1# Changes/Release Notes 2 3On this page we provide a summary of the main API changes, new features and examples 4for each release of libCEED. 5 6(main)= 7 8## Current `main` branch 9 10### Interface changes 11 12- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields. 13- 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`. 14- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added. 15- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients. 16- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces. 17- Warning added when compiling OCCA backend to alert users that this backend is experimental. 18- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`. 19- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name. 20- 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`. 21- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitalized arrays; require initalized data for {c:func}`CeedVectorGetArray`. 22- 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. 23- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retreive user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`. 24- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums. 25- 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. 26 27### New features 28 29- `CeedScalar` can now be set as `float` or `double` at compile time. 30- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends. 31- 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. 32- Remove need to guard library headers in QFunction source for code generation backends. 33- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present. 34- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally. 35- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file. 36- Added support for non-tensor H(div) elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor. 37- 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. 38 39### Maintainability 40 41- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage. 42- `Include-what-you-use` makefile target added as `make iwyu`. 43- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase. 44- Put GPU JiTed kernel source code into separate files. 45- 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. 46 47(v0-9)= 48 49## v0.9 (Jul 6, 2021) 50 51### Interface changes 52 53- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact. 54 55### New features 56 57- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use. 58- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends. 59- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations. 60- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed). 61 62### Performance improvements 63 64### Examples 65 66- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations. 67- Fluid mechanics example adds GPU support and improves modularity. 68 69### Deprecated backends 70 71- 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. 72 73(v0-8)= 74 75## v0.8 (Mar 31, 2021) 76 77### Interface changes 78 79- Error handling improved to include enumerated error codes for C interface return values. 80- 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. 81 82### New features 83 84- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features. 85- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly. 86- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators. 87 88### Performance improvements 89 90- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`. 91- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`. 92 93### Examples 94 95- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions. 96- {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. 97- {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. 98- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree. 99 100(v0-7)= 101 102## v0.7 (Sep 29, 2020) 103 104### Interface changes 105 106- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors. 107 As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`. 108 These changes improve support for mixed finite element methods. 109- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature. 110- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity. 111- 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. 112- 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. 113- 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. 114- 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. 115 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`. 116- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory. 117- 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. 118 119### New features 120 121- New HIP backend: `/gpu/hip/ref`. 122- CeedQFunction support for user `CUfunction`s in some backends 123 124### Performance improvements 125 126- OCCA backend rebuilt to facilitate future performance enhancements. 127- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`. 128 129### Examples 130 131- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions. 132 133### Deprecated backends 134 135- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`. 136 137(v0-6)= 138 139## v0.6 (Mar 29, 2020) 140 141libCEED v0.6 contains numerous new features and examples, as well as expanded 142documentation in [this new website](https://libceed.readthedocs.io). 143 144### New features 145 146- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly 147 1-1 correspondence with the C interface, plus some convenience features. For instance, 148 data stored in the {cpp:type}`CeedVector` structure are available without copy as 149 {py:class}`numpy.ndarray`. Short tutorials are provided in 150 [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/). 151- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point, 152 {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal 153 ({c:func}`CeedOperatorAssembleLinearDiagonal`). These operations are useful for 154 preconditioning ingredients and are used in the libCEED's multigrid examples. 