xref: /libCEED/doc/sphinx/source/releasenotes.md (revision 42971d38b94f0ed7610b44682a014e3204aa7efd)
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
13### Bugfix
14
15- Fix storing of indices for `CeedElemRestriction` on the host with GPU backends.
16
17### Examples
18
19- Added various performance enhancements for {ref}`example-petsc-navier-stokes`
20
21(v0-10-1)=
22
23## v0.10.1 (Apr 11, 2022)
24
25### Interface changes
26
27- Added {c:func}`CeedQFunctionSetUserFlopsEstimate` and {c:func}`CeedOperatorGetFlopsEstimate` to facilitate estimating FLOPs in operator application.
28
29### Bugfix
30
31- Install JiT source files in install directory to fix GPU functionality for installed libCEED.
32
33(v0-10)=
34
35## v0.10 (Mar 21, 2022)
36
37### Interface changes
38
39- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields.
40- 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`.
41- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added.
42- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients.
43- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces.
44- Warning added when compiling OCCA backend to alert users that this backend is experimental.
45- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`.
46- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name.
47- 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`.
48- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitalized arrays; require initalized data for {c:func}`CeedVectorGetArray`.
49- 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.
50- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retreive user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`.
51- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums.
52- 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.
53- Added {c:func}`CeedOperatorSetQFunctionUpdated` to facilitate QFunction data re-use between operators sharing the same quadrature space, such as in a multigrid hierarchy.
54- Added {c:func}`CeedOperatorGetActiveVectorLengths` to get shape of CeedOperator.
55
56### New features
57
58- `CeedScalar` can now be set as `float` or `double` at compile time.
59- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends.
60- 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.
61- Remove need to guard library headers in QFunction source for code generation backends.
62- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present.
63- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally.
64- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file.
65- Added support for non-tensor H(div) elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor.
66- 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.
67- Added support for element matrix assembly in GPU backends.
68
69### Maintainability
70
71- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage.
72- `Include-what-you-use` makefile target added as `make iwyu`.
73- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase.
74- Put GPU JiTed kernel source code into separate files.
75- 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.
76
77(v0-9)=
78
79## v0.9 (Jul 6, 2021)
80
81### Interface changes
82
83- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact.
84
85### New features
86
87- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use.
88- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends.
89- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations.
90- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed).
91
92### Performance improvements
93
94### Examples
95
96- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations.
97- Fluid mechanics example adds GPU support and improves modularity.
98
99### Deprecated backends
100
101- 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.
102
103(v0-8)=
104
105## v0.8 (Mar 31, 2021)
106
107### Interface changes
108
109- Error handling improved to include enumerated error codes for C interface return values.
110- 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.
111
112### New features
113
114- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features.
115- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly.
116- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators.
117
118### Performance improvements
119
120- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`.
121- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`.
122
123### Examples
124
125- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions.
126- {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.
127- {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.
128- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree.
129
130(v0-7)=
131
132## v0.7 (Sep 29, 2020)
133
134### Interface changes
135
136- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors.
137  As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`.
138  These changes improve support for mixed finite element methods.
139- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature.
140- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity.
141- 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.
142- 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.
143- 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.
144- 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.
145  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`.
146- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory.
147- 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.
148
149### New features
150
151- New HIP backend: `/gpu/hip/ref`.
152- CeedQFunction support for user `CUfunction`s in some backends
153
154### Performance improvements
155
156- OCCA backend rebuilt to facilitate future performance enhancements.
157- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`.
158
159### Examples
160
161- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions.
162
163### Deprecated backends
164
165- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`.
166
167(v0-6)=
168
169## v0.6 (Mar 29, 2020)
170
171libCEED v0.6 contains numerous new features and examples, as well as expanded
172documentation in [this new website](https://libceed.org).
