xref: /libCEED/doc/sphinx/source/releasenotes.md (revision ed094490f53e580908aa80e9fe815a6fd76d7526)
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- Change QFunction source include file handling in JiT compilers
15    - Add `CEED_RUNNING_JIT_PASS` compiler definition for wrapping header files that device JiT compilers cannot read
16    - Users should now prefer `#include <ceed/types.h>` rather than `#include <ceed.h>` in QFunction source files
17- Require use of `Ceed*Destroy()` on Ceed objects returned from `Ceed*Get*()`.
18- Rename `CeedCompositeOperatorCreate()` to `CeedOperatorCreateComposite()` for uniformity.
19- Rename `CeedCompositeOperator*()` to `CeedOperatorComposite*()` for uniformity.
20
21### New features
22
23- Add `CeedOperatorCreateAtPoints` which evaluates the `CeedQFunction` at arbitrary locations in each element, for use in Particle in Cell, Material Point Method, and similar methods.
24- Add `CeedElemRestrictionGetLLayout` to provide L-vector layout for strided `CeedElemRestriction` created with `CEED_BACKEND_STRIDES`.
25- Add `CeedVectorReturnCeed` and similar when parent `Ceed` context for a libCEED object is only needed once in a calling scope.
26- Enable `#pragma once` for all JiT source; remove duplicate includes in JiT source string before compilation.
27- Allow user to set additional compiler options for CUDA and HIP JiT.
28Specifically, directories set with `CeedAddJitSourceRoot(ceed, "foo/bar")` will be used to set `-Ifoo/bar` and defines set with `CeedAddJitDefine(ceed, "foo=bar")` will be used to set `-Dfoo=bar`.
29- Added non-tensor basis support to code generation backends `/gpu/cuda/gen` and `/gpu/hip/gen`.
30- Added support to code generation backends `/gpu/cuda/gen` and `/gpu/hip/gen` for operators with both tensor and non-tensor bases.
31- Add `CeedGetGitVersion()` to access the Git commit and dirty state of the repository at build time.
32- Add `CeedGetBuildConfiguration()` to access compilers, flags, and related information about the build environment.
33
34### Examples
35
36- Add deal.II example with CEED BP suite.
37
38(v0-12)=
39
40## v0.12 (Oct 31, 2023)
41
42### Interface changes
43
44- Update `CeedOperatorContext*` functions to `CeedOperator*Context*` functions for consistency.
45For example, `CeedOperatorContextGetFieldLabel` was renamed to `CeedOperatorGetContextFieldLabel`.
46- Removed `CeedBasisSetNumQuadraturePoints` as redundant and bug-prone interface.
47
48### New features
49
50- Added {c:func}`CeedOperatorGetFieldByName` to access a specific `CeedOperatorField` by its name.
51- Update `/cpu/self/memcheck/*` backends to help verify `CeedVector` array access assumptions and `CeedQFunction` user output assumptions.
52- Update {c:func}`CeedOperatorLinearAssembleDiagonal` to provide default implementation that supports `CeedOperator` with multiple active bases.
53- Added Sycl backends `/gpu/sycl/ref`, `/gpu/sycl/shared`, and `/gpu/sycl/gen`.
54- Added {c:func}`CeedBasisApplyAtPoints` for evaluation of values and derivatives at arbitrary points inside elements.
55- Added support for non-tensor $H(\text{curl})$ finite element spaces with {c:func}`CeedBasisCreateHcurl`.
56- 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)).
57- Added {c:func}`CeedOperatorLinearAssemblePointBlockDiagonalSymbolic` to create COO mapping for mapping out of {c:func}`CeedOperatorLinearAssemblePointBlockDiagonal`.
58- Added support for application codes which manage multiple {ref}`Ceed` objects, parallelized across OpenMP threads.
59
60### Examples
61
62- Add `DMSwarm` example demonstrating interpolation from background mesh to swarm points and projection from swarm points to background mesh.
