameters for which the MAGMA tensor basis kernels will work.
46
47### Bugfix
48
49- Install JiT source files in install directory to fix GPU functionality for installed libCEED.
50
51(v0-10)=
52
53## v0.10 (Mar 21, 2022)
54
55### Interface changes
56
57- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields.
58- 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`.
59- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added.
60- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients.
61- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces.
62- Warning added when compiling OCCA backend to alert users that this backend is experimental.
63- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`.
64- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name.
65- 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`.
66- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitalized arrays; require initalized data for {c:func}`CeedVectorGetArray`.
67- 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.
68- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retreive user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`.
69- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums.
70- 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.
71- Added {c:func}`CeedOperatorSetQFunctionUpdated` to facilitate QFunction data re-use between operators sharing the same quadrature space, such as in a multigrid hierarchy.
72- Added {c:func}`CeedOperatorGetActiveVectorLengths` to get shape of CeedOperator.
73
74### New features
75
76- `CeedScalar` can now be set as `float` or `double` at compile time.
77- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends.
78- 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.
79- Remove need to guard library headers in QFunction source for code generation backends.
80- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present.
81- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally.
82- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file.
83- Added support for non-tensor H(div) elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor.
84- 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.
85- Added support for element matrix assembly in GPU backends.
86
87### Maintainability
88
89- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage.
90- `Include-what-you-use` makefile target added as `make iwyu`.
91- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase.
92- Put GPU JiTed kernel source code into separate files.
93- 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.
94
95(v0-9)=
96
97## v0.9 (Jul 6, 2021)
98
99### Interface changes
100
101- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact.
102
103### New features
104
105- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use.
106- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends.
107- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations.
108- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed).
109
110### Performance improvements
111
112### Examples
113
114- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations.
115- Fluid mechanics example adds GPU support and improves modularity.
116
117### Deprecated backends
118
119- 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.
120
121(v0-8)=
122
123## v0.8 (Mar 31, 2021)
124
125### Interface changes
126
127- Error handling improved to include enumerated error codes for C interface return values.
128- 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.
129
130### New features
131
132- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features.
133- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly.
134- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators.
135
136### Performance improvements
137
138- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`.
139- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`.
140
141### Examples
142
143- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions.
144- {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.
145- {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.
146- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree.
147
148(v0-7)=
149
150## v0.7 (Sep 29, 2020)
151
152### Interface changes
153
154- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors.
155  As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`.
156  These changes improve support for mixed finite element methods.
157- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature.
158- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity.
159- 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.
160- 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.
161- 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.
162- 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.
163  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`.
164- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory.
165- 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.
166
167### New features
168
169- New HIP backend: `/gpu/hip/ref`.
170- CeedQFunction support for user `CUfunction`s in some backends
171
172### Performance improvements
173
174- OCCA backend rebuilt to facilitate future performance enhancements.
175- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`.
176
177### Examples
178
179- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions.
180
181### Deprecated backends
182
183- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`.
184
185(v0-6)=
186
187## v0.6 (Mar 29, 2020)
188
189libCEED v0.6 contains numerous new features and examples, as well as expanded
190documentation in [this new website](https://libceed.org).
191
192### New features
193
194- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly
195  1-1 correspondence with the C interface, plus some convenience features.  For instance,
196  data stored in the {cpp:type}`CeedVector` structure are available without copy as
197  {py:class}`numpy.ndarray`.  Short tutorials are provided in
198  [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/).
199- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point,
200  {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal
201  ({c:func}`CeedOperatorAssembleLinearDiagonal`).  These operations are useful for
202  preconditioning ingredients and are used in the libCEED's multigrid examples.
203- The inverse of separable operators can be obtained using
204  {c:func}`CeedOperatorCreateFDMElementInverse` and applied with
205  {c:func}`CeedOperatorApply`.  This is a useful preconditioning ingredient,
206  especially for Laplacians and related operators.
207- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`,
208  {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`.
209- Make public accessors for various attributes to facilitate writing composable code.
210- New backend: `/cpu/self/memcheck/serial`.
211- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA
212  backends.  (Single source is coming soon for OCCA backends.)
213- Fix some missing edge cases in CUDA backend.
214
215### Performance Improvements
216
217- MAGMA backend performance optimization and non-tensor bases.
218- No-copy optimization in {c:func}`CeedOperatorApply`.
219
220### Interface changes
221
222- Replace {code}`CeedElemRestrictionCreateIdentity` and
223  {code}`CeedElemRestrictionCreateBlocked` with more flexible
224  {c:func}`CeedElemRestrictionCreateStrided` and
225  {c:func}`CeedElemRestrictionCreateBlockedStrided`.
