1# Changes/Release Notes 2 3On this page we provide a summary of the main API changes, new features and examples for each release of libCEED. 4 5(main)= 6 7## Current `main` branch 8 9### Interface changes 10 11- Added {c:func}`CeedOperatorSetName` for more readable {c:func}`CeedOperatorView` output. 12- Added {c:func}`CeedBasisCreateProjection` to facilitate interpolation between nodes for separate `CeedBases`. 13 14### New features 15 16- Update `/cpu/self/memcheck/*` backends to help verify `CeedQFunctionContext` data sizes provided by user. 17- Added `CeedInt_FMT` to support potential future use of larger interger sizes. 18 19### Bugfix 20 21- Fix bug in setting device id for GPU backends. 22- Fix storing of indices for `CeedElemRestriction` on the host with GPU backends. 23- Fix `CeedElemRestriction` sizing for {c:func}`CeedOperatorAssemblePointBlockDiagonal`. 24- Fix bugs in CPU implementation of {c:func}`CeedOperatorLinearAssemble` when there are different number of active input modes and active output modes. 25 26### Examples 27 28- Added various performance enhancements for {ref}`example-petsc-navier-stokes`. 29- Refactored {ref}`example-petsc-navier-stokes` to improve code reuse. 30- Added Shock Tube, Channel, and Flat Plate boundary layer problems to {ref}`example-petsc-navier-stokes`. 31- Added ability to use QFunctions for strong STG inflow in {ref}`example-petsc-navier-stokes`. 32 33(v0-10-1)= 34 35## v0.10.1 (Apr 11, 2022) 36 37### Interface changes 38 39- Added {c:func}`CeedQFunctionSetUserFlopsEstimate` and {c:func}`CeedOperatorGetFlopsEstimate` to facilitate estimating FLOPs in operator application. 40 41### New features 42 43- Switched MAGMA backends to use runtime compilation for tensor basis kernels (and element restriction kernels, in non-deterministic `/gpu/*/magma` backends). 44This reduces time to compile the library and increases the range of parameters for which the MAGMA tensor basis kernels will work. 45 46### Bugfix 47 48- Install JiT source files in install directory to fix GPU functionality for installed libCEED. 49 50(v0-10)= 51 52## v0.10 (Mar 21, 2022) 53 54### Interface changes 55 56- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields. 57- 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`. 58- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added. 59- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients. 60- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces. 61- Warning added when compiling OCCA backend to alert users that this backend is experimental. 62- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`. 63- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name. 64- 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`. 65- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitalized arrays; require initalized data for {c:func}`CeedVectorGetArray`. 66- 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. 67- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retreive user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`. 68- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums. 69- 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. 70- Added {c:func}`CeedOperatorSetQFunctionUpdated` to facilitate QFunction data re-use between operators sharing the same quadrature space, such as in a multigrid hierarchy. 71- Added {c:func}`CeedOperatorGetActiveVectorLengths` to get shape of CeedOperator. 72 73### New features 74 75- `CeedScalar` can now be set as `float` or `double` at compile time. 76- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends. 77- 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. 78- Remove need to guard library headers in QFunction source for code generation backends. 79- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present. 80- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally. 81- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file. 82- Added support for non-tensor H(div) elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor. 83- 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. 84- Added support for element matrix assembly in GPU backends. 85 86### Maintainability 87 88- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage. 89- `Include-what-you-use` makefile target added as `make iwyu`. 90- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase. 91- Put GPU JiTed kernel source code into separate files. 92- 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. 93 94(v0-9)= 95 96## v0.9 (Jul 6, 2021) 97 98### Interface changes 99 100- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact. 101 102### New features 103 104- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use. 105- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends. 106- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations. 107- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed). 108 109### Performance improvements 110 111### Examples 112 113- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations. 114- Fluid mechanics example adds GPU support and improves modularity. 