xref: /libCEED/doc/sphinx/source/releasenotes.md (revision b13efd58b277efef1db70d6f06eaaf4d415a7642)
1# Changes/Release Notes
2
3On this page we provide a summary of the main API changes, new features and examples for each release of libCEED.
4
5(main)=
6
7## Current `main` branch
8
9### Interface changes
10
11- Add `bool` field type for `CeedQFunctionContext` and related interfaces to use `bool` fields.
12- `CEED_BASIS_COLLOCATED` removed; users should only use `CEED_BASIS_NONE`.
13- Remove unneeded pointer for `CeedElemRestrictionGetELayout`.
14- Require use of `Ceed*Destroy()` on Ceed objects returned from `CeedOperatorFieldGet*()`;
15
16### New features
17
18- Add `CeedOperatorCreateAtPoints` which evaluates the `CeedQFunction` at arbitrary locations in each element, for use in Particle in Cell, Material Point Method, and similar methods.
19- Add `CeedElemRestrictionGetLLayout` to provide L-vector layout for strided `CeedElemRestriction` created with `CEED_BACKEND_STRIDES`.
20- Add `CeedVectorReturnCeed` and similar when parent `Ceed` context for a libCEED object is only needed once in a calling scope.
21- Enable `#pragma once` for all JiT source; remove duplicate includes in JiT source string before compilation.
22
23### Examples
24
25- Add deal.II example with CEED BP suite.
26
27(v0-12)=
28
29## v0.12 (Oct 31, 2023)
30
31### Interface changes
32
33- Update `CeedOperatorContext*` functions to `CeedOperator*Context*` functions for consistency.
34For example, `CeedOperatorContextGetFieldLabel` was renamed to `CeedOperatorGetContextFieldLabel`.
35- Removed `CeedBasisSetNumQuadraturePoints` as redundant and bug-prone interface.
36
37### New features
38
39- Added {c:func}`CeedOperatorGetFieldByName` to access a specific `CeedOperatorField` by its name.
40- Update `/cpu/self/memcheck/*` backends to help verify `CeedVector` array access assumptions and `CeedQFunction` user output assumptions.
41- Update {c:func}`CeedOperatorLinearAssembleDiagonal` to provide default implementation that supports `CeedOperator` with multiple active bases.
42- Added Sycl backends `/gpu/sycl/ref`, `/gpu/sycl/shared`, and `/gpu/sycl/gen`.
43- Added {c:func}`CeedBasisApplyAtPoints` for evaluation of values and derivatives at arbitrary points inside elements.
44- Added support for non-tensor $H(\text{curl})$ finite element spaces with {c:func}`CeedBasisCreateHcurl`.
45- Added {c:func}`CeedElemRestrictionCreateCurlOriented`, similar to {c:func}`CeedElemRestrictionCreateOriented`, for element restrictions requiring more general element transformations such as those for high-order $H(\text{curl})$ spaces on tetrahedra (see [https://dl.acm.org/doi/pdf/10.1145/3524456](https://dl.acm.org/doi/pdf/10.1145/3524456)).
46- Added {c:func}`CeedOperatorLinearAssemblePointBlockDiagonalSymbolic` to create COO mapping for mapping out of {c:func}`CeedOperatorLinearAssemblePointBlockDiagonal`.
47- Added support for application codes which manage multiple {ref}`Ceed` objects, parallelized across OpenMP threads.
48
49### Examples
50
51- Add `DMSwarm` example demonstrating interpolation from background mesh to swarm points and projection from swarm points to background mesh.
52
53#### {ref}`example-petsc-bps`
54
55- Requires PETSc version 3.19 or later.
56
57#### {ref}`example-petsc-navier-stokes`
58
59- Updated restart and checkpointing interface.
60- Add data-driven subgrid-stress model.
61- Add differential filtering of solution.
62- Add turbulence statistics collection over spanwise-symmetric geometries.
63- Add Taylor-Green vortex initial condition.
64- Add Riemann-based outflow boundary conditions.
65- Added vortex shedding and flow past cylinder example, including calculations for lift, drag, and heat transfer.
66- Add Internal Damping Layer (IDL) for helping turbulent simulation stability.
67- Derive `CeedBasis` from `PetscFE`, and various other internal maintainability updates.
68
69(v0-11)=
70
71## v0.11 (Dec 24, 2022)
72
73### Interface changes
74
75- Added {c:func}`CeedOperatorSetName` for more readable {c:func}`CeedOperatorView` output.