155- The inverse of separable operators can be obtained using 156 {c:func}`CeedOperatorCreateFDMElementInverse` and applied with 157 {c:func}`CeedOperatorApply`. This is a useful preconditioning ingredient, 158 especially for Laplacians and related operators. 159- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`, 160 {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`. 161- Make public accessors for various attributes to facilitate writing composable code. 162- New backend: `/cpu/self/memcheck/serial`. 163- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA 164 backends. (Single source is coming soon for OCCA backends.) 165- Fix some missing edge cases in CUDA backend. 166 167### Performance Improvements 168 169- MAGMA backend performance optimization and non-tensor bases. 170- No-copy optimization in {c:func}`CeedOperatorApply`. 171 172### Interface changes 173 174- Replace {code}`CeedElemRestrictionCreateIdentity` and 175 {code}`CeedElemRestrictionCreateBlocked` with more flexible 176 {c:func}`CeedElemRestrictionCreateStrided` and 177 {c:func}`CeedElemRestrictionCreateBlockedStrided`. 178- Add arguments to {c:func}`CeedQFunctionCreateIdentity`. 179- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification 180 with {cpp:enum}`CeedInterlaceMode`. This is now an attribute of the 181 {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no 182 longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and 183 {c:func}`CeedElemRestrictionApply`. 184 185### Examples 186 187libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`. 188Notable additions include: 189 190- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of 191 a domain in 1, 2, and 3 dimensions by applying a Laplacian. 192 193- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area 194 of domains (like the cube and sphere) by direct integration on a surface mesh; 195 demonstrates geometric dimension different from topological dimension. 196 197- PETSc {ref}`example-petsc-bps`: 198 199 - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support. 200 - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements 201 and transparent CUDA support. 202 - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`): 203 generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric 204 dimension different from topological dimension. 205 206- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid 207 solver with algebraic multigrid coarse solve. 208 209- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly 210 `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`), 211 implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and 212 quasi-2D computational domain support. 213 214- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for 215 linear elasticity, small-strain hyperelasticity, and globalized finite-strain 216 hyperelasticity using p-multigrid with algebraic multigrid coarse solve. 217 218(v0-5)= 219 220## v0.5 (Sep 18, 2019) 221 222For this release, several improvements were made. Two new CUDA backends were added to 223the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art 224performance using single-source {ref}`CeedQFunction`. From this release, users 225can define Q-Functions in a single source code independently of the targeted backend 226with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU 227compilation of the user provided {ref}`CeedQFunction` code. To allow a unified 228declaration, the {ref}`CeedQFunction` API has undergone a slight change: 229the `QFunctionField` parameter `ncomp` has been changed to `size`. This change 230requires setting the previous value of `ncomp` to `ncomp*dim` when adding a 231`QFunctionField` with eval mode `CEED EVAL GRAD`. 232 233Additionally, new CPU backends 234were included in this release, such as the `/cpu/self/opt/*` backends (which are 235written in pure C and use partial **E-vectors** to improve performance) and the 236`/cpu/self/ref/memcheck` backend (which relies upon the 237[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user 238{ref}`CeedQFunction` have no undefined values). 239This release also included various performance improvements, bug fixes, new examples, 240and improved tests. Among these improvements, vectorized instructions for 241{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD` 242instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and 243identity Q-Functions were introduced, and the PETSc benchmark problems were expanded 244to include unstructured meshes handling were. For this expansion, the prior version of 245the PETSc BPs, which only included data associated with structured geometries, were 246renamed `bpsraw`, and the new version of the BPs, which can handle data associated 247with any unstructured geometry, were called `bps`. Additionally, other benchmark 248problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively), 249and BP5 and BP6 (the collocated versions---for which the quadrature points are the same 250as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc 251examples. Furthermoew, another standalone libCEED example, called `ex2`, which 252computes the surface area of a given mesh was added to this release. 