115 116### Deprecated backends 117 118- 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. 119 120(v0-8)= 121 122## v0.8 (Mar 31, 2021) 123 124### Interface changes 125 126- Error handling improved to include enumerated error codes for C interface return values. 127- 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. 128 129### New features 130 131- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features. 132- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly. 133- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators. 134 135### Performance improvements 136 137- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`. 138- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`. 139 140### Examples 141 142- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions. 143- {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. 144- {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. 145- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree. 146 147(v0-7)= 148 149## v0.7 (Sep 29, 2020) 150 151### Interface changes 152 153- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors. 154 As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`. 155 These changes improve support for mixed finite element methods. 156- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature. 157- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity. 158- 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. 159- 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. 160- 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. 161- 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. 162 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`. 163- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory. 164- 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. 165 166### New features 167 168- New HIP backend: `/gpu/hip/ref`. 169- CeedQFunction support for user `CUfunction`s in some backends 170 171### Performance improvements 172 173- OCCA backend rebuilt to facilitate future performance enhancements. 174- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`. 175 176### Examples 177 178- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions. 179 180### Deprecated backends 181 182- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`. 183 184(v0-6)= 185 186## v0.6 (Mar 29, 2020) 187 188libCEED v0.6 contains numerous new features and examples, as well as expanded 189documentation in [this new website](https://libceed.org). 190 191### New features 192 193- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly 194 1-1 correspondence with the C interface, plus some convenience features. For instance, 195 data stored in the {cpp:type}`CeedVector` structure are available without copy as 196 {py:class}`numpy.ndarray`. Short tutorials are provided in 197 [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/). 198- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point, 199 {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal 200 ({c:func}`CeedOperatorAssembleLinearDiagonal`). These operations are useful for 201 preconditioning ingredients and are used in the libCEED's multigrid examples. 202- The inverse of separable operators can be obtained using 203 {c:func}`CeedOperatorCreateFDMElementInverse` and applied with 204 {c:func}`CeedOperatorApply`. This is a useful preconditioning ingredient, 205 especially for Laplacians and related operators. 206- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`, 207 {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`. 208- Make public accessors for various attributes to facilitate writing composable code. 209- New backend: `/cpu/self/memcheck/serial`. 210- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA 211 backends. (Single source is coming soon for OCCA backends.) 212- Fix some missing edge cases in CUDA backend. 213 214### Performance Improvements 215 216- MAGMA backend performance optimization and non-tensor bases. 217- No-copy optimization in {c:func}`CeedOperatorApply`. 218 219### Interface changes 220 221- Replace {code}`CeedElemRestrictionCreateIdentity` and 222 {code}`CeedElemRestrictionCreateBlocked` with more flexible 223 {c:func}`CeedElemRestrictionCreateStrided` and 224 {c:func}`CeedElemRestrictionCreateBlockedStrided`. 225- Add arguments to {c:func}`CeedQFunctionCreateIdentity`. 226- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification 227 with {cpp:enum}`CeedInterlaceMode`. This is now an attribute of the 228 {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no 229 longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and 230 {c:func}`CeedElemRestrictionApply`. 231 232### Examples 233 234libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`. 235Notable additions include: 236 237- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of 238 a domain in 1, 2, and 3 dimensions by applying a Laplacian. 239 240- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area 241 of domains (like the cube and sphere) by direct integration on a surface mesh; 242 demonstrates geometric dimension different from topological dimension. 243 244- PETSc {ref}`example-petsc-bps`: 245 246 - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support. 