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- Update `CeedOperatorContext*` functions to `CeedOperator*Context*` functions for consistency. 12For example, `CeedOperatorContextGetFieldLabel` was renamed to `CeedOperatorGetContextFieldLabel`. 13 14### New features 15 16- Added {c:func}`CeedOperatorGetFieldByName` to access a specific `CeedOperatorField` by its name. 17- Update `/cpu/self/memcheck/*` backends to help verify `CeedVector` array access assumptions and `CeedQFunction` user output assumptions. 18- Update {c:func}`CeedOperatorLinearAssembleDiagonal` to provide default implementation that supports `CeedOperator` with multiple active bases. 19- Added Sycl backends `/gpu/sycl/ref` and `/gpu/sycl/shared`. 20- Added {c:func}`CeedBasisApplyAtPoints` for evalution of values and derivaties at arbitrary points inside elements. 21- Added support for non-tensor $H(\text{curl})$ finite element spaces with {c:func}`CeedBasisCreateHcurl`. 22- Added {c:func}`CeedElemRestrictionCreateCurlOriented`, similar to {c:func}`CeedElemRestrictionCreateOriented`, for element restrictions requiring more general element transformations such as those for high-order $H(\text{curl})$ spaces on tetrahedra (see [https://dl.acm.org/doi/pdf/10.1145/3524456](https://dl.acm.org/doi/pdf/10.1145/3524456)). 23- Added {c:func}`CeedOperatorLinearAssemblePointBlockDiagonalSymbolic` to create COO mapping for mapping out of {c:func}`CeedOperatorLinearAssemblePointBlockDiagonal`. 24- Added support for application codes which manage multiple {ref}`Ceed` objects, parallelized across OpenMP threads. 25 26### Examples 27 28#### {ref}`example-petsc-bps` 29 30- Requires PETSc version 3.19 or later. 31 32(v0-11)= 33 34## v0.11 (Dec 24, 2022) 35 36### Interface changes 37 38- Added {c:func}`CeedOperatorSetName` for more readable {c:func}`CeedOperatorView` output. 39- Added {c:func}`CeedBasisCreateProjection` to facilitate interpolation between nodes for separate `CeedBases`. 40- Rename and move {c:func}`CeedCompositeOperatorGetNumSub` and {c:func}`CeedCompositeOperatorGetSubList` to public interface. 41- Renamed `CEED_BASIS_COLLOCATED` to `CEED_BASIS_NONE` for clarity. 42Some users previously misinterpreted a `CeedOperator` field using `CEED_BASIS_COLLOCATED` as meaning that the entire `CeedOperator` used a quadrature space that is collocated with the nodal space of the active bases. 43 44### New features 45 46- Update `/cpu/self/memcheck/*` backends to help verify `CeedQFunctionContext` data sizes provided by user. 47- Improved support for $H(\text{div})$ bases. 48- Added `CeedInt_FMT` to support potential future use of larger integer sizes. 49- Added `CEED_QFUNCTION_ATTR` for setting compiler attributes/pragmas to `CEED_QFUNCTION_HELPER` and `CEED_QFUNCTION`. 50- OCCA backend updated to latest OCCA release; DPC++ and OMP OCCA modes enabled. 51Due to a limitation of the OCCA parser, typedefs are required to use pointers to arrays in QFunctions with the OCCA backend. 52This issue will be fixed in a future OCCA release. 53 54### Bugfix 55 56- Fix bug in setting device id for GPU backends. 57- Fix storing of indices for `CeedElemRestriction` on the host with GPU backends. 58- Fix `CeedElemRestriction` sizing for {c:func}`CeedOperatorAssemblePointBlockDiagonal`. 59- Fix bugs in CPU implementation of {c:func}`CeedOperatorLinearAssemble` when there are different number of active input modes and active output modes. 60 61### Examples 62 63#### {ref}`example-petsc-navier-stokes` 64 65- Various performance enhancements, analytic matrix-free and assembled Jacobian, and PETSc solver configurations for GPUs. 66- Refactored to improve code reuse and modularity. 67- Support for primitive variables for more accurate boundary layers and all-speed flow. 68- Added $YZ\beta$ shock capturing scheme and Shock Tube example. 69- Added Channel example, with comparison to analytic solutions. 70- Added Flat Plate with boundary layer mesh and compressible Blasius inflow condition based on Chebyshev collocation solution of the Blasius equations. 