63
64#### {ref}`example-petsc-bps`
65
66- Requires PETSc version 3.19 or later.
67
68#### {ref}`example-petsc-navier-stokes`
69
70- Updated restart and checkpointing interface.
71- Add data-driven subgrid-stress model.
72- Add differential filtering of solution.
73- Add turbulence statistics collection over spanwise-symmetric geometries.
74- Add Taylor-Green vortex initial condition.
75- Add Riemann-based outflow boundary conditions.
76- Added vortex shedding and flow past cylinder example, including calculations for lift, drag, and heat transfer.
77- Add Internal Damping Layer (IDL) for helping turbulent simulation stability.
78- Derive `CeedBasis` from `PetscFE`, and various other internal maintainability updates.
79
80(v0-11)=
81
82## v0.11 (Dec 24, 2022)
83
84### Interface changes
85
86- Added {c:func}`CeedOperatorSetName` for more readable {c:func}`CeedOperatorView` output.
87- Added {c:func}`CeedBasisCreateProjection` to facilitate interpolation between nodes for separate `CeedBases`.
88- Rename and move {c:func}`CeedCompositeOperatorGetNumSub` and {c:func}`CeedCompositeOperatorGetSubList` to public interface.
89- Renamed `CEED_BASIS_COLLOCATED` to `CEED_BASIS_NONE` for clarity.
90Some 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.
91
92### New features
93
94- Update `/cpu/self/memcheck/*` backends to help verify `CeedQFunctionContext` data sizes provided by user.
95- Improved support for $H(\text{div})$ bases.
96- Added `CeedInt_FMT` to support potential future use of larger integer sizes.
97- Added `CEED_QFUNCTION_ATTR` for setting compiler attributes/pragmas to `CEED_QFUNCTION_HELPER` and `CEED_QFUNCTION`.
98- OCCA backend updated to latest OCCA release; DPC++ and OMP OCCA modes enabled.
99Due to a limitation of the OCCA parser, typedefs are required to use pointers to arrays in QFunctions with the OCCA backend.
100This issue will be fixed in a future OCCA release.
101
102### Bugfix
103
104- Fix bug in setting device id for GPU backends.
105- Fix storing of indices for `CeedElemRestriction` on the host with GPU backends.
106- Fix `CeedElemRestriction` sizing for {c:func}`CeedOperatorAssemblePointBlockDiagonal`.
107- Fix bugs in CPU implementation of {c:func}`CeedOperatorLinearAssemble` when there are different number of active input modes and active output modes.
108
109### Examples
110
111#### {ref}`example-petsc-navier-stokes`
112
113- Various performance enhancements, analytic matrix-free and assembled Jacobian, and PETSc solver configurations for GPUs.
114- Refactored to improve code reuse and modularity.
115- Support for primitive variables for more accurate boundary layers and all-speed flow.
116- Added $YZ\beta$ shock capturing scheme and Shock Tube example.
117- Added Channel example, with comparison to analytic solutions.
118- Added Flat Plate with boundary layer mesh and compressible Blasius inflow condition based on Chebyshev collocation solution of the Blasius equations.
119- Added strong and weak synthetic turbulence generation (STG) inflow boundary conditions.
120- Added "freestream" boundary conditions based on HLLC Riemann solver.
121- Automated stabilization coefficients for different basis degree.
122
123#### {ref}`example-petsc-bps`
124
125- Support for convergence studies.
126
127### Maintainability
128
129- 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.
130- Enabled support for `p > 8` for `/gpu/*/shared` backends.
131- Switch to `clang-format` over `astyle` for automatic formatting; Makefile command changed to `make format` from `make style`.
132- Improved test harness.
133
134(v0-10-1)=
135
136## v0.10.1 (Apr 11, 2022)
137
138### Interface changes
139
140- Added {c:func}`CeedQFunctionSetUserFlopsEstimate` and {c:func}`CeedOperatorGetFlopsEstimate` to facilitate estimating FLOPs in operator application.