247 - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements 248 and transparent CUDA support. 249 - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`): 250 generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric 251 dimension different from topological dimension. 252 253- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid 254 solver with algebraic multigrid coarse solve. 255 256- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly 257 `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`), 258 implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and 259 quasi-2D computational domain support. 260 261- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for 262 linear elasticity, small-strain hyperelasticity, and globalized finite-strain 263 hyperelasticity using p-multigrid with algebraic multigrid coarse solve. 264 265(v0-5)= 266 267## v0.5 (Sep 18, 2019) 268 269For this release, several improvements were made. Two new CUDA backends were added to 270the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art 271performance using single-source {ref}`CeedQFunction`. From this release, users 272can define Q-Functions in a single source code independently of the targeted backend 273with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU 274compilation of the user provided {ref}`CeedQFunction` code. To allow a unified 275declaration, the {ref}`CeedQFunction` API has undergone a slight change: 276the `QFunctionField` parameter `ncomp` has been changed to `size`. This change 277requires setting the previous value of `ncomp` to `ncomp*dim` when adding a 278`QFunctionField` with eval mode `CEED EVAL GRAD`. 279 280Additionally, new CPU backends 281were included in this release, such as the `/cpu/self/opt/*` backends (which are 282written in pure C and use partial **E-vectors** to improve performance) and the 283`/cpu/self/ref/memcheck` backend (which relies upon the 284[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user 285{ref}`CeedQFunction` have no undefined values). 286This release also included various performance improvements, bug fixes, new examples, 287and improved tests. Among these improvements, vectorized instructions for 288{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD` 289instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and 290identity Q-Functions were introduced, and the PETSc benchmark problems were expanded 291to include unstructured meshes handling were. For this expansion, the prior version of 292the PETSc BPs, which only included data associated with structured geometries, were 293renamed `bpsraw`, and the new version of the BPs, which can handle data associated 294with any unstructured geometry, were called `bps`. Additionally, other benchmark 295problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively), 296and BP5 and BP6 (the collocated versions---for which the quadrature points are the same 297as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc 298examples. Furthermoew, another standalone libCEED example, called `ex2`, which 299computes the surface area of a given mesh was added to this release. 300 301Backends available in this release: 302 303| CEED resource (`-ceed`) | Backend | 304|--------------------------|-----------------------------------------------------| 305| `/cpu/self/ref/serial` | Serial reference implementation | 306| `/cpu/self/ref/blocked` | Blocked reference implementation | 307| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks | 308| `/cpu/self/opt/serial` | Serial optimized C implementation | 309| `/cpu/self/opt/blocked` | Blocked optimized C implementation | 310| `/cpu/self/avx/serial` | Serial AVX implementation | 311| `/cpu/self/avx/blocked` | Blocked AVX implementation | 312| `/cpu/self/xsmm/serial` | Serial LIBXSMM implementation | 313| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation | 314| `/cpu/occa` | Serial OCCA kernels | 315| `/gpu/occa` | CUDA OCCA kernels | 316| `/omp/occa` | OpenMP OCCA kernels | 317| `/ocl/occa` | OpenCL OCCA kernels | 318| `/gpu/cuda/ref` | Reference pure CUDA kernels | 319| `/gpu/cuda/reg` | Pure CUDA kernels using one thread per element | 320| `/gpu/cuda/shared` | Optimized pure CUDA kernels using shared memory | 321| `/gpu/cuda/gen` | Optimized pure CUDA kernels using code generation | 322| `/gpu/magma` | CUDA MAGMA kernels | 323 324Examples available in this release: 325 326:::{list-table} 327:header-rows: 1 328:widths: auto 329* - User code 330 - Example 331* - `ceed` 332 - * ex1 (volume) 333 * ex2 (surface) 334* - `mfem` 335 - * BP1 (scalar mass operator) 336 * BP3 (scalar Laplace operator) 337* - `petsc` 338 - * BP1 (scalar mass operator) 339 * BP2 (vector mass operator) 340 * BP3 (scalar Laplace operator) 341 * BP4 (vector Laplace operator) 342 * BP5 (collocated scalar Laplace operator) 343 * BP6 (collocated vector Laplace operator) 344 * Navier-Stokes 345* - `nek5000` 346 - * BP1 (scalar mass operator) 347 * BP3 (scalar Laplace operator) 348::: 349 350(v0-4)= 351 352## v0.4 (Apr 1, 2019) 353 354libCEED v0.4 was made again publicly available in the second full CEED software 355distribution, release CEED 2.0. This release contained notable features, such as 356four new CPU backends, two new GPU backends, CPU backend optimizations, initial 357support for operator composition, performance benchmarking, and a Navier-Stokes demo. 358The new CPU backends in this release came in two families. The `/cpu/self/*/serial` 359backends process one element at a time and are intended for meshes with a smaller number 360of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of 361eight interlaced elements and are intended for meshes with higher numbers of elements. 