71- Added strong and weak synthetic turbulence generation (STG) inflow boundary conditions. 72- Added "freestream" boundary conditions based on HLLC Riemann solver. 73- Automated stabilization coefficients for different basis degree. 74 75#### {ref}`example-petsc-bps` 76 77- Support for convergence studies. 78 79### Maintainability 80 81- Refactored `/gpu/cuda/shared` and `/gpu/cuda/gen` as well as `/gpu/hip/shared` and `/gpu/hip/gen` backend to improve maintainablity and reduce duplicated code. 82- Enabled support for `p > 8` for `/gpu/*/shared` backends. 83- Switch to `clang-format` over `astyle` for automatic formatting; Makefile command changed to `make format` from `make style`. 84- Improved test harness. 85 86(v0-10-1)= 87 88## v0.10.1 (Apr 11, 2022) 89 90### Interface changes 91 92- Added {c:func}`CeedQFunctionSetUserFlopsEstimate` and {c:func}`CeedOperatorGetFlopsEstimate` to facilitate estimating FLOPs in operator application. 93 94### New features 95 96- Switched MAGMA backends to use runtime compilation for tensor basis kernels (and element restriction kernels, in non-deterministic `/gpu/*/magma` backends). 97This reduces time to compile the library and increases the range of parameters for which the MAGMA tensor basis kernels will work. 98 99### Bugfix 100 101- Install JiT source files in install directory to fix GPU functionality for installed libCEED. 102 103(v0-10)= 104 105## v0.10 (Mar 21, 2022) 106 107### Interface changes 108 109- Update {c:func}`CeedQFunctionGetFields` and {c:func}`CeedOperatorGetFields` to include number of fields. 110- 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`. 111- Clarify and document conditions where `CeedQFunction` and `CeedOperator` become immutable and no further fields or suboperators can be added. 112- Add {c:func}`CeedOperatorLinearAssembleQFunctionBuildOrUpdate` to reduce object creation overhead in assembly of CeedOperator preconditioning ingredients. 113- Promote {c:func}`CeedOperatorCheckReady`to the public API to facilitate interactive interfaces. 114- Warning added when compiling OCCA backend to alert users that this backend is experimental. 115- `ceed-backend.h`, `ceed-hash.h`, and `ceed-khash.h` removed. Users should use `ceed/backend.h`, `ceed/hash.h`, and `ceed/khash.h`. 116- Added {c:func}`CeedQFunctionGetKernelName`; refactored {c:func}`CeedQFunctionGetSourcePath` to exclude function kernel name. 117- 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`. 118- Added {c:func}`CeedVectorGetArrayWrite` that allows access to uninitalized arrays; require initalized data for {c:func}`CeedVectorGetArray`. 119- 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. 120- Added {c:func}`CeedQFunctionContextGetFieldDescriptions` to retreive user defined descriptions of fields that are registered with `CeedQFunctionContextRegister*`. 121- Renamed `CeedElemTopology` entries for clearer namespacing between libCEED enums. 122- 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. 123- Added {c:func}`CeedOperatorSetQFunctionUpdated` to facilitate QFunction data re-use between operators sharing the same quadrature space, such as in a multigrid hierarchy. 124- Added {c:func}`CeedOperatorGetActiveVectorLengths` to get shape of CeedOperator. 125 126### New features 127 128- `CeedScalar` can now be set as `float` or `double` at compile time. 129- Added JiT utilities in `ceed/jit-tools.h` to reduce duplicated code in GPU backends. 130- 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. 131- Remove need to guard library headers in QFunction source for code generation backends. 132- `CeedDebugEnv()` macro created to provide debugging outputs when Ceed context is not present. 133- Added {c:func}`CeedStringAllocCopy` to reduce repeated code for copying strings internally. 