253 254Backends available in this release: 255 256| CEED resource (`-ceed`) | Backend | 257|--------------------------|-----------------------------------------------------| 258| `/cpu/self/ref/serial` | Serial reference implementation | 259| `/cpu/self/ref/blocked` | Blocked reference implementation | 260| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks | 261| `/cpu/self/opt/serial` | Serial optimized C implementation | 262| `/cpu/self/opt/blocked` | Blocked optimized C implementation | 263| `/cpu/self/avx/serial` | Serial AVX implementation | 264| `/cpu/self/avx/blocked` | Blocked AVX implementation | 265| `/cpu/self/xsmm/serial` | Serial LIBXSMM implementation | 266| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation | 267| `/cpu/occa` | Serial OCCA kernels | 268| `/gpu/occa` | CUDA OCCA kernels | 269| `/omp/occa` | OpenMP OCCA kernels | 270| `/ocl/occa` | OpenCL OCCA kernels | 271| `/gpu/cuda/ref` | Reference pure CUDA kernels | 272| `/gpu/cuda/reg` | Pure CUDA kernels using one thread per element | 273| `/gpu/cuda/shared` | Optimized pure CUDA kernels using shared memory | 274| `/gpu/cuda/gen` | Optimized pure CUDA kernels using code generation | 275| `/gpu/magma` | CUDA MAGMA kernels | 276 277Examples available in this release: 278 279:::{list-table} 280:header-rows: 1 281:widths: auto 282* - User code 283 - Example 284* - `ceed` 285 - * ex1 (volume) 286 * ex2 (surface) 287* - `mfem` 288 - * BP1 (scalar mass operator) 289 * BP3 (scalar Laplace operator) 290* - `petsc` 291 - * BP1 (scalar mass operator) 292 * BP2 (vector mass operator) 293 * BP3 (scalar Laplace operator) 294 * BP4 (vector Laplace operator) 295 * BP5 (collocated scalar Laplace operator) 296 * BP6 (collocated vector Laplace operator) 297 * Navier-Stokes 298* - `nek5000` 299 - * BP1 (scalar mass operator) 300 * BP3 (scalar Laplace operator) 301::: 302 303(v0-4)= 304 305## v0.4 (Apr 1, 2019) 306 307libCEED v0.4 was made again publicly available in the second full CEED software 308distribution, release CEED 2.0. This release contained notable features, such as 309four new CPU backends, two new GPU backends, CPU backend optimizations, initial 310support for operator composition, performance benchmarking, and a Navier-Stokes demo. 311The new CPU backends in this release came in two families. The `/cpu/self/*/serial` 312backends process one element at a time and are intended for meshes with a smaller number 313of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of 314eight interlaced elements and are intended for meshes with higher numbers of elements. 315The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU 316performance. The `/cpu/self/xsmm/*` backends rely upon the 317[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU 318performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA. 319The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable 320performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple 321parallelization approach, where each thread treats a finite element. Using just in time 322compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this 323backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend 324achieve good peak performance for 1D, 2D, and low order 3D problems, but performance 325deteriorates very quickly when threads run out of registers. 326 327A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED 328examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`). 329This example solves the time-dependent Navier-Stokes equations of compressible gas 330dynamics in a static Eulerian three-dimensional frame, using structured high-order 331finite/spectral element spatial discretizations and explicit high-order time-stepping 332(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc, 333so that the pointwise physics (defined at quadrature points) is separated from the 334parallelization and meshing concerns. 335 336Backends available in this release: 337 338| CEED resource (`-ceed`) | Backend | 339|--------------------------|-----------------------------------------------------| 340| `/cpu/self/ref/serial` | Serial reference implementation | 341| `/cpu/self/ref/blocked` | Blocked reference implementation | 342| `/cpu/self/tmpl` | Backend template, defaults to `/cpu/self/blocked` | 343| `/cpu/self/avx/serial` | Serial AVX implementation | 344| `/cpu/self/avx/blocked` | Blocked AVX implementation | 345| `/cpu/self/xsmm/serial` | Serial LIBXSMM implementation | 346| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation | 347| `/cpu/occa` | Serial OCCA kernels | 348| `/gpu/occa` | CUDA OCCA kernels | 349| `/omp/occa` | OpenMP OCCA kernels | 350| `/ocl/occa` | OpenCL OCCA kernels | 351| `/gpu/cuda/ref` | Reference pure CUDA kernels | 352| `/gpu/cuda/reg` | Pure CUDA kernels using one thread per element | 353| `/gpu/magma` | CUDA MAGMA kernels | 354 355Examples available in this release: 356 357:::{list-table} 358:header-rows: 1 359:widths: auto 360* - User code 361 - Example 362* - `ceed` 363 - * ex1 (volume) 364* - `mfem` 365 - * BP1 (scalar mass operator) 366 * BP3 (scalar Laplace operator) 367* - `petsc` 368 - * BP1 (scalar mass operator) 369 * BP3 (scalar Laplace operator) 370 * Navier-Stokes 371* - `nek5000` 372 - * BP1 (scalar mass operator) 373 * BP3 (scalar Laplace operator) 374::: 375 376(v0-3)= 377 378## v0.3 (Sep 30, 2018) 379 380Notable features in this release include active/passive field interface, support for 381non-tensor bases, backend optimization, and improved Fortran interface. This release 382also focused on providing improved continuous integration, and many new tests with code 383coverage reports of about 90%. This release also provided a significant change to the 384public interface: a {ref}`CeedQFunction` can take any number of named input and output 385arguments while {ref}`CeedOperator` connects them to the actual data, which may be 386supplied explicitly to `CeedOperatorApply()` (active) or separately via 387`CeedOperatorSetField()` (passive). This interface change enables reusable libraries 388of CeedQFunctions and composition of block solvers constructed using 389{ref}`CeedOperator`. A concept of blocked restriction was added to this release and 390used in an optimized CPU backend. Although this is typically not visible to the user, 391it enables effective use of arbitrary-length SIMD while maintaining cache locality. 392This CPU backend also implements an algebraic factorization of tensor product gradients 393to perform fewer operations than standard application of interpolation and 394differentiation from nodes to quadrature points. This algebraic formulation 395automatically supports non-polynomial and non-interpolatory bases, thus is more general 396than the more common derivation in terms of Lagrange polynomials on the quadrature points. 