362The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU 363performance. The `/cpu/self/xsmm/*` backends rely upon the 364[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU 365performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA. 366The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable 367performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple 368parallelization approach, where each thread treats a finite element. Using just in time 369compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this 370backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend 371achieve good peak performance for 1D, 2D, and low order 3D problems, but performance 372deteriorates very quickly when threads run out of registers. 373 374A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED 375examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`). 376This example solves the time-dependent Navier-Stokes equations of compressible gas 377dynamics in a static Eulerian three-dimensional frame, using structured high-order 378finite/spectral element spatial discretizations and explicit high-order time-stepping 379(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc, 380so that the pointwise physics (defined at quadrature points) is separated from the 381parallelization and meshing concerns. 382 383Backends available in this release: 384 385| CEED resource (`-ceed`) | Backend | 386|--------------------------|-----------------------------------------------------| 387| `/cpu/self/ref/serial` | Serial reference implementation | 388| `/cpu/self/ref/blocked` | Blocked reference implementation | 389| `/cpu/self/tmpl` | Backend template, defaults to `/cpu/self/blocked` | 390| `/cpu/self/avx/serial` | Serial AVX implementation | 391| `/cpu/self/avx/blocked` | Blocked AVX implementation | 392| `/cpu/self/xsmm/serial` | Serial LIBXSMM implementation | 393| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation | 394| `/cpu/occa` | Serial OCCA kernels | 395| `/gpu/occa` | CUDA OCCA kernels | 396| `/omp/occa` | OpenMP OCCA kernels | 397| `/ocl/occa` | OpenCL OCCA kernels | 398| `/gpu/cuda/ref` | Reference pure CUDA kernels | 399| `/gpu/cuda/reg` | Pure CUDA kernels using one thread per element | 400| `/gpu/magma` | CUDA MAGMA kernels | 401 402Examples available in this release: 403 404:::{list-table} 405:header-rows: 1 406:widths: auto 407* - User code 408 - Example 409* - `ceed` 410 - * ex1 (volume) 411* - `mfem` 412 - * BP1 (scalar mass operator) 413 * BP3 (scalar Laplace operator) 414* - `petsc` 415 - * BP1 (scalar mass operator) 416 * BP3 (scalar Laplace operator) 417 * Navier-Stokes 418* - `nek5000` 419 - * BP1 (scalar mass operator) 420 * BP3 (scalar Laplace operator) 421::: 422 423(v0-3)= 424 425## v0.3 (Sep 30, 2018) 426 427Notable features in this release include active/passive field interface, support for 428non-tensor bases, backend optimization, and improved Fortran interface. This release 429also focused on providing improved continuous integration, and many new tests with code 430coverage reports of about 90%. This release also provided a significant change to the 431public interface: a {ref}`CeedQFunction` can take any number of named input and output 432arguments while {ref}`CeedOperator` connects them to the actual data, which may be 433supplied explicitly to `CeedOperatorApply()` (active) or separately via 434`CeedOperatorSetField()` (passive). This interface change enables reusable libraries 435of CeedQFunctions and composition of block solvers constructed using 436{ref}`CeedOperator`. A concept of blocked restriction was added to this release and 437used in an optimized CPU backend. Although this is typically not visible to the user, 438it enables effective use of arbitrary-length SIMD while maintaining cache locality. 439This CPU backend also implements an algebraic factorization of tensor product gradients 440to perform fewer operations than standard application of interpolation and 441differentiation from nodes to quadrature points. This algebraic formulation 442automatically supports non-polynomial and non-interpolatory bases, thus is more general 443than the more common derivation in terms of Lagrange polynomials on the quadrature points. 