76- Added {c:func}`CeedBasisCreateProjection` to facilitate interpolation between nodes for separate `CeedBases`.
77- Rename and move {c:func}`CeedCompositeOperatorGetNumSub` and {c:func}`CeedCompositeOperatorGetSubList` to public interface.
78- Renamed `CEED_BASIS_COLLOCATED` to `CEED_BASIS_NONE` for clarity.
79Some users previously misinterpreted a `CeedOperator` field using `CEED_BASIS_COLLOCATED` as meaning that the entire `CeedOperator` used a quadrature space that is collocated with the nodal space of the active bases.
80
81### New features
82
83- Update `/cpu/self/memcheck/*` backends to help verify `CeedQFunctionContext` data sizes provided by user.
84- Improved support for $H(\text{div})$ bases.
85- Added `CeedInt_FMT` to support potential future use of larger integer sizes.
86- Added `CEED_QFUNCTION_ATTR` for setting compiler attributes/pragmas to `CEED_QFUNCTION_HELPER` and `CEED_QFUNCTION`.
87- OCCA backend updated to latest OCCA release; DPC++ and OMP OCCA modes enabled.
88Due to a limitation of the OCCA parser, typedefs are required to use pointers to arrays in QFunctions with the OCCA backend.
89This issue will be fixed in a future OCCA release.
90
91### Bugfix
92
93- Fix bug in setting device id for GPU backends.
94- Fix storing of indices for `CeedElemRestriction` on the host with GPU backends.
95- Fix `CeedElemRestriction` sizing for {c:func}`CeedOperatorAssemblePointBlockDiagonal`.
96- Fix bugs in CPU implementation of {c:func}`CeedOperatorLinearAssemble` when there are different number of active input modes and active output modes.
97
98### Examples
99
100#### {ref}`example-petsc-navier-stokes`
101
102- Various performance enhancements, analytic matrix-free and assembled Jacobian, and PETSc solver configurations for GPUs.
103- Refactored to improve code reuse and modularity.
104- Support for primitive variables for more accurate boundary layers and all-speed flow.
105- Added $YZ\beta$ shock capturing scheme and Shock Tube example.
106- Added Channel example, with comparison to analytic solutions.
107- Added Flat Plate with boundary layer mesh and compressible Blasius inflow condition based on Chebyshev collocation solution of the Blasius equations.
108- Added strong and weak synthetic turbulence generation (STG) inflow boundary conditions.
109- Added "freestream" boundary conditions based on HLLC Riemann solver.
110- Automated stabilization coefficients for different basis degree.
111
112#### {ref}`example-petsc-bps`
113
114- Support for convergence studies.
115
116### Maintainability
117
118- Refactored `/gpu/cuda/shared` and `/gpu/cuda/gen` as well as `/gpu/hip/shared` and `/gpu/hip/gen` backend to improve maintainablity and reduce duplicated code.
119- Enabled support for `p > 8` for `/gpu/*/shared` backends.
120- Switch to `clang-format` over `astyle` for automatic formatting; Makefile command changed to `make format` from `make style`.
121- Improved test harness.
122
123(v0-10-1)=
124
125## v0.10.1 (Apr 11, 2022)
126
127### Interface changes
128
129- Added {c:func}`CeedQFunctionSetUserFlopsEstimate` and {c:func}`CeedOperatorGetFlopsEstimate` to facilitate estimating FLOPs in operator application.
130
131### New features
132
133- Switched MAGMA backends to use runtime compilation for tensor basis kernels (and element restriction kernels, in non-deterministic `/gpu/*/magma` backends).
134This reduces time to compile the library and increases the range of parameters for which the MAGMA tensor basis kernels will work.
135
136### Bugfix
137
138- Install JiT source files in install directory to fix GPU functionality for installed libCEED.
139
140(v0-10)=
141
142## v0.10 (Mar 21, 2022)
143
144### Interface changes
145
146- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields.
147- 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`.
148- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added.
149- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients.
150- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces.
151- Warning added when compiling OCCA backend to alert users that this backend is experimental.
152- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`.
153- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name.
154- 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`.
155- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitialized arrays; require initialized data for {c:func}`CeedVectorGetArray`.
156- 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.
157- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retrieve user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`.
158- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums.
159- 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.
160- Added {c:func}`CeedOperatorSetQFunctionUpdated` to facilitate QFunction data re-use between operators sharing the same quadrature space, such as in a multigrid hierarchy.