226- Add arguments to {c:func}`CeedQFunctionCreateIdentity`.
227- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification
228  with {cpp:enum}`CeedInterlaceMode`.  This is now an attribute of the
229  {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no
230  longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and
231  {c:func}`CeedElemRestrictionApply`.
232
233### Examples
234
235libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`.
236Notable additions include:
237
238- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of
239  a domain in 1, 2, and 3 dimensions by applying a Laplacian.
240
241- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area
242  of domains (like the cube and sphere) by direct integration on a surface mesh;
243  demonstrates geometric dimension different from topological dimension.
244
245- PETSc {ref}`example-petsc-bps`:
246
247  - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support.
248  - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements
249    and transparent CUDA support.
250  - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`):
251    generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric
252    dimension different from topological dimension.
253
254- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid
255  solver with algebraic multigrid coarse solve.
256
257- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly
258  `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`),
259  implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and
260  quasi-2D computational domain support.
261
262- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for
263  linear elasticity, small-strain hyperelasticity, and globalized finite-strain
264  hyperelasticity using p-multigrid with algebraic multigrid coarse solve.
265
266(v0-5)=
267
268## v0.5 (Sep 18, 2019)
269
270For this release, several improvements were made. Two new CUDA backends were added to
271the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art
272performance using single-source {ref}`CeedQFunction`. From this release, users
273can define Q-Functions in a single source code independently of the targeted backend
274with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU
275compilation of the user provided {ref}`CeedQFunction` code. To allow a unified
276declaration, the {ref}`CeedQFunction` API has undergone a slight change:
277the `QFunctionField` parameter `ncomp` has been changed to `size`. This change
278requires setting the previous value of `ncomp` to `ncomp*dim` when adding a
279`QFunctionField` with eval mode `CEED EVAL GRAD`.
280
281Additionally, new CPU backends
282were included in this release, such as the `/cpu/self/opt/*` backends (which are
283written in pure C and use partial **E-vectors** to improve performance) and the
284`/cpu/self/ref/memcheck` backend (which relies upon the
285[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user
286{ref}`CeedQFunction` have no undefined values).
287This release also included various performance improvements, bug fixes, new examples,
288and improved tests. Among these improvements, vectorized instructions for
289{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD`
290instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and
291identity Q-Functions were introduced, and the PETSc benchmark problems were expanded
292to include unstructured meshes handling were. For this expansion, the prior version of
293the PETSc BPs, which only included data associated with structured geometries, were
294renamed `bpsraw`, and the new version of the BPs, which can handle data associated
295with any unstructured geometry, were called `bps`. Additionally, other benchmark
296problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively),
297and BP5 and BP6 (the collocated versions---for which the quadrature points are the same
298as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc
299examples. Furthermoew, another standalone libCEED example, called `ex2`, which
300computes the surface area of a given mesh was added to this release.
301
302Backends available in this release:
303
304| CEED resource (`-ceed`)  | Backend                                             |
305|--------------------------|-----------------------------------------------------|
306| `/cpu/self/ref/serial`   | Serial reference implementation                     |
307| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
308| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks            |
309| `/cpu/self/opt/serial`   | Serial optimized C implementation                   |
310| `/cpu/self/opt/blocked`  | Blocked optimized C implementation                  |
311| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
312| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
313| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
314| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
315| `/cpu/occa`              | Serial OCCA kernels                                 |
316| `/gpu/occa`              | CUDA OCCA kernels                                   |
317| `/omp/occa`              | OpenMP OCCA kernels                                 |
318| `/ocl/occa`              | OpenCL OCCA kernels                                 |
319| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
320| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
321| `/gpu/cuda/shared`       | Optimized pure CUDA kernels using shared memory     |
322| `/gpu/cuda/gen`          | Optimized pure CUDA kernels using code generation   |
323| `/gpu/magma`             | CUDA MAGMA kernels                                  |
324
325Examples available in this release:
326
327:::{list-table}
328:header-rows: 1
329:widths: auto
330* - User code
331  - Example
332* - `ceed`
333  - * ex1 (volume)
334    * ex2 (surface)
335* - `mfem`
336  - * BP1 (scalar mass operator)
337    * BP3 (scalar Laplace operator)
338* - `petsc`
339  - * BP1 (scalar mass operator)
340    * BP2 (vector mass operator)
341    * BP3 (scalar Laplace operator)
342    * BP4 (vector Laplace operator)
343    * BP5 (collocated scalar Laplace operator)
344    * BP6 (collocated vector Laplace operator)
345    * Navier-Stokes
346* - `nek5000`
347  - * BP1 (scalar mass operator)
348    * BP3 (scalar Laplace operator)
349:::
350
351(v0-4)=
352
353## v0.4 (Apr 1, 2019)
354
355libCEED v0.4 was made again publicly available in the second full CEED software
356distribution, release CEED 2.0. This release contained notable features, such as
357four new CPU backends, two new GPU backends, CPU backend optimizations, initial
358support for operator composition, performance benchmarking, and a Navier-Stokes demo.