173
174### New features
175
176- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly
177  1-1 correspondence with the C interface, plus some convenience features.  For instance,
178  data stored in the {cpp:type}`CeedVector` structure are available without copy as
179  {py:class}`numpy.ndarray`.  Short tutorials are provided in
180  [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/).
181- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point,
182  {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal
183  ({c:func}`CeedOperatorAssembleLinearDiagonal`).  These operations are useful for
184  preconditioning ingredients and are used in the libCEED's multigrid examples.
185- The inverse of separable operators can be obtained using
186  {c:func}`CeedOperatorCreateFDMElementInverse` and applied with
187  {c:func}`CeedOperatorApply`.  This is a useful preconditioning ingredient,
188  especially for Laplacians and related operators.
189- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`,
190  {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`.
191- Make public accessors for various attributes to facilitate writing composable code.
192- New backend: `/cpu/self/memcheck/serial`.
193- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA
194  backends.  (Single source is coming soon for OCCA backends.)
195- Fix some missing edge cases in CUDA backend.
196
197### Performance Improvements
198
199- MAGMA backend performance optimization and non-tensor bases.
200- No-copy optimization in {c:func}`CeedOperatorApply`.
201
202### Interface changes
203
204- Replace {code}`CeedElemRestrictionCreateIdentity` and
205  {code}`CeedElemRestrictionCreateBlocked` with more flexible
206  {c:func}`CeedElemRestrictionCreateStrided` and
207  {c:func}`CeedElemRestrictionCreateBlockedStrided`.
208- Add arguments to {c:func}`CeedQFunctionCreateIdentity`.
209- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification
210  with {cpp:enum}`CeedInterlaceMode`.  This is now an attribute of the
211  {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no
212  longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and
213  {c:func}`CeedElemRestrictionApply`.
214
215### Examples
216
217libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`.
218Notable additions include:
219
220- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of
221  a domain in 1, 2, and 3 dimensions by applying a Laplacian.
222
223- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area
224  of domains (like the cube and sphere) by direct integration on a surface mesh;
225  demonstrates geometric dimension different from topological dimension.
226
227- PETSc {ref}`example-petsc-bps`:
228
229  - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support.
230  - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements
231    and transparent CUDA support.
232  - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`):
233    generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric
234    dimension different from topological dimension.
235
236- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid
237  solver with algebraic multigrid coarse solve.
238
239- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly
240  `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`),
241  implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and
242  quasi-2D computational domain support.
243
244- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for
245  linear elasticity, small-strain hyperelasticity, and globalized finite-strain
246  hyperelasticity using p-multigrid with algebraic multigrid coarse solve.
247
248(v0-5)=
249
250## v0.5 (Sep 18, 2019)
251
252For this release, several improvements were made. Two new CUDA backends were added to
253the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art
254performance using single-source {ref}`CeedQFunction`. From this release, users
255can define Q-Functions in a single source code independently of the targeted backend
256with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU
257compilation of the user provided {ref}`CeedQFunction` code. To allow a unified
258declaration, the {ref}`CeedQFunction` API has undergone a slight change:
259the `QFunctionField` parameter `ncomp` has been changed to `size`. This change
260requires setting the previous value of `ncomp` to `ncomp*dim` when adding a
261`QFunctionField` with eval mode `CEED EVAL GRAD`.
262
263Additionally, new CPU backends
264were included in this release, such as the `/cpu/self/opt/*` backends (which are
265written in pure C and use partial **E-vectors** to improve performance) and the
266`/cpu/self/ref/memcheck` backend (which relies upon the
267[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user
268{ref}`CeedQFunction` have no undefined values).
269This release also included various performance improvements, bug fixes, new examples,
270and improved tests. Among these improvements, vectorized instructions for
271{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD`
272instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and
273identity Q-Functions were introduced, and the PETSc benchmark problems were expanded
274to include unstructured meshes handling were. For this expansion, the prior version of
275the PETSc BPs, which only included data associated with structured geometries, were
276renamed `bpsraw`, and the new version of the BPs, which can handle data associated
277with any unstructured geometry, were called `bps`. Additionally, other benchmark
278problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively),
279and BP5 and BP6 (the collocated versions---for which the quadrature points are the same
280as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc
281examples. Furthermoew, another standalone libCEED example, called `ex2`, which
282computes the surface area of a given mesh was added to this release.