161- Added {c:func}`CeedOperatorGetActiveVectorLengths` to get shape of CeedOperator.
162
163### New features
164
165- `CeedScalar` can now be set as `float` or `double` at compile time.
166- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends.
167- 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.
168- Remove need to guard library headers in QFunction source for code generation backends.
169- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present.
170- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally.
171- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file.
172- Added support for non-tensor $H(\text{div})$ elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor.
173- 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.
174- Added support for element matrix assembly in GPU backends.
175
176### Maintainability
177
178- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage.
179- `Include-what-you-use` makefile target added as `make iwyu`.
180- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase.
181- Put GPU JiTed kernel source code into separate files.
182- 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.
183
184(v0-9)=
185
186## v0.9 (Jul 6, 2021)
187
188### Interface changes
189
190- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact.
191
192### New features
193
194- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use.
195- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends.
196- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations.
197- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed).
198
199### Performance improvements
200
201### Examples
202
203- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations.
204- Fluid mechanics example adds GPU support and improves modularity.
205
206### Deprecated backends
207
208- 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.
209
210(v0-8)=
211
212## v0.8 (Mar 31, 2021)
213
214### Interface changes
215
216- Error handling improved to include enumerated error codes for C interface return values.
217- 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.
218
219### New features
220
221- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features.
222- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly.
223- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators.
224
225### Performance improvements
226
227- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`.
228- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`.
229
230### Examples
231
232- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions.
233- {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.
234- {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.
235- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree.
236
237(v0-7)=
238
239## v0.7 (Sep 29, 2020)
240
241### Interface changes
242
243- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors.
244  As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`.
245  These changes improve support for mixed finite element methods.
246- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature.
247- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity.
248- 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.
249- 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.
250- 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.
251- 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.
252  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`.
253- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory.
254- 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.
255
256### New features
257
258- New HIP backend: `/gpu/hip/ref`.
259- CeedQFunction support for user `CUfunction`s in some backends
260
261### Performance improvements
262
263- OCCA backend rebuilt to facilitate future performance enhancements.
264- PETSc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`.
265
266### Examples
267
268- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions.
269
270### Deprecated backends
271
272- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`.
273
274(v0-6)=
275
276## v0.6 (Mar 29, 2020)
277
278libCEED v0.6 contains numerous new features and examples, as well as expanded
279documentation in [this new website](https://libceed.org).
280
281### New features
282
283- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly
284  1-1 correspondence with the C interface, plus some convenience features.  For instance,
285  data stored in the {cpp:type}`CeedVector` structure are available without copy as
286  {py:class}`numpy.ndarray`.  Short tutorials are provided in
287  [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/).
288- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point,
289  {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal
290  ({c:func}`CeedOperatorAssembleLinearDiagonal`).  These operations are useful for
291  preconditioning ingredients and are used in the libCEED's multigrid examples.
292- The inverse of separable operators can be obtained using
293  {c:func}`CeedOperatorCreateFDMElementInverse` and applied with
294  {c:func}`CeedOperatorApply`.  This is a useful preconditioning ingredient,
295  especially for Laplacians and related operators.
296- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`,
297  {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`.
298- Make public accessors for various attributes to facilitate writing composable code.
299- New backend: `/cpu/self/memcheck/serial`.
300- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA
301  backends.  (Single source is coming soon for OCCA backends.)
302- Fix some missing edge cases in CUDA backend.
303
304### Performance Improvements
305
306- MAGMA backend performance optimization and non-tensor bases.
307- No-copy optimization in {c:func}`CeedOperatorApply`.
308
309### Interface changes
310
311- Replace {code}`CeedElemRestrictionCreateIdentity` and
312  {code}`CeedElemRestrictionCreateBlocked` with more flexible
313  {c:func}`CeedElemRestrictionCreateStrided` and
314  {c:func}`CeedElemRestrictionCreateBlockedStrided`.
315- Add arguments to {c:func}`CeedQFunctionCreateIdentity`.
316- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification
317  with {cpp:enum}`CeedInterlaceMode`.  This is now an attribute of the
318  {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no
319  longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and
320  {c:func}`CeedElemRestrictionApply`.
321
322### Examples
323
324libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`.
325Notable additions include:
326
327- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of
328  a domain in 1, 2, and 3 dimensions by applying a Laplacian.
329
330- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area
331  of domains (like the cube and sphere) by direct integration on a surface mesh;
332  demonstrates geometric dimension different from topological dimension.