397 398Backends available in this release: 399 400| CEED resource (`-ceed`) | Backend | 401|-------------------------|-----------------------------------------------------| 402| `/cpu/self/blocked` | Blocked reference implementation | 403| `/cpu/self/ref` | Serial reference implementation | 404| `/cpu/self/tmpl` | Backend template, defaults to `/cpu/self/blocked` | 405| `/cpu/occa` | Serial OCCA kernels | 406| `/gpu/occa` | CUDA OCCA kernels | 407| `/omp/occa` | OpenMP OCCA kernels | 408| `/ocl/occa` | OpenCL OCCA kernels | 409| `/gpu/magma` | CUDA MAGMA kernels | 410 411Examples available in this release: 412 413:::{list-table} 414:header-rows: 1 415:widths: auto 416* - User code 417 - Example 418* - `ceed` 419 - * ex1 (volume) 420* - `mfem` 421 - * BP1 (scalar mass operator) 422 * BP3 (scalar Laplace operator) 423* - `petsc` 424 - * BP1 (scalar mass operator) 425 * BP3 (scalar Laplace operator) 426* - `nek5000` 427 - * BP1 (scalar mass operator) 428 * BP3 (scalar Laplace operator) 429::: 430 431(v0-21)= 432 433## v0.21 (Sep 30, 2018) 434 435A MAGMA backend (which relies upon the 436[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this 437release. This initial integration set up the framework of using MAGMA and provided the 438libCEED functionality through MAGMA kernels as one of libCEED’s computational backends. 439As any other backend, the MAGMA backend provides extended basic data structures for 440{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements 441the fundamental CEED building blocks to work with the new data structures. 442In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data 443but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled 444internally, and thus seamlessly to the user, through the functions/methods that are 445provided to support them. 446 447Backends available in this release: 448 449| CEED resource (`-ceed`) | Backend | 450|-------------------------|---------------------------------| 451| `/cpu/self` | Serial reference implementation | 452| `/cpu/occa` | Serial OCCA kernels | 453| `/gpu/occa` | CUDA OCCA kernels | 454| `/omp/occa` | OpenMP OCCA kernels | 455| `/ocl/occa` | OpenCL OCCA kernels | 456| `/gpu/magma` | CUDA MAGMA kernels | 457 458Examples available in this release: 459 460:::{list-table} 461:header-rows: 1 462:widths: auto 463* - User code 464 - Example 465* - `ceed` 466 - * ex1 (volume) 467* - `mfem` 468 - * BP1 (scalar mass operator) 469 * BP3 (scalar Laplace operator) 470* - `petsc` 471 - * BP1 (scalar mass operator) 472* - `nek5000` 473 - * BP1 (scalar mass operator) 474::: 475 476(v0-2)= 477 478## v0.2 (Mar 30, 2018) 479 480libCEED was made publicly available the first full CEED software distribution, release 481CEED 1.0. The distribution was made available using the Spack package manager to provide 482a common, easy-to-use build environment, where the user can build the CEED distribution 483with all dependencies. This release included a new Fortran interface for the library. 484This release also contained major improvements in the OCCA backend (including a new 485`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to 486compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an 487`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application 488of the Laplace operator) was also added to this release. 489 490Backends available in this release: 491 492| CEED resource (`-ceed`) | Backend | 493|-------------------------|---------------------------------| 494| `/cpu/self` | Serial reference 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 500Examples available in this release: 501 502:::{list-table} 503:header-rows: 1 504:widths: auto 505* - User code 506 - Example 507* - `ceed` 508 - * ex1 (volume) 509* - `mfem` 510 - * BP1 (scalar mass operator) 511 * BP3 (scalar Laplace operator) 512* - `petsc` 513 - * BP1 (scalar mass operator) 514* - `nek5000` 515 - * BP1 (scalar mass operator) 516::: 517 518(v0-1)= 519 520## v0.1 (Jan 3, 2018) 521 522Initial low-level API of the CEED project. The low-level API provides a set of Finite 523Elements kernels and components for writing new low-level kernels. Examples include: 524vector and sparse linear algebra, element matrix assembly over a batch of elements, 525partial assembly and action for efficient high-order operators like mass, diffusion, 526advection, etc. The main goal of the low-level API is to establish the basis for the 527high-level API. Also, identifying such low-level kernels and providing a reference 528implementation for them serves as the basis for specialized backend implementations. 529This release contained several backends: `/cpu/self`, and backends which rely upon the 530[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`, 531`/gpu/occa`, and `/omp/occa`. 532It also included several examples, in the `examples` folder: 533A standalone code that shows the usage of libCEED (with no external 534dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1 535(with the application of the mass operator); and a `petsc` example to perform BP1 536(with the application of the mass operator). 537 538Backends available in this release: 539 540| CEED resource (`-ceed`) | Backend | 541|-------------------------|---------------------------------| 542| `/cpu/self` | Serial reference implementation | 543| `/cpu/occa` | Serial OCCA kernels | 544| `/gpu/occa` | CUDA OCCA kernels | 545| `/omp/occa` | OpenMP OCCA kernels | 546 547Examples available in this release: 548 549| User code | Example | 550|-----------------------|-----------------------------------| 551| `ceed` | ex1 (scalar Laplace operator) | 552| `mfem` | BP1 (scalar mass operator) | 553| `petsc` | BP1 (scalar mass operator) | 554``` 555