134- Added {c:func}`CeedPathConcatenate` to facilitate loading kernel source files with a path relative to the current file. 135- Added support for non-tensor $H(\text{div})$ elements, to include CPU backend implementations and {c:func}`CeedBasisCreateHdiv` convenience constructor. 136- 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. 137- Added support for element matrix assembly in GPU backends. 138 139### Maintainability 140 141- Refactored preconditioner support internally to facilitate future development and improve GPU completeness/test coverage. 142- `Include-what-you-use` makefile target added as `make iwyu`. 143- Create backend constant `CEED_FIELD_MAX` to reduce magic numbers in codebase. 144- Put GPU JiTed kernel source code into separate files. 145- 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. 146 147(v0-9)= 148 149## v0.9 (Jul 6, 2021) 150 151### Interface changes 152 153- Minor modification in error handling macro to silence pedantic warnings when compiling with Clang, but no functional impact. 154 155### New features 156 157- Add {c:func}`CeedVectorAXPY` and {c:func}`CeedVectorPointwiseMult` as a convenience for stand-alone testing and internal use. 158- Add `CEED_QFUNCTION_HELPER` macro to properly annotate QFunction helper functions for code generation backends. 159- Add `CeedPragmaOptimizeOff` macro for code that is sensitive to floating point errors from fast math optimizations. 160- Rust support: split `libceed-sys` crate out of `libceed` and [publish both on crates.io](https://crates.io/crates/libceed). 161 162### Performance improvements 163 164### Examples 165 166- Solid mechanics mini-app updated to explore the performance impacts of various formulations in the initial and current configurations. 167- Fluid mechanics example adds GPU support and improves modularity. 168 169### Deprecated backends 170 171- 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. 172 173(v0-8)= 174 175## v0.8 (Mar 31, 2021) 176 177### Interface changes 178 179- Error handling improved to include enumerated error codes for C interface return values. 180- 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. 181 182### New features 183 184- Julia and Rust interfaces added, providing a nearly 1-1 correspondence with the C interface, plus some convenience features. 185- Static libraries can be built with `make STATIC=1` and the pkg-config file is installed accordingly. 186- Add {c:func}`CeedOperatorLinearAssembleSymbolic` and {c:func}`CeedOperatorLinearAssemble` to support full assembly of libCEED operators. 187 188### Performance improvements 189 190- New HIP MAGMA backends for hipMAGMA library users: `/gpu/hip/magma` and `/gpu/hip/magma/det`. 191- New HIP backends for improved tensor basis performance: `/gpu/hip/shared` and `/gpu/hip/gen`. 192 193### Examples 194 195- {ref}`example-petsc-elasticity` example updated with traction boundary conditions and improved Dirichlet boundary conditions. 196- {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. 197- {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. 198- {ref}`example-petsc-navier-stokes` example updated with support for performing convergence study and plotting order of convergence by polynomial degree. 199 200(v0-7)= 201 202## v0.7 (Sep 29, 2020) 203 204### Interface changes 205 206- Replace limited {code}`CeedInterlaceMode` with more flexible component stride {code}`compstride` in {code}`CeedElemRestriction` constructors. 207 As a result, the {code}`indices` parameter has been replaced with {code}`offsets` and the {code}`nnodes` parameter has been replaced with {code}`lsize`. 208 These changes improve support for mixed finite element methods. 209- Replace various uses of {code}`Ceed*Get*Status` with {code}`Ceed*Is*` in the backend API to match common nomenclature. 210- Replace {code}`CeedOperatorAssembleLinearDiagonal` with {c:func}`CeedOperatorLinearAssembleDiagonal` for clarity. 211- 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. 212- 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. 213- 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. 