141
142### New features
143
144- Switched MAGMA backends to use runtime compilation for tensor basis kernels (and element restriction kernels, in non-deterministic `/gpu/*/magma` backends).
145This reduces time to compile the library and increases the range of parameters for which the MAGMA tensor basis kernels will work.
146
147### Bugfix
148
149- Install JiT source files in install directory to fix GPU functionality for installed libCEED.
150
151(v0-10)=
152
153## v0.10 (Mar 21, 2022)
154
155### Interface changes
156
157- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields.
158- 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`.
159- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added.
160- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients.
161- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces.
162- Warning added when compiling OCCA backend to alert users that this backend is experimental.
163- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`.
164- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name.
165- 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`.
166- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitialized arrays; require initialized data for {c:func}`CeedVectorGetArray`.
167- 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.
168- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retrieve user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`.
169- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums.
170- 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.
171- Added {c:func}`CeedOperatorSetQFunctionUpdated` to facilitate QFunction data re-use between operators sharing the same quadrature space, such as in a multigrid hierarchy.
172- Added {c:func}`CeedOperatorGetActiveVectorLengths` to get shape of CeedOperator.
173
174### New features
175
176- `CeedScalar` can now be set as `float` or `double` at compile time.
177- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends.
178- 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.
179- Remove need to guard library headers in QFunction source for code generation backends.
180- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present.
181- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally.
182- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file.
183- Added support for non-tensor $H(\text{div})$ elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor.
184- 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.
185- Added support for element matrix assembly in GPU backends.
186
187### Maintainability
188
189- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage.
190- `Include-what-you-use` makefile target added as `make iwyu`.
191- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase.
192- Put GPU JiTed kernel source code into separate files.
193- 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.
194
195(v0-9)=
196
197## v0.9 (Jul 6, 2021)
198
199### Interface changes
200
201- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact.
202
203### New features
204
205- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use.
206- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends.
207- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations.
208- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed).
209
210### Performance improvements
211
212### Examples
213
214- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations.
215- Fluid mechanics example adds GPU support and improves modularity.
216
217### Deprecated backends
218
219- 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.
220
221(v0-8)=
222
223## v0.8 (Mar 31, 2021)
224
225### Interface changes
226
227- Error handling improved to include enumerated error codes for C interface return values.
228- 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.
229
230### New features
231
232- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features.
233- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly.
234- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators.
235
236### Performance improvements
237
238- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`.
239- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`.
240
241### Examples
242
243- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions.
244- {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.
245- {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.
246- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree.
247
248(v0-7)=
249
250## v0.7 (Sep 29, 2020)
251
252### Interface changes
253
254- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors.
255  As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`.
256  These changes improve support for mixed finite element methods.
257- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature.
258- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity.
259- 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.
260- 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.
261- 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.
262- 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.
263  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`.
264- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory.
265- 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.
266
267### New features
268
269- New HIP backend: `/gpu/hip/ref`.
270- CeedQFunction support for user `CUfunction`s in some backends
271
272### Performance improvements
273
274- OCCA backend rebuilt to facilitate future performance enhancements.
275- PETSc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`.
276
277### Examples
278
279- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions.
280
281### Deprecated backends
282
283- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`.
284
285(v0-6)=
286
287## v0.6 (Mar 29, 2020)
288
289libCEED v0.6 contains numerous new features and examples, as well as expanded
290documentation in [this new website](https://libceed.org).
291
292### New features
293
294- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly
295  1-1 correspondence with the C interface, plus some convenience features.  For instance,
296  data stored in the {cpp:type}`CeedVector` structure are available without copy as
297  {py:class}`numpy.ndarray`.  Short tutorials are provided in
298  [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/).
299- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point,
300  {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal
301  ({c:func}`CeedOperatorAssembleLinearDiagonal`).  These operations are useful for
302  preconditioning ingredients and are used in the libCEED's multigrid examples.