283
284Backends available in this release:
285
286| CEED resource (`-ceed`)  | Backend                                             |
287|--------------------------|-----------------------------------------------------|
288| `/cpu/self/ref/serial`   | Serial reference implementation                     |
289| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
290| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks            |
291| `/cpu/self/opt/serial`   | Serial optimized C implementation                   |
292| `/cpu/self/opt/blocked`  | Blocked optimized C implementation                  |
293| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
294| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
295| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
296| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
297| `/cpu/occa`              | Serial OCCA kernels                                 |
298| `/gpu/occa`              | CUDA OCCA kernels                                   |
299| `/omp/occa`              | OpenMP OCCA kernels                                 |
300| `/ocl/occa`              | OpenCL OCCA kernels                                 |
301| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
302| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
303| `/gpu/cuda/shared`       | Optimized pure CUDA kernels using shared memory     |
304| `/gpu/cuda/gen`          | Optimized pure CUDA kernels using code generation   |
305| `/gpu/magma`             | CUDA MAGMA kernels                                  |
306
307Examples available in this release:
308
309:::{list-table}
310:header-rows: 1
311:widths: auto
312* - User code
313  - Example
314* - `ceed`
315  - * ex1 (volume)
316    * ex2 (surface)
317* - `mfem`
318  - * BP1 (scalar mass operator)
319    * BP3 (scalar Laplace operator)
320* - `petsc`
321  - * BP1 (scalar mass operator)
322    * BP2 (vector mass operator)
323    * BP3 (scalar Laplace operator)
324    * BP4 (vector Laplace operator)
325    * BP5 (collocated scalar Laplace operator)
326    * BP6 (collocated vector Laplace operator)
327    * Navier-Stokes
328* - `nek5000`
329  - * BP1 (scalar mass operator)
330    * BP3 (scalar Laplace operator)
331:::
332
333(v0-4)=
334
335## v0.4 (Apr 1, 2019)
336
337libCEED v0.4 was made again publicly available in the second full CEED software
338distribution, release CEED 2.0. This release contained notable features, such as
339four new CPU backends, two new GPU backends, CPU backend optimizations, initial
340support for operator composition, performance benchmarking, and a Navier-Stokes demo.
341The new CPU backends in this release came in two families. The `/cpu/self/*/serial`
342backends process one element at a time and are intended for meshes with a smaller number
343of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of
344eight interlaced elements and are intended for meshes with higher numbers of elements.
345The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU
346performance. The `/cpu/self/xsmm/*` backends rely upon the
347[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU
348performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA.
349The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable
350performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple
351parallelization approach, where each thread treats a finite element. Using just in time
352compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this
353backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend
354achieve good peak performance for 1D, 2D, and low order 3D problems, but performance
355deteriorates very quickly when threads run out of registers.
356
357A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED
358examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`).
359This example solves the time-dependent Navier-Stokes equations of compressible gas
360dynamics in a static Eulerian three-dimensional frame, using structured high-order
361finite/spectral element spatial discretizations and explicit high-order time-stepping
362(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc,
363so that the pointwise physics (defined at quadrature points) is separated from the
364parallelization and meshing concerns.