214- 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. 215 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`. 216- Added {code}`CeedQFunctionContext` object to manage user QFunction context data and reduce copies between device and host memory. 217- 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. 218 219### New features 220 221- New HIP backend: `/gpu/hip/ref`. 222- CeedQFunction support for user `CUfunction`s in some backends 223 224### Performance improvements 225 226- OCCA backend rebuilt to facilitate future performance enhancements. 227- Petsc BPs suite improved to reduce noise due to multiple calls to {code}`mpiexec`. 228 229### Examples 230 231- {ref}`example-petsc-elasticity` example updated with strain energy computation and more flexible boundary conditions. 232 233### Deprecated backends 234 235- The `/gpu/cuda/reg` backend has been removed, with its core features moved into `/gpu/cuda/ref` and `/gpu/cuda/shared`. 236 237(v0-6)= 238 239## v0.6 (Mar 29, 2020) 240 241libCEED v0.6 contains numerous new features and examples, as well as expanded 242documentation in [this new website](https://libceed.org). 243 244### New features 245 246- New Python interface using [CFFI](https://cffi.readthedocs.io/) provides a nearly 247 1-1 correspondence with the C interface, plus some convenience features. For instance, 248 data stored in the {cpp:type}`CeedVector` structure are available without copy as 249 {py:class}`numpy.ndarray`. Short tutorials are provided in 250 [Binder](https://mybinder.org/v2/gh/CEED/libCEED/main?urlpath=lab/tree/examples/tutorials/). 251- Linear QFunctions can be assembled as block-diagonal matrices (per quadrature point, 252 {c:func}`CeedOperatorAssembleLinearQFunction`) or to evaluate the diagonal 253 ({c:func}`CeedOperatorAssembleLinearDiagonal`). These operations are useful for 254 preconditioning ingredients and are used in the libCEED's multigrid examples. 255- The inverse of separable operators can be obtained using 256 {c:func}`CeedOperatorCreateFDMElementInverse` and applied with 257 {c:func}`CeedOperatorApply`. This is a useful preconditioning ingredient, 258 especially for Laplacians and related operators. 259- New functions: {c:func}`CeedVectorNorm`, {c:func}`CeedOperatorApplyAdd`, 260 {c:func}`CeedQFunctionView`, {c:func}`CeedOperatorView`. 261- Make public accessors for various attributes to facilitate writing composable code. 262- New backend: `/cpu/self/memcheck/serial`. 263- QFunctions using variable-length array (VLA) pointer constructs can be used with CUDA 264 backends. (Single source is coming soon for OCCA backends.) 265- Fix some missing edge cases in CUDA backend. 266 267### Performance Improvements 268 269- MAGMA backend performance optimization and non-tensor bases. 270- No-copy optimization in {c:func}`CeedOperatorApply`. 271 272### Interface changes 273 274- Replace {code}`CeedElemRestrictionCreateIdentity` and 275 {code}`CeedElemRestrictionCreateBlocked` with more flexible 276 {c:func}`CeedElemRestrictionCreateStrided` and 277 {c:func}`CeedElemRestrictionCreateBlockedStrided`. 278- Add arguments to {c:func}`CeedQFunctionCreateIdentity`. 279- Replace ambiguous uses of {cpp:enum}`CeedTransposeMode` for L-vector identification 280 with {cpp:enum}`CeedInterlaceMode`. This is now an attribute of the 281 {cpp:type}`CeedElemRestriction` (see {c:func}`CeedElemRestrictionCreate`) and no 282 longer passed as `lmode` arguments to {c:func}`CeedOperatorSetField` and 283 {c:func}`CeedElemRestrictionApply`. 284 285### Examples 286 287libCEED-0.6 contains greatly expanded examples with {ref}`new documentation <Examples>`. 288Notable additions include: 289 290- Standalone {ref}`ex2-surface` ({file}`examples/ceed/ex2-surface`): compute the area of 291 a domain in 1, 2, and 3 dimensions by applying a Laplacian. 292 293- PETSc {ref}`example-petsc-area` ({file}`examples/petsc/area.c`): computes surface area 294 of domains (like the cube and sphere) by direct integration on a surface mesh; 295 demonstrates geometric dimension different from topological dimension. 