359The new CPU backends in this release came in two families. The `/cpu/self/*/serial`
360backends process one element at a time and are intended for meshes with a smaller number
361of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of
362eight interlaced elements and are intended for meshes with higher numbers of elements.
363The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU
364performance. The `/cpu/self/xsmm/*` backends rely upon the
365[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU
366performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA.
367The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable
368performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple
369parallelization approach, where each thread treats a finite element. Using just in time
370compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this
371backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend
372achieve good peak performance for 1D, 2D, and low order 3D problems, but performance
373deteriorates very quickly when threads run out of registers.
374
375A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED
376examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`).
377This example solves the time-dependent Navier-Stokes equations of compressible gas
378dynamics in a static Eulerian three-dimensional frame, using structured high-order
379finite/spectral element spatial discretizations and explicit high-order time-stepping
380(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc,
381so that the pointwise physics (defined at quadrature points) is separated from the
382parallelization and meshing concerns.
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/tmpl`         | Backend template, defaults to `/cpu/self/blocked`   |
391| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
392| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
393| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
394| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
395| `/cpu/occa`              | Serial OCCA kernels                                 |
396| `/gpu/occa`              | CUDA OCCA kernels                                   |
397| `/omp/occa`              | OpenMP OCCA kernels                                 |
398| `/ocl/occa`              | OpenCL OCCA kernels                                 |
399| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
400| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
401| `/gpu/magma`             | CUDA MAGMA kernels                                  |
402
403Examples available in this release:
404
405:::{list-table}
406:header-rows: 1
407:widths: auto
408* - User code
409  - Example
410* - `ceed`
411  - * ex1 (volume)
412* - `mfem`
413  - * BP1 (scalar mass operator)
414    * BP3 (scalar Laplace operator)
415* - `petsc`
416  - * BP1 (scalar mass operator)
417    * BP3 (scalar Laplace operator)
418    * Navier-Stokes
419* - `nek5000`
420  - * BP1 (scalar mass operator)
421    * BP3 (scalar Laplace operator)
422:::
423
424(v0-3)=
425
426## v0.3 (Sep 30, 2018)
427
428Notable features in this release include active/passive field interface, support for
429non-tensor bases, backend optimization, and improved Fortran interface. This release
430also focused on providing improved continuous integration, and many new tests with code
431coverage reports of about 90%. This release also provided a significant change to the
432public interface: a {ref}`CeedQFunction` can take any number of named input and output
433arguments while {ref}`CeedOperator` connects them to the actual data, which may be
434supplied explicitly to `CeedOperatorApply()` (active) or separately via
435`CeedOperatorSetField()` (passive). This interface change enables reusable libraries
436of CeedQFunctions and composition of block solvers constructed using
437{ref}`CeedOperator`. A concept of blocked restriction was added to this release and
438used in an optimized CPU backend. Although this is typically not visible to the user,
439it enables effective use of arbitrary-length SIMD while maintaining cache locality.
440This CPU backend also implements an algebraic factorization of tensor product gradients
441to perform fewer operations than standard application of interpolation and
442differentiation from nodes to quadrature points. This algebraic formulation
443automatically supports non-polynomial and non-interpolatory bases, thus is more general
444than the more common derivation in terms of Lagrange polynomials on the quadrature points.
445
446Backends available in this release:
447
448| CEED resource (`-ceed`) | Backend                                             |
449|-------------------------|-----------------------------------------------------|
450| `/cpu/self/blocked`     | Blocked reference implementation                    |
451| `/cpu/self/ref`         | Serial reference implementation                     |
452| `/cpu/self/tmpl`        | Backend template, defaults to `/cpu/self/blocked`   |
453| `/cpu/occa`             | Serial OCCA kernels                                 |
454| `/gpu/occa`             | CUDA OCCA kernels                                   |
455| `/omp/occa`             | OpenMP OCCA kernels                                 |
456| `/ocl/occa`             | OpenCL OCCA kernels                                 |
457| `/gpu/magma`            | CUDA MAGMA kernels                                  |
458
459Examples available in this release:
460
461:::{list-table}
462:header-rows: 1
463:widths: auto
464* - User code
465  - Example
466* - `ceed`
467  - * ex1 (volume)
468* - `mfem`
469  - * BP1 (scalar mass operator)
470    * BP3 (scalar Laplace operator)
471* - `petsc`
472  - * BP1 (scalar mass operator)
473    * BP3 (scalar Laplace operator)
474* - `nek5000`
475  - * BP1 (scalar mass operator)
476    * BP3 (scalar Laplace operator)
477:::
478
479(v0-21)=
480
481## v0.21 (Sep 30, 2018)
482
483A MAGMA backend (which relies upon the
484[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this
485release. This initial integration set up the framework of using MAGMA and provided the
486libCEED functionality through MAGMA kernels as one of libCEED’s computational backends.