365
366Backends available in this release:
367
368| CEED resource (`-ceed`)  | Backend                                             |
369|--------------------------|-----------------------------------------------------|
370| `/cpu/self/ref/serial`   | Serial reference implementation                     |
371| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
372| `/cpu/self/tmpl`         | Backend template, defaults to `/cpu/self/blocked`   |
373| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
374| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
375| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
376| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
377| `/cpu/occa`              | Serial OCCA kernels                                 |
378| `/gpu/occa`              | CUDA OCCA kernels                                   |
379| `/omp/occa`              | OpenMP OCCA kernels                                 |
380| `/ocl/occa`              | OpenCL OCCA kernels                                 |
381| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
382| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
383| `/gpu/magma`             | CUDA MAGMA kernels                                  |
384
385Examples available in this release:
386
387:::{list-table}
388:header-rows: 1
389:widths: auto
390* - User code
391  - Example
392* - `ceed`
393  - * ex1 (volume)
394* - `mfem`
395  - * BP1 (scalar mass operator)
396    * BP3 (scalar Laplace operator)
397* - `petsc`
398  - * BP1 (scalar mass operator)
399    * BP3 (scalar Laplace operator)
400    * Navier-Stokes
401* - `nek5000`
402  - * BP1 (scalar mass operator)
403    * BP3 (scalar Laplace operator)
404:::
405
406(v0-3)=
407
408## v0.3 (Sep 30, 2018)
409
410Notable features in this release include active/passive field interface, support for
411non-tensor bases, backend optimization, and improved Fortran interface. This release
412also focused on providing improved continuous integration, and many new tests with code
413coverage reports of about 90%. This release also provided a significant change to the
414public interface: a {ref}`CeedQFunction` can take any number of named input and output
415arguments while {ref}`CeedOperator` connects them to the actual data, which may be
416supplied explicitly to `CeedOperatorApply()` (active) or separately via
417`CeedOperatorSetField()` (passive). This interface change enables reusable libraries
418of CeedQFunctions and composition of block solvers constructed using
419{ref}`CeedOperator`. A concept of blocked restriction was added to this release and
420used in an optimized CPU backend. Although this is typically not visible to the user,
421it enables effective use of arbitrary-length SIMD while maintaining cache locality.
422This CPU backend also implements an algebraic factorization of tensor product gradients
423to perform fewer operations than standard application of interpolation and
424differentiation from nodes to quadrature points. This algebraic formulation
425automatically supports non-polynomial and non-interpolatory bases, thus is more general
426than the more common derivation in terms of Lagrange polynomials on the quadrature points.
427
428Backends available in this release:
429
430| CEED resource (`-ceed`) | Backend                                             |
431|-------------------------|-----------------------------------------------------|
432| `/cpu/self/blocked`     | Blocked reference implementation                    |
433| `/cpu/self/ref`         | Serial reference implementation                     |
434| `/cpu/self/tmpl`        | Backend template, defaults to `/cpu/self/blocked`   |
435| `/cpu/occa`             | Serial OCCA kernels                                 |
436| `/gpu/occa`             | CUDA OCCA kernels                                   |
437| `/omp/occa`             | OpenMP OCCA kernels                                 |
438| `/ocl/occa`             | OpenCL OCCA kernels                                 |
439| `/gpu/magma`            | CUDA MAGMA kernels                                  |
440
441Examples available in this release:
442
443:::{list-table}
444:header-rows: 1
445:widths: auto
446* - User code
447  - Example
448* - `ceed`
449  - * ex1 (volume)
450* - `mfem`
451  - * BP1 (scalar mass operator)
452    * BP3 (scalar Laplace operator)
453* - `petsc`
454  - * BP1 (scalar mass operator)
455    * BP3 (scalar Laplace operator)
456* - `nek5000`
457  - * BP1 (scalar mass operator)
458    * BP3 (scalar Laplace operator)
459:::
460
461(v0-21)=
462
463## v0.21 (Sep 30, 2018)
464
465A MAGMA backend (which relies upon the
466[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this
467release. This initial integration set up the framework of using MAGMA and provided the
468libCEED functionality through MAGMA kernels as one of libCEED’s computational backends.
469As any other backend, the MAGMA backend provides extended basic data structures for
470{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements
471the fundamental CEED building blocks to work with the new data structures.
472In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data
473but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled
474internally, and thus seamlessly to the user, through the functions/methods that are
475provided to support them.