296 297- PETSc {ref}`example-petsc-bps`: 298 299 - {file}`examples/petsc/bpsraw.c` (formerly `bps.c`): transparent CUDA support. 300 - {file}`examples/petsc/bps.c` (formerly `bpsdmplex.c`): performance improvements 301 and transparent CUDA support. 302 - {ref}`example-petsc-bps-sphere` ({file}`examples/petsc/bpssphere.c`): 303 generalizations of all CEED BPs to the surface of the sphere; demonstrates geometric 304 dimension different from topological dimension. 305 306- {ref}`example-petsc-multigrid` ({file}`examples/petsc/multigrid.c`): new p-multigrid 307 solver with algebraic multigrid coarse solve. 308 309- {ref}`example-petsc-navier-stokes` ({file}`examples/fluids/navierstokes.c`; formerly 310 `examples/navier-stokes`): unstructured grid support (using PETSc's `DMPlex`), 311 implicit time integration, SU/SUPG stabilization, free-slip boundary conditions, and 312 quasi-2D computational domain support. 313 314- {ref}`example-petsc-elasticity` ({file}`examples/solids/elasticity.c`): new solver for 315 linear elasticity, small-strain hyperelasticity, and globalized finite-strain 316 hyperelasticity using p-multigrid with algebraic multigrid coarse solve. 317 318(v0-5)= 319 320## v0.5 (Sep 18, 2019) 321 322For this release, several improvements were made. Two new CUDA backends were added to 323the family of backends, of which, the new `cuda-gen` backend achieves state-of-the-art 324performance using single-source {ref}`CeedQFunction`. From this release, users 325can define Q-Functions in a single source code independently of the targeted backend 326with the aid of a new macro `CEED QFUNCTION` to support JIT (Just-In-Time) and CPU 327compilation of the user provided {ref}`CeedQFunction` code. To allow a unified 328declaration, the {ref}`CeedQFunction` API has undergone a slight change: 329the `QFunctionField` parameter `ncomp` has been changed to `size`. This change 330requires setting the previous value of `ncomp` to `ncomp*dim` when adding a 331`QFunctionField` with eval mode `CEED EVAL GRAD`. 332 333Additionally, new CPU backends 334were included in this release, such as the `/cpu/self/opt/*` backends (which are 335written in pure C and use partial **E-vectors** to improve performance) and the 336`/cpu/self/ref/memcheck` backend (which relies upon the 337[Valgrind](http://valgrind.org/) Memcheck tool to help verify that user 338{ref}`CeedQFunction` have no undefined values). 339This release also included various performance improvements, bug fixes, new examples, 340and improved tests. Among these improvements, vectorized instructions for 341{ref}`CeedQFunction` code compiled for CPU were enhanced by using `CeedPragmaSIMD` 342instead of `CeedPragmaOMP`, implementation of a {ref}`CeedQFunction` gallery and 343identity Q-Functions were introduced, and the PETSc benchmark problems were expanded 344to include unstructured meshes handling were. For this expansion, the prior version of 345the PETSc BPs, which only included data associated with structured geometries, were 346renamed `bpsraw`, and the new version of the BPs, which can handle data associated 347with any unstructured geometry, were called `bps`. Additionally, other benchmark 348problems, namely BP2 and BP4 (the vector-valued versions of BP1 and BP3, respectively), 349and BP5 and BP6 (the collocated versions---for which the quadrature points are the same 350as the Gauss Lobatto nodes---of BP3 and BP4 respectively) were added to the PETSc 351examples. Furthermoew, another standalone libCEED example, called `ex2`, which 352computes the surface area of a given mesh was added to this release. 