303- The inverse of separable operators can be obtained using
304  {c:func}`CeedOperatorCreateFDMElementInverse` and applied with
305  {c:func}`CeedOperatorApply`.  This is a useful preconditioning ingredient,
306  especially for Laplacians and related operators.
307- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`,
308  {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`.
309- Make public accessors for various attributes to facilitate writing composable code.
310- New backend: `/cpu/self/memcheck/serial`.
311- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA
312  backends.  (Single source is coming soon for OCCA backends.)
313- Fix some missing edge cases in CUDA backend.
314
315### Performance Improvements
316
317- MAGMA backend performance optimization and non-tensor bases.
318- No-copy optimization in {c:func}`CeedOperatorApply`.
319
320### Interface changes
321
322- Replace {code}`CeedElemRestrictionCreateIdentity` and
323  {code}`CeedElemRestrictionCreateBlocked` with more flexible
324  {c:func}`CeedElemRestrictionCreateStrided` and
325  {c:func}`CeedElemRestrictionCreateBlockedStrided`.
326- Add arguments to {c:func}`CeedQFunctionCreateIdentity`.
327- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification
328  with {cpp:enum}`CeedInterlaceMode`.  This is now an attribute of the
329  {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no
330  longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and
331  {c:func}`CeedElemRestrictionApply`.
332
333### Examples
334
335libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`.
336Notable additions include:
337
338- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of
339  a domain in 1, 2, and 3 dimensions by applying a Laplacian.
340
341- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area
342  of domains (like the cube and sphere) by direct integration on a surface mesh;
343  demonstrates geometric dimension different from topological dimension.
344
345- PETSc {ref}`example-petsc-bps`:
346
347  - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support.
348  - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements
349    and transparent CUDA support.
350  - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`):
351    generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric
352    dimension different from topological dimension.
353
354- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid
355  solver with algebraic multigrid coarse solve.
356
357- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly
358  `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`),
359  implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and
360  quasi-2D computational domain support.
361
362- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for
363  linear elasticity, small-strain hyperelasticity, and globalized finite-strain
364  hyperelasticity using p-multigrid with algebraic multigrid coarse solve.
365
366(v0-5)=
367
368## v0.5 (Sep 18, 2019)
369
370For this release, several improvements were made. Two new CUDA backends were added to
371the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art
372performance using single-source {ref}`CeedQFunction`. From this release, users
373can define Q-Functions in a single source code independently of the targeted backend
374with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU
375compilation of the user provided {ref}`CeedQFunction` code. To allow a unified
376declaration, the {ref}`CeedQFunction` API has undergone a slight change:
377the `QFunctionField` parameter `ncomp` has been changed to `size`. This change
378requires setting the previous value of `ncomp` to `ncomp*dim` when adding a
379`QFunctionField` with eval mode `CEED EVAL GRAD`.
380
381Additionally, new CPU backends
382were included in this release, such as the `/cpu/self/opt/*` backends (which are
383written in pure C and use partial **E-vectors** to improve performance) and the
384`/cpu/self/ref/memcheck` backend (which relies upon the
385[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user
386{ref}`CeedQFunction` have no undefined values).
387This release also included various performance improvements, bug fixes, new examples,
388and improved tests. Among these improvements, vectorized instructions for
389{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD`
390instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and
391identity Q-Functions were introduced, and the PETSc benchmark problems were expanded
392to include unstructured meshes handling were. For this expansion, the prior version of
393the PETSc BPs, which only included data associated with structured geometries, were
394renamed `bpsraw`, and the new version of the BPs, which can handle data associated
395with any unstructured geometry, were called `bps`. Additionally, other benchmark
396problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively),
397and BP5 and BP6 (the collocated versions---for which the quadrature points are the same
398as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc
399examples. Furthermoew, another standalone libCEED example, called `ex2`, which
400computes the surface area of a given mesh was added to this release.