487As any other backend, the MAGMA backend provides extended basic data structures for
488{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements
489the fundamental CEED building blocks to work with the new data structures.
490In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data
491but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled
492internally, and thus seamlessly to the user, through the functions/methods that are
493provided to support them.
494
495Backends available in this release:
496
497| CEED resource (`-ceed`) | Backend                         |
498|-------------------------|---------------------------------|
499| `/cpu/self`             | Serial reference implementation |
500| `/cpu/occa`             | Serial OCCA kernels             |
501| `/gpu/occa`             | CUDA OCCA kernels               |
502| `/omp/occa`             | OpenMP OCCA kernels             |
503| `/ocl/occa`             | OpenCL OCCA kernels             |
504| `/gpu/magma`            | CUDA MAGMA kernels              |
505
506Examples available in this release:
507
508:::{list-table}
509:header-rows: 1
510:widths: auto
511* - User code
512  - Example
513* - `ceed`
514  - * ex1 (volume)
515* - `mfem`
516  - * BP1 (scalar mass operator)
517    * BP3 (scalar Laplace operator)
518* - `petsc`
519  - * BP1 (scalar mass operator)
520* - `nek5000`
521  - * BP1 (scalar mass operator)
522:::
523
524(v0-2)=
525
526## v0.2 (Mar 30, 2018)
527
528libCEED was made publicly available the first full CEED software distribution, release
529CEED 1.0. The distribution was made available using the Spack package manager to provide
530a common, easy-to-use build environment, where the user can build the CEED distribution
531with all dependencies. This release included a new Fortran interface for the library.
532This release also contained major improvements in the OCCA backend (including a new
533`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to
534compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an
535`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application
536of the Laplace operator) was also added to this release.
537
538Backends available in this release:
539
540| CEED resource (`-ceed`) | Backend                         |
541|-------------------------|---------------------------------|
542| `/cpu/self`             | Serial reference implementation |
543| `/cpu/occa`             | Serial OCCA kernels             |
544| `/gpu/occa`             | CUDA OCCA kernels               |
545| `/omp/occa`             | OpenMP OCCA kernels             |
546| `/ocl/occa`             | OpenCL OCCA kernels             |
547
548Examples available in this release:
549
550:::{list-table}
551:header-rows: 1
552:widths: auto
553* - User code
554  - Example
555* - `ceed`
556  - * ex1 (volume)
557* - `mfem`
558  - * BP1 (scalar mass operator)
559    * BP3 (scalar Laplace operator)
560* - `petsc`
561  - * BP1 (scalar mass operator)
562* - `nek5000`
563  - * BP1 (scalar mass operator)
564:::
565
566(v0-1)=
567
568## v0.1 (Jan 3, 2018)
569
570Initial low-level API of the CEED project. The low-level API provides a set of Finite
571Elements kernels and components for writing new low-level kernels. Examples include:
572vector and sparse linear algebra, element matrix assembly over a batch of elements,
573partial assembly and action for efficient high-order operators like mass, diffusion,
574advection, etc. The main goal of the low-level API is to establish the basis for the
575high-level API. Also, identifying such low-level kernels and providing a reference
576implementation for them serves as the basis for specialized backend implementations.
577This release contained several backends: `/cpu/self`, and backends which rely upon the
578[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`,
579`/gpu/occa`, and `/omp/occa`.
580It also included several examples, in the `examples` folder:
581A standalone code that shows the usage of libCEED (with no external
582dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1
583(with the application of the mass operator); and a `petsc` example to perform BP1
584(with the application of the mass operator).
585
586Backends available in this release:
587
588| CEED resource (`-ceed`) | Backend                         |
589|-------------------------|---------------------------------|
590| `/cpu/self`             | Serial reference implementation |
591| `/cpu/occa`             | Serial OCCA kernels             |
592| `/gpu/occa`             | CUDA OCCA kernels               |
593| `/omp/occa`             | OpenMP OCCA kernels             |
594
595Examples available in this release:
596
597| User code             | Example                           |
598|-----------------------|-----------------------------------|
599| `ceed`                | ex1 (scalar Laplace operator)     |
600| `mfem`                | BP1 (scalar mass operator)        |
601| `petsc`               | BP1 (scalar mass operator)        |
602```
603