353 354Backends available in this release: 355 356| CEED resource (`-ceed`) | Backend | 357|--------------------------|-----------------------------------------------------| 358| `/cpu/self/ref/serial` | Serial reference implementation | 359| `/cpu/self/ref/blocked` | Blocked reference implementation | 360| `/cpu/self/ref/memcheck` | Memcheck backend, undefined value checks | 361| `/cpu/self/opt/serial` | Serial optimized C implementation | 362| `/cpu/self/opt/blocked` | Blocked optimized C implementation | 363| `/cpu/self/avx/serial` | Serial AVX implementation | 364| `/cpu/self/avx/blocked` | Blocked AVX implementation | 365| `/cpu/self/xsmm/serial` | Serial LIBXSMM implementation | 366| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation | 367| `/cpu/occa` | Serial OCCA kernels | 368| `/gpu/occa` | CUDA OCCA kernels | 369| `/omp/occa` | OpenMP OCCA kernels | 370| `/ocl/occa` | OpenCL OCCA kernels | 371| `/gpu/cuda/ref` | Reference pure CUDA kernels | 372| `/gpu/cuda/reg` | Pure CUDA kernels using one thread per element | 373| `/gpu/cuda/shared` | Optimized pure CUDA kernels using shared memory | 374| `/gpu/cuda/gen` | Optimized pure CUDA kernels using code generation | 375| `/gpu/magma` | CUDA MAGMA kernels | 376 377Examples available in this release: 378 379:::{list-table} 380:header-rows: 1 381:widths: auto 382* - User code 383 - Example 384* - `ceed` 385 - * ex1 (volume) 386 * ex2 (surface) 387* - `mfem` 388 - * BP1 (scalar mass operator) 389 * BP3 (scalar Laplace operator) 390* - `petsc` 391 - * BP1 (scalar mass operator) 392 * BP2 (vector mass operator) 393 * BP3 (scalar Laplace operator) 394 * BP4 (vector Laplace operator) 395 * BP5 (collocated scalar Laplace operator) 396 * BP6 (collocated vector Laplace operator) 397 * Navier-Stokes 398* - `nek5000` 399 - * BP1 (scalar mass operator) 400 * BP3 (scalar Laplace operator) 401::: 402 403(v0-4)= 404 405## v0.4 (Apr 1, 2019) 406 407libCEED v0.4 was made again publicly available in the second full CEED software 408distribution, release CEED 2.0. This release contained notable features, such as 409four new CPU backends, two new GPU backends, CPU backend optimizations, initial 410support for operator composition, performance benchmarking, and a Navier-Stokes demo. 411The new CPU backends in this release came in two families. The `/cpu/self/*/serial` 412backends process one element at a time and are intended for meshes with a smaller number 413of high order elements. The `/cpu/self/*/blocked` backends process blocked batches of 414eight interlaced elements and are intended for meshes with higher numbers of elements. 415The `/cpu/self/avx/*` backends rely upon AVX instructions to provide vectorized CPU 416performance. The `/cpu/self/xsmm/*` backends rely upon the 417[LIBXSMM](http://github.com/hfp/libxsmm) package to provide vectorized CPU 418performance. The `/gpu/cuda/*` backends provide GPU performance strictly using CUDA. 419The `/gpu/cuda/ref` backend is a reference CUDA backend, providing reasonable 420performance for most problem configurations. The `/gpu/cuda/reg` backend uses a simple 421parallelization approach, where each thread treats a finite element. Using just in time 422compilation, provided by nvrtc (NVidia Runtime Compiler), and runtime parameters, this 423backend unroll loops and map memory address to registers. The `/gpu/cuda/reg` backend 424achieve good peak performance for 1D, 2D, and low order 3D problems, but performance 425deteriorates very quickly when threads run out of registers. 426 427A new explicit time-stepping Navier-Stokes solver was added to the family of libCEED 428examples in the `examples/petsc` directory (see {ref}`example-petsc-navier-stokes`). 429This example solves the time-dependent Navier-Stokes equations of compressible gas 430dynamics in a static Eulerian three-dimensional frame, using structured high-order 431finite/spectral element spatial discretizations and explicit high-order time-stepping 432(available in PETSc). Moreover, the Navier-Stokes example was developed using PETSc, 433so that the pointwise physics (defined at quadrature points) is separated from the 434parallelization and meshing concerns. 