333
334- PETSc {ref}`example-petsc-bps`:
335
336  - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support.
337  - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements
338    and transparent CUDA support.
339  - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`):
340    generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric
341    dimension different from topological dimension.
342
343- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid
344  solver with algebraic multigrid coarse solve.
345
346- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly
347  `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`),
348  implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and
349  quasi-2D computational domain support.
350
351- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for
352  linear elasticity, small-strain hyperelasticity, and globalized finite-strain
353  hyperelasticity using p-multigrid with algebraic multigrid coarse solve.
354
355(v0-5)=
356
357## v0.5 (Sep 18, 2019)
358
359For this release, several improvements were made. Two new CUDA backends were added to
360the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art
361performance using single-source {ref}`CeedQFunction`. From this release, users
362can define Q-Functions in a single source code independently of the targeted backend
363with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU
364compilation of the user provided {ref}`CeedQFunction` code. To allow a unified
365declaration, the {ref}`CeedQFunction` API has undergone a slight change:
366the `QFunctionField` parameter `ncomp` has been changed to `size`. This change
367requires setting the previous value of `ncomp` to `ncomp*dim` when adding a
368`QFunctionField` with eval mode `CEED EVAL GRAD`.
369
370Additionally, new CPU backends
371were included in this release, such as the `/cpu/self/opt/*` backends (which are
372written in pure C and use partial **E-vectors** to improve performance) and the
373`/cpu/self/ref/memcheck` backend (which relies upon the
374[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user
375{ref}`CeedQFunction` have no undefined values).
376This release also included various performance improvements, bug fixes, new examples,
377and improved tests. Among these improvements, vectorized instructions for
378{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD`
379instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and
380identity Q-Functions were introduced, and the PETSc benchmark problems were expanded
381to include unstructured meshes handling were. For this expansion, the prior version of
382the PETSc BPs, which only included data associated with structured geometries, were
383renamed `bpsraw`, and the new version of the BPs, which can handle data associated
384with any unstructured geometry, were called `bps`. Additionally, other benchmark
385problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively),
386and BP5 and BP6 (the collocated versions---for which the quadrature points are the same
387as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc
388examples. Furthermoew, another standalone libCEED example, called `ex2`, which
389computes the surface area of a given mesh was added to this release.
390
391Backends available in this release:
392
393| CEED resource (`-ceed`)  | Backend                                             |
394|--------------------------|-----------------------------------------------------|
395| `/cpu/self/ref/serial`   | Serial reference implementation                     |
396| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
397| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks            |
398| `/cpu/self/opt/serial`   | Serial optimized C implementation                   |
399| `/cpu/self/opt/blocked`  | Blocked optimized C implementation                  |
400| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
401| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
402| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
403| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
404| `/cpu/occa`              | Serial OCCA kernels                                 |
405| `/gpu/occa`              | CUDA OCCA kernels                                   |
406| `/omp/occa`              | OpenMP OCCA kernels                                 |
407| `/ocl/occa`              | OpenCL OCCA kernels                                 |
408| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
409| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
410| `/gpu/cuda/shared`       | Optimized pure CUDA kernels using shared memory     |
411| `/gpu/cuda/gen`          | Optimized pure CUDA kernels using code generation   |
412| `/gpu/magma`             | CUDA MAGMA kernels                                  |
413
414Examples available in this release:
415
416:::{list-table}
417:header-rows: 1
418:widths: auto
419* - User code
420  - Example
421* - `ceed`
422  - * ex1 (volume)
423    * ex2 (surface)
424* - `mfem`
425  - * BP1 (scalar mass operator)
426    * BP3 (scalar Laplace operator)
427* - `petsc`
428  - * BP1 (scalar mass operator)
429    * BP2 (vector mass operator)
430    * BP3 (scalar Laplace operator)
431    * BP4 (vector Laplace operator)
432    * BP5 (collocated scalar Laplace operator)
433    * BP6 (collocated vector Laplace operator)
434    * Navier-Stokes
435* - `nek5000`
436  - * BP1 (scalar mass operator)
437    * BP3 (scalar Laplace operator)
438:::
439
440(v0-4)=
441
442## v0.4 (Apr 1, 2019)
443
444libCEED v0.4 was made again publicly available in the second full CEED software
445distribution, release CEED 2.0. This release contained notable features, such as
446four new CPU backends, two new GPU backends, CPU backend optimizations, initial
447support for operator composition, performance benchmarking, and a Navier-Stokes demo.
448The new CPU backends in this release came in two families. The `/cpu/self/*/serial`
449backends process one element at a time and are intended for meshes with a smaller number
450of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of
451eight interlaced elements and are intended for meshes with higher numbers of elements.
452The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU
453performance. The `/cpu/self/xsmm/*` backends rely upon the
454[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU
455performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA.
456The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable
457performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple
458parallelization approach, where each thread treats a finite element. Using just in time
459compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this
460backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend
461achieve good peak performance for 1D, 2D, and low order 3D problems, but performance
462deteriorates very quickly when threads run out of registers.
463
464A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED
465examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`).
466This example solves the time-dependent Navier-Stokes equations of compressible gas
467dynamics in a static Eulerian three-dimensional frame, using structured high-order
468finite/spectral element spatial discretizations and explicit high-order time-stepping
469(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc,
470so that the pointwise physics (defined at quadrature points) is separated from the
471parallelization and meshing concerns.
472
473Backends available in this release:
474
475| CEED resource (`-ceed`)  | Backend                                             |
476|--------------------------|-----------------------------------------------------|
477| `/cpu/self/ref/serial`   | Serial reference implementation                     |
478| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
479| `/cpu/self/tmpl`         | Backend template, defaults to `/cpu/self/blocked`   |
480| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
481| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
482| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
483| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
484| `/cpu/occa`              | Serial OCCA kernels                                 |
485| `/gpu/occa`              | CUDA OCCA kernels                                   |
486| `/omp/occa`              | OpenMP OCCA kernels                                 |
487| `/ocl/occa`              | OpenCL OCCA kernels                                 |
488| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
489| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
490| `/gpu/magma`             | CUDA MAGMA kernels                                  |
491
492Examples available in this release:
493
494:::{list-table}
495:header-rows: 1
496:widths: auto
497* - User code
498  - Example
499* - `ceed`
500  - * ex1 (volume)
501* - `mfem`
502  - * BP1 (scalar mass operator)
503    * BP3 (scalar Laplace operator)
504* - `petsc`
505  - * BP1 (scalar mass operator)
506    * BP3 (scalar Laplace operator)
507    * Navier-Stokes
508* - `nek5000`
509  - * BP1 (scalar mass operator)
510    * BP3 (scalar Laplace operator)
511:::
512
513(v0-3)=
514
515## v0.3 (Sep 30, 2018)
516
517Notable features in this release include active/passive field interface, support for
518non-tensor bases, backend optimization, and improved Fortran interface. This release
519also focused on providing improved continuous integration, and many new tests with code
520coverage reports of about 90%. This release also provided a significant change to the
521public interface: a {ref}`CeedQFunction` can take any number of named input and output
522arguments while {ref}`CeedOperator` connects them to the actual data, which may be
523supplied explicitly to `CeedOperatorApply()` (active) or separately via
524`CeedOperatorSetField()` (passive). This interface change enables reusable libraries
525of CeedQFunctions and composition of block solvers constructed using
526{ref}`CeedOperator`. A concept of blocked restriction was added to this release and
527used in an optimized CPU backend. Although this is typically not visible to the user,
528it enables effective use of arbitrary-length SIMD while maintaining cache locality.
529This CPU backend also implements an algebraic factorization of tensor product gradients
530to perform fewer operations than standard application of interpolation and
531differentiation from nodes to quadrature points. This algebraic formulation
532automatically supports non-polynomial and non-interpolatory bases, thus is more general
533than the more common derivation in terms of Lagrange polynomials on the quadrature points.
534
535Backends available in this release:
536
537| CEED resource (`-ceed`) | Backend                                             |
538|-------------------------|-----------------------------------------------------|
539| `/cpu/self/blocked`     | Blocked reference implementation                    |
540| `/cpu/self/ref`         | Serial reference implementation                     |
541| `/cpu/self/tmpl`        | Backend template, defaults to `/cpu/self/blocked`   |
542| `/cpu/occa`             | Serial OCCA kernels                                 |
543| `/gpu/occa`             | CUDA OCCA kernels                                   |
544| `/omp/occa`             | OpenMP OCCA kernels                                 |
545| `/ocl/occa`             | OpenCL OCCA kernels                                 |
546| `/gpu/magma`            | CUDA MAGMA 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    * BP3 (scalar Laplace operator)
563* - `nek5000`
564  - * BP1 (scalar mass operator)
565    * BP3 (scalar Laplace operator)
566:::
567
568(v0-21)=
569
570## v0.21 (Sep 30, 2018)
571
572A MAGMA backend (which relies upon the
573[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this
574release. This initial integration set up the framework of using MAGMA and provided the
575libCEED functionality through MAGMA kernels as one of libCEED’s computational backends.
576As any other backend, the MAGMA backend provides extended basic data structures for
577{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements
578the fundamental CEED building blocks to work with the new data structures.
579In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data
580but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled
581internally, and thus seamlessly to the user, through the functions/methods that are
582provided to support them.
583
584Backends available in this release:
585
586| CEED resource (`-ceed`) | Backend                         |
587|-------------------------|---------------------------------|
588| `/cpu/self`             | Serial reference implementation |
589| `/cpu/occa`             | Serial OCCA kernels             |
590| `/gpu/occa`             | CUDA OCCA kernels               |
591| `/omp/occa`             | OpenMP OCCA kernels             |
592| `/ocl/occa`             | OpenCL OCCA kernels             |
593| `/gpu/magma`            | CUDA MAGMA kernels              |
594
595Examples available in this release:
596
597:::{list-table}
598:header-rows: 1
599:widths: auto
600* - User code
601  - Example
602* - `ceed`
603  - * ex1 (volume)
604* - `mfem`
605  - * BP1 (scalar mass operator)
606    * BP3 (scalar Laplace operator)
607* - `petsc`
608  - * BP1 (scalar mass operator)
609* - `nek5000`
610  - * BP1 (scalar mass operator)
611:::
612
613(v0-2)=
614
615## v0.2 (Mar 30, 2018)
616
617libCEED was made publicly available the first full CEED software distribution, release
618CEED 1.0. The distribution was made available using the Spack package manager to provide
619a common, easy-to-use build environment, where the user can build the CEED distribution
620with all dependencies. This release included a new Fortran interface for the library.
621This release also contained major improvements in the OCCA backend (including a new
622`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to
623compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an
624`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application
625of the Laplace operator) was also added to this release.
626
627Backends available in this release:
628
629| CEED resource (`-ceed`) | Backend                         |
630|-------------------------|---------------------------------|
631| `/cpu/self`             | Serial reference implementation |
632| `/cpu/occa`             | Serial OCCA kernels             |
633| `/gpu/occa`             | CUDA OCCA kernels               |
634| `/omp/occa`             | OpenMP OCCA kernels             |
635| `/ocl/occa`             | OpenCL OCCA kernels             |
636
637Examples available in this release:
638
639:::{list-table}
640:header-rows: 1
641:widths: auto
642* - User code
643  - Example
644* - `ceed`
645  - * ex1 (volume)
646* - `mfem`
647  - * BP1 (scalar mass operator)
648    * BP3 (scalar Laplace operator)
649* - `petsc`
650  - * BP1 (scalar mass operator)
651* - `nek5000`
652  - * BP1 (scalar mass operator)
653:::
654
655(v0-1)=
656
657## v0.1 (Jan 3, 2018)
658
659Initial low-level API of the CEED project. The low-level API provides a set of Finite
660Elements kernels and components for writing new low-level kernels. Examples include:
661vector and sparse linear algebra, element matrix assembly over a batch of elements,
662partial assembly and action for efficient high-order operators like mass, diffusion,
663advection, etc. The main goal of the low-level API is to establish the basis for the
664high-level API. Also, identifying such low-level kernels and providing a reference
665implementation for them serves as the basis for specialized backend implementations.
666This release contained several backends: `/cpu/self`, and backends which rely upon the
667[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`,
668`/gpu/occa`, and `/omp/occa`.
669It also included several examples, in the `examples` folder:
670A standalone code that shows the usage of libCEED (with no external
671dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1
672(with the application of the mass operator); and a `petsc` example to perform BP1
673(with the application of the mass operator).
674
675Backends available in this release:
676
677| CEED resource (`-ceed`) | Backend                         |
678|-------------------------|---------------------------------|
679| `/cpu/self`             | Serial reference implementation |
680| `/cpu/occa`             | Serial OCCA kernels             |
681| `/gpu/occa`             | CUDA OCCA kernels               |
682| `/omp/occa`             | OpenMP OCCA kernels             |
683
684Examples available in this release:
685
686| User code             | Example                           |
687|-----------------------|-----------------------------------|
688| `ceed`                | ex1 (scalar Laplace operator)     |
689| `mfem`                | BP1 (scalar mass operator)        |
690| `petsc`               | BP1 (scalar mass operator)        |
691```
692