444 445Backends available in this release: 446 447| CEED resource (`-ceed`) | Backend | 448|-------------------------|-----------------------------------------------------| 449| `/cpu/self/blocked` | Blocked reference implementation | 450| `/cpu/self/ref` | Serial reference implementation | 451| `/cpu/self/tmpl` | Backend template, defaults to `/cpu/self/blocked` | 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 * BP3 (scalar Laplace operator) 473* - `nek5000` 474 - * BP1 (scalar mass operator) 475 * BP3 (scalar Laplace operator) 476::: 477 478(v0-21)= 479 480## v0.21 (Sep 30, 2018) 481 482A MAGMA backend (which relies upon the 483[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this 484release. This initial integration set up the framework of using MAGMA and provided the 485libCEED functionality through MAGMA kernels as one of libCEED’s computational backends. 486As any other backend, the MAGMA backend provides extended basic data structures for 487{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements 488the fundamental CEED building blocks to work with the new data structures. 489In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data 490but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled 491internally, and thus seamlessly to the user, through the functions/methods that are 492provided to support them. 493 494Backends available in this release: 495 496| CEED resource (`-ceed`) | Backend | 497|-------------------------|---------------------------------| 498| `/cpu/self` | Serial reference implementation | 499| `/cpu/occa` | Serial OCCA kernels | 500| `/gpu/occa` | CUDA OCCA kernels | 501| `/omp/occa` | OpenMP OCCA kernels | 502| `/ocl/occa` | OpenCL OCCA kernels | 503| `/gpu/magma` | CUDA MAGMA kernels | 504 505Examples available in this release: 506 507:::{list-table} 508:header-rows: 1 509:widths: auto 510* - User code 511 - Example 512* - `ceed` 513 - * ex1 (volume) 514* - `mfem` 515 - * BP1 (scalar mass operator) 516 * BP3 (scalar Laplace operator) 517* - `petsc` 518 - * BP1 (scalar mass operator) 519* - `nek5000` 520 - * BP1 (scalar mass operator) 521::: 522 523(v0-2)= 524 525## v0.2 (Mar 30, 2018) 526 527libCEED was made publicly available the first full CEED software distribution, release 528CEED 1.0. The distribution was made available using the Spack package manager to provide 529a common, easy-to-use build environment, where the user can build the CEED distribution 530with all dependencies. This release included a new Fortran interface for the library. 531This release also contained major improvements in the OCCA backend (including a new 532`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to 533compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an 534`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application 535of the Laplace operator) was also added to this release. 536 537Backends available in this release: 538 539| CEED resource (`-ceed`) | Backend | 540|-------------------------|---------------------------------| 541| `/cpu/self` | Serial reference implementation | 542| `/cpu/occa` | Serial OCCA kernels | 543| `/gpu/occa` | CUDA OCCA kernels | 544| `/omp/occa` | OpenMP OCCA kernels | 545| `/ocl/occa` | OpenCL OCCA kernels | 546 547Examples available in this release: 548 549:::{list-table} 550:header-rows: 1 551:widths: auto 552* - User code 553 - Example 554* - `ceed` 555 - * ex1 (volume) 556* - `mfem` 557 - * BP1 (scalar mass operator) 558 * BP3 (scalar Laplace operator) 559* - `petsc` 560 - * BP1 (scalar mass operator) 561* - `nek5000` 562 - * BP1 (scalar mass operator) 563::: 564 565(v0-1)= 566 567## v0.1 (Jan 3, 2018) 568 569Initial low-level API of the CEED project. The low-level API provides a set of Finite 570Elements kernels and components for writing new low-level kernels. Examples include: 571vector and sparse linear algebra, element matrix assembly over a batch of elements, 572partial assembly and action for efficient high-order operators like mass, diffusion, 573advection, etc. The main goal of the low-level API is to establish the basis for the 574high-level API. Also, identifying such low-level kernels and providing a reference 575implementation for them serves as the basis for specialized backend implementations. 576This release contained several backends: `/cpu/self`, and backends which rely upon the 577[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`, 578`/gpu/occa`, and `/omp/occa`. 579It also included several examples, in the `examples` folder: 580A standalone code that shows the usage of libCEED (with no external 581dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1 582(with the application of the mass operator); and a `petsc` example to perform BP1 583(with the application of the mass operator). 584 585Backends available in this release: 586 587| CEED resource (`-ceed`) | Backend | 588|-------------------------|---------------------------------| 589| `/cpu/self` | Serial reference implementation | 590| `/cpu/occa` | Serial OCCA kernels | 591| `/gpu/occa` | CUDA OCCA kernels | 592| `/omp/occa` | OpenMP OCCA kernels | 593 594Examples available in this release: 595 596| User code | Example | 597|-----------------------|-----------------------------------| 598| `ceed` | ex1 (scalar Laplace operator) | 599| `mfem` | BP1 (scalar mass operator) | 600| `petsc` | BP1 (scalar mass operator) | 601``` 602