435 436Backends available in this release: 437 438| CEED resource (`-ceed`) | Backend | 439|--------------------------|-----------------------------------------------------| 440| `/cpu/self/ref/serial` | Serial reference implementation | 441| `/cpu/self/ref/blocked` | Blocked reference implementation | 442| `/cpu/self/tmpl` | Backend template, defaults to `/cpu/self/blocked` | 443| `/cpu/self/avx/serial` | Serial AVX implementation | 444| `/cpu/self/avx/blocked` | Blocked AVX implementation | 445| `/cpu/self/xsmm/serial` | Serial LIBXSMM implementation | 446| `/cpu/self/xsmm/blocked` | Blocked LIBXSMM implementation | 447| `/cpu/occa` | Serial OCCA kernels | 448| `/gpu/occa` | CUDA OCCA kernels | 449| `/omp/occa` | OpenMP OCCA kernels | 450| `/ocl/occa` | OpenCL OCCA kernels | 451| `/gpu/cuda/ref` | Reference pure CUDA kernels | 452| `/gpu/cuda/reg` | Pure CUDA kernels using one thread per element | 453| `/gpu/magma` | CUDA MAGMA kernels | 454 455Examples available in this release: 456 457:::{list-table} 458:header-rows: 1 459:widths: auto 460* - User code 461 - Example 462* - `ceed` 463 - * ex1 (volume) 464* - `mfem` 465 - * BP1 (scalar mass operator) 466 * BP3 (scalar Laplace operator) 467* - `petsc` 468 - * BP1 (scalar mass operator) 469 * BP3 (scalar Laplace operator) 470 * Navier-Stokes 471* - `nek5000` 472 - * BP1 (scalar mass operator) 473 * BP3 (scalar Laplace operator) 474::: 475 476(v0-3)= 477 478## v0.3 (Sep 30, 2018) 479 480Notable features in this release include active/passive field interface, support for 481non-tensor bases, backend optimization, and improved Fortran interface. This release 482also focused on providing improved continuous integration, and many new tests with code 483coverage reports of about 90%. This release also provided a significant change to the 484public interface: a {ref}`CeedQFunction` can take any number of named input and output 485arguments while {ref}`CeedOperator` connects them to the actual data, which may be 486supplied explicitly to `CeedOperatorApply()` (active) or separately via 487`CeedOperatorSetField()` (passive). This interface change enables reusable libraries 488of CeedQFunctions and composition of block solvers constructed using 489{ref}`CeedOperator`. A concept of blocked restriction was added to this release and 490used in an optimized CPU backend. Although this is typically not visible to the user, 491it enables effective use of arbitrary-length SIMD while maintaining cache locality. 492This CPU backend also implements an algebraic factorization of tensor product gradients 493to perform fewer operations than standard application of interpolation and 494differentiation from nodes to quadrature points. This algebraic formulation 495automatically supports non-polynomial and non-interpolatory bases, thus is more general 496than the more common derivation in terms of Lagrange polynomials on the quadrature points. 497 498Backends available in this release: 499 500| CEED resource (`-ceed`) | Backend | 501|-------------------------|-----------------------------------------------------| 502| `/cpu/self/blocked` | Blocked reference implementation | 503| `/cpu/self/ref` | Serial reference implementation | 504| `/cpu/self/tmpl` | Backend template, defaults to `/cpu/self/blocked` | 505| `/cpu/occa` | Serial OCCA kernels | 506| `/gpu/occa` | CUDA OCCA kernels | 507| `/omp/occa` | OpenMP OCCA kernels | 508| `/ocl/occa` | OpenCL OCCA kernels | 509| `/gpu/magma` | CUDA MAGMA kernels | 510 511Examples available in this release: 512 513:::{list-table} 514:header-rows: 1 515:widths: auto 516* - User code 517 - Example 518* - `ceed` 519 - * ex1 (volume) 520* - `mfem` 521 - * BP1 (scalar mass operator) 522 * BP3 (scalar Laplace operator) 523* - `petsc` 524 - * BP1 (scalar mass operator) 525 * BP3 (scalar Laplace operator) 526* - `nek5000` 527 - * BP1 (scalar mass operator) 528 * BP3 (scalar Laplace operator) 529::: 530 531(v0-21)= 532 533## v0.21 (Sep 30, 2018) 534 535A MAGMA backend (which relies upon the 536[MAGMA](https://bitbucket.org/icl/magma) package) was integrated in libCEED for this 537release. This initial integration set up the framework of using MAGMA and provided the 538libCEED functionality through MAGMA kernels as one of libCEED’s computational backends. 