401
402Backends available in this release:
403
404| CEED resource (`-ceed`)  | Backend                                             |
405|--------------------------|-----------------------------------------------------|
406| `/cpu/self/ref/serial`   | Serial reference implementation                     |
407| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
408| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks            |
409| `/cpu/self/opt/serial`   | Serial optimized C implementation                   |
410| `/cpu/self/opt/blocked`  | Blocked optimized C implementation                  |
411| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
412| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
413| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
414| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
415| `/cpu/occa`              | Serial OCCA kernels                                 |
416| `/gpu/occa`              | CUDA OCCA kernels                                   |
417| `/omp/occa`              | OpenMP OCCA kernels                                 |
418| `/ocl/occa`              | OpenCL OCCA kernels                                 |
419| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
420| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
421| `/gpu/cuda/shared`       | Optimized pure CUDA kernels using shared memory     |
422| `/gpu/cuda/gen`          | Optimized pure CUDA kernels using code generation   |
423| `/gpu/magma`             | CUDA MAGMA kernels                                  |
424
425Examples available in this release:
426
427:::{list-table}
428:header-rows: 1
429:widths: auto
430* - User code
431  - Example
432* - `ceed`
433  - * ex1 (volume)
434    * ex2 (surface)
435* - `mfem`
436  - * BP1 (scalar mass operator)
437    * BP3 (scalar Laplace operator)
438* - `petsc`
439  - * BP1 (scalar mass operator)
440    * BP2 (vector mass operator)
441    * BP3 (scalar Laplace operator)
442    * BP4 (vector Laplace operator)
443    * BP5 (collocated scalar Laplace operator)
444    * BP6 (collocated vector Laplace operator)
445    * Navier-Stokes
446* - `nek5000`
447  - * BP1 (scalar mass operator)
448    * BP3 (scalar Laplace operator)
449:::
450
451(v0-4)=
452
453## v0.4 (Apr 1, 2019)
454
455libCEED v0.4 was made again publicly available in the second full CEED software
456distribution, release CEED 2.0. This release contained notable features, such as
457four new CPU backends, two new GPU backends, CPU backend optimizations, initial
458support for operator composition, performance benchmarking, and a Navier-Stokes demo.
459The new CPU backends in this release came in two families. The `/cpu/self/*/serial`
460backends process one element at a time and are intended for meshes with a smaller number
461of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of
462eight interlaced elements and are intended for meshes with higher numbers of elements.
463The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU
464performance. The `/cpu/self/xsmm/*` backends rely upon the
465[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU
466performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA.
467The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable
468performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple
469parallelization approach, where each thread treats a finite element. Using just in time
470compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this
471backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend
472achieve good peak performance for 1D, 2D, and low order 3D problems, but performance
473deteriorates very quickly when threads run out of registers.
474
475A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED
476examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`).
477This example solves the time-dependent Navier-Stokes equations of compressible gas
478dynamics in a static Eulerian three-dimensional frame, using structured high-order
479finite/spectral element spatial discretizations and explicit high-order time-stepping
480(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc,
481so that the pointwise physics (defined at quadrature points) is separated from the
482parallelization and meshing concerns.