476
477Backends available in this release:
478
479| CEED resource (`-ceed`) | Backend                         |
480|-------------------------|---------------------------------|
481| `/cpu/self`             | Serial reference implementation |
482| `/cpu/occa`             | Serial OCCA kernels             |
483| `/gpu/occa`             | CUDA OCCA kernels               |
484| `/omp/occa`             | OpenMP OCCA kernels             |
485| `/ocl/occa`             | OpenCL OCCA kernels             |
486| `/gpu/magma`            | CUDA MAGMA kernels              |
487
488Examples available in this release:
489
490:::{list-table}
491:header-rows: 1
492:widths: auto
493* - User code
494  - Example
495* - `ceed`
496  - * ex1 (volume)
497* - `mfem`
498  - * BP1 (scalar mass operator)
499    * BP3 (scalar Laplace operator)
500* - `petsc`
501  - * BP1 (scalar mass operator)
502* - `nek5000`
503  - * BP1 (scalar mass operator)
504:::
505
506(v0-2)=
507
508## v0.2 (Mar 30, 2018)
509
510libCEED was made publicly available the first full CEED software distribution, release
511CEED 1.0. The distribution was made available using the Spack package manager to provide
512a common, easy-to-use build environment, where the user can build the CEED distribution
513with all dependencies. This release included a new Fortran interface for the library.
514This release also contained major improvements in the OCCA backend (including a new
515`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to
516compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an
517`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application
518of the Laplace operator) was also added to this release.
519
520Backends available in this release:
521
522| CEED resource (`-ceed`) | Backend                         |
523|-------------------------|---------------------------------|
524| `/cpu/self`             | Serial reference implementation |
525| `/cpu/occa`             | Serial OCCA kernels             |
526| `/gpu/occa`             | CUDA OCCA kernels               |
527| `/omp/occa`             | OpenMP OCCA kernels             |
528| `/ocl/occa`             | OpenCL OCCA kernels             |
529
530Examples available in this release:
531
532:::{list-table}
533:header-rows: 1
534:widths: auto
535* - User code
536  - Example
537* - `ceed`
538  - * ex1 (volume)
539* - `mfem`
540  - * BP1 (scalar mass operator)
541    * BP3 (scalar Laplace operator)
542* - `petsc`
543  - * BP1 (scalar mass operator)
544* - `nek5000`
545  - * BP1 (scalar mass operator)
546:::
547
548(v0-1)=
549
550## v0.1 (Jan 3, 2018)
551
552Initial low-level API of the CEED project. The low-level API provides a set of Finite
553Elements kernels and components for writing new low-level kernels. Examples include:
554vector and sparse linear algebra, element matrix assembly over a batch of elements,
555partial assembly and action for efficient high-order operators like mass, diffusion,
556advection, etc. The main goal of the low-level API is to establish the basis for the
557high-level API. Also, identifying such low-level kernels and providing a reference
558implementation for them serves as the basis for specialized backend implementations.
559This release contained several backends: `/cpu/self`, and backends which rely upon the
560[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`,
561`/gpu/occa`, and `/omp/occa`.
562It also included several examples, in the `examples` folder:
563A standalone code that shows the usage of libCEED (with no external
564dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1
565(with the application of the mass operator); and a `petsc` example to perform BP1
566(with the application of the mass operator).
567
568Backends available in this release:
569
570| CEED resource (`-ceed`) | Backend                         |
571|-------------------------|---------------------------------|
572| `/cpu/self`             | Serial reference implementation |
573| `/cpu/occa`             | Serial OCCA kernels             |
574| `/gpu/occa`             | CUDA OCCA kernels               |
575| `/omp/occa`             | OpenMP OCCA kernels             |
576
577Examples available in this release:
578
579| User code             | Example                           |
580|-----------------------|-----------------------------------|
581| `ceed`                | ex1 (scalar Laplace operator)     |
582| `mfem`                | BP1 (scalar mass operator)        |
583| `petsc`               | BP1 (scalar mass operator)        |
584```
585