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