483
484Backends available in this release:
485
486| CEED resource (`-ceed`)  | Backend                                             |
487|--------------------------|-----------------------------------------------------|
488| `/cpu/self/ref/serial`   | Serial reference implementation                     |
489| `/cpu/self/ref/blocked`  | Blocked reference implementation                    |
490| `/cpu/self/tmpl`         | Backend template, defaults to `/cpu/self/blocked`   |
491| `/cpu/self/avx/serial`   | Serial AVX implementation                           |
492| `/cpu/self/avx/blocked`  | Blocked AVX implementation                          |
493| `/cpu/self/xsmm/serial`  | Serial LIBXSMM implementation                       |
494| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation                      |
495| `/cpu/occa`              | Serial OCCA kernels                                 |
496| `/gpu/occa`              | CUDA OCCA kernels                                   |
497| `/omp/occa`              | OpenMP OCCA kernels                                 |
498| `/ocl/occa`              | OpenCL OCCA kernels                                 |
499| `/gpu/cuda/ref`          | Reference pure CUDA kernels                         |
500| `/gpu/cuda/reg`          | Pure CUDA kernels using one thread per element      |
501| `/gpu/magma`             | CUDA MAGMA kernels                                  |
502
503Examples available in this release:
504
505:::{list-table}
506:header-rows: 1
507:widths: auto
508* - User code
509  - Example
510* - `ceed`
511  - * ex1 (volume)
512* - `mfem`
513  - * BP1 (scalar mass operator)
514    * BP3 (scalar Laplace operator)
515* - `petsc`
516  - * BP1 (scalar mass operator)
517    * BP3 (scalar Laplace operator)
518    * Navier-Stokes
519* - `nek5000`
520  - * BP1 (scalar mass operator)
521    * BP3 (scalar Laplace operator)
522:::
523
524(v0-3)=
525
526## v0.3 (Sep 30, 2018)
527
528Notable features in this release include active/passive field interface, support for
529non-tensor bases, backend optimization, and improved Fortran interface. This release
530also focused on providing improved continuous integration, and many new tests with code
531coverage reports of about 90%. This release also provided a significant change to the
532public interface: a {ref}`CeedQFunction` can take any number of named input and output
533arguments while {ref}`CeedOperator` connects them to the actual data, which may be
534supplied explicitly to `CeedOperatorApply()` (active) or separately via
535`CeedOperatorSetField()` (passive). This interface change enables reusable libraries
536of CeedQFunctions and composition of block solvers constructed using
537{ref}`CeedOperator`. A concept of blocked restriction was added to this release and
538used in an optimized CPU backend. Although this is typically not visible to the user,
539it enables effective use of arbitrary-length SIMD while maintaining cache locality.
540This CPU backend also implements an algebraic factorization of tensor product gradients
541to perform fewer operations than standard application of interpolation and
542differentiation from nodes to quadrature points. This algebraic formulation
543automatically supports non-polynomial and non-interpolatory bases, thus is more general
544than the more common derivation in terms of Lagrange polynomials on the quadrature points.
545
546Backends available in this release:
547
548| CEED resource (`-ceed`) | Backend                                             |
549|-------------------------|-----------------------------------------------------|
550| `/cpu/self/blocked`     | Blocked reference implementation                    |
551| `/cpu/self/ref`         | Serial reference implementation                     |
552| `/cpu/self/tmpl`        | Backend template, defaults to `/cpu/self/blocked`   |
553| `/cpu/occa`             | Serial OCCA kernels                                 |
554| `/gpu/occa`             | CUDA OCCA kernels                                   |
555| `/omp/occa`             | OpenMP OCCA kernels                                 |
556| `/ocl/occa`             | OpenCL OCCA kernels                                 |
557| `/gpu/magma`            | CUDA MAGMA kernels                                  |
558
559Examples available in this release:
560
561:::{list-table}
562:header-rows: 1
563:widths: auto
564* - User code
565  - Example
566* - `ceed`
567  - * ex1 (volume)
568* - `mfem`
569  - * BP1 (scalar mass operator)
570    * BP3 (scalar Laplace operator)
571* - `petsc`
572  - * BP1 (scalar mass operator)
573    * BP3 (scalar Laplace operator)
574* - `nek5000`
575  - * BP1 (scalar mass operator)
576    * BP3 (scalar Laplace operator)
577:::
578
579(v0-21)=
580
581## v0.21 (Sep 30, 2018)
582
583A MAGMA backend (which relies upon the
584[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this
585release. This initial integration set up the framework of using MAGMA and provided the
586libCEED functionality through MAGMA kernels as one of libCEED’s computational backends.
587As any other backend, the MAGMA backend provides extended basic data structures for
588{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements
589the fundamental CEED building blocks to work with the new data structures.
590In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data
591but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled
592internally, and thus seamlessly to the user, through the functions/methods that are
593provided to support them.
594
595Backends available in this release:
596
597| CEED resource (`-ceed`) | Backend                         |
598|-------------------------|---------------------------------|
599| `/cpu/self`             | Serial reference implementation |
600| `/cpu/occa`             | Serial OCCA kernels             |
601| `/gpu/occa`             | CUDA OCCA kernels               |
602| `/omp/occa`             | OpenMP OCCA kernels             |
603| `/ocl/occa`             | OpenCL OCCA kernels             |
604| `/gpu/magma`            | CUDA MAGMA kernels              |
605
606Examples available in this release:
607
608:::{list-table}
609:header-rows: 1
610:widths: auto
611* - User code
612  - Example
613* - `ceed`
614  - * ex1 (volume)
615* - `mfem`
616  - * BP1 (scalar mass operator)
617    * BP3 (scalar Laplace operator)
618* - `petsc`
619  - * BP1 (scalar mass operator)
620* - `nek5000`
621  - * BP1 (scalar mass operator)
622:::
623
624(v0-2)=
625
626## v0.2 (Mar 30, 2018)
627
628libCEED was made publicly available the first full CEED software distribution, release
629CEED 1.0. The distribution was made available using the Spack package manager to provide
630a common, easy-to-use build environment, where the user can build the CEED distribution
631with all dependencies. This release included a new Fortran interface for the library.
632This release also contained major improvements in the OCCA backend (including a new
633`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to
634compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an
635`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application
636of the Laplace operator) was also added to this release.
637
638Backends available in this release:
639
640| CEED resource (`-ceed`) | Backend                         |
641|-------------------------|---------------------------------|
642| `/cpu/self`             | Serial reference implementation |
643| `/cpu/occa`             | Serial OCCA kernels             |
644| `/gpu/occa`             | CUDA OCCA kernels               |
645| `/omp/occa`             | OpenMP OCCA kernels             |
646| `/ocl/occa`             | OpenCL OCCA kernels             |
647
648Examples available in this release:
649
650:::{list-table}
651:header-rows: 1
652:widths: auto
653* - User code
654  - Example
655* - `ceed`
656  - * ex1 (volume)
657* - `mfem`
658  - * BP1 (scalar mass operator)
659    * BP3 (scalar Laplace operator)
660* - `petsc`
661  - * BP1 (scalar mass operator)
662* - `nek5000`
663  - * BP1 (scalar mass operator)
664:::
665
666(v0-1)=
667
668## v0.1 (Jan 3, 2018)
669
670Initial low-level API of the CEED project. The low-level API provides a set of Finite
671Elements kernels and components for writing new low-level kernels. Examples include:
672vector and sparse linear algebra, element matrix assembly over a batch of elements,
673partial assembly and action for efficient high-order operators like mass, diffusion,
674advection, etc. The main goal of the low-level API is to establish the basis for the
675high-level API. Also, identifying such low-level kernels and providing a reference
676implementation for them serves as the basis for specialized backend implementations.
677This release contained several backends: `/cpu/self`, and backends which rely upon the
678[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`,
679`/gpu/occa`, and `/omp/occa`.
680It also included several examples, in the `examples` folder:
681A standalone code that shows the usage of libCEED (with no external
682dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1
683(with the application of the mass operator); and a `petsc` example to perform BP1
684(with the application of the mass operator).
685
686Backends available in this release:
687
688| CEED resource (`-ceed`) | Backend                         |
689|-------------------------|---------------------------------|
690| `/cpu/self`             | Serial reference implementation |
691| `/cpu/occa`             | Serial OCCA kernels             |
692| `/gpu/occa`             | CUDA OCCA kernels               |
693| `/omp/occa`             | OpenMP OCCA kernels             |
694
695Examples available in this release:
696
697| User code             | Example                           |
698|-----------------------|-----------------------------------|
699| `ceed`                | ex1 (scalar Laplace operator)     |
700| `mfem`                | BP1 (scalar mass operator)        |
701| `petsc`               | BP1 (scalar mass operator)        |
702```
703