539As any other backend, the MAGMA backend provides extended basic data structures for 540{ref}`CeedVector`, {ref}`CeedElemRestriction`, and {ref}`CeedOperator`, and implements 541the fundamental CEED building blocks to work with the new data structures. 542In general, the MAGMA-specific data structures keep the libCEED pointers to CPU data 543but also add corresponding device (e.g., GPU) pointers to the data. Coherency is handled 544internally, and thus seamlessly to the user, through the functions/methods that are 545provided to support them. 546 547Backends available in this release: 548 549| CEED resource (`-ceed`) | Backend | 550|-------------------------|---------------------------------| 551| `/cpu/self` | Serial reference implementation | 552| `/cpu/occa` | Serial OCCA kernels | 553| `/gpu/occa` | CUDA OCCA kernels | 554| `/omp/occa` | OpenMP OCCA kernels | 555| `/ocl/occa` | OpenCL OCCA kernels | 556| `/gpu/magma` | CUDA MAGMA kernels | 557 558Examples available in this release: 559 560:::{list-table} 561:header-rows: 1 562:widths: auto 563* - User code 564 - Example 565* - `ceed` 566 - * ex1 (volume) 567* - `mfem` 568 - * BP1 (scalar mass operator) 569 * BP3 (scalar Laplace operator) 570* - `petsc` 571 - * BP1 (scalar mass operator) 572* - `nek5000` 573 - * BP1 (scalar mass operator) 574::: 575 576(v0-2)= 577 578## v0.2 (Mar 30, 2018) 579 580libCEED was made publicly available the first full CEED software distribution, release 581CEED 1.0. The distribution was made available using the Spack package manager to provide 582a common, easy-to-use build environment, where the user can build the CEED distribution 583with all dependencies. This release included a new Fortran interface for the library. 584This release also contained major improvements in the OCCA backend (including a new 585`/ocl/occa` backend) and new examples. The standalone libCEED example was modified to 586compute the volume volume of a given mesh (in 1D, 2D, or 3D) and placed in an 587`examples/ceed` subfolder. A new `mfem` example to perform BP3 (with the application 588of the Laplace operator) was also added to this release. 589 590Backends available in this release: 591 592| CEED resource (`-ceed`) | Backend | 593|-------------------------|---------------------------------| 594| `/cpu/self` | Serial reference implementation | 595| `/cpu/occa` | Serial OCCA kernels | 596| `/gpu/occa` | CUDA OCCA kernels | 597| `/omp/occa` | OpenMP OCCA kernels | 598| `/ocl/occa` | OpenCL OCCA kernels | 599 600Examples available in this release: 601 602:::{list-table} 603:header-rows: 1 604:widths: auto 605* - User code 606 - Example 607* - `ceed` 608 - * ex1 (volume) 609* - `mfem` 610 - * BP1 (scalar mass operator) 611 * BP3 (scalar Laplace operator) 612* - `petsc` 613 - * BP1 (scalar mass operator) 614* - `nek5000` 615 - * BP1 (scalar mass operator) 616::: 617 618(v0-1)= 619 620## v0.1 (Jan 3, 2018) 621 622Initial low-level API of the CEED project. The low-level API provides a set of Finite 623Elements kernels and components for writing new low-level kernels. Examples include: 624vector and sparse linear algebra, element matrix assembly over a batch of elements, 625partial assembly and action for efficient high-order operators like mass, diffusion, 626advection, etc. The main goal of the low-level API is to establish the basis for the 627high-level API. Also, identifying such low-level kernels and providing a reference 628implementation for them serves as the basis for specialized backend implementations. 629This release contained several backends: `/cpu/self`, and backends which rely upon the 630[OCCA](http://github.com/libocca/occa) package, such as `/cpu/occa`, 631`/gpu/occa`, and `/omp/occa`. 632It also included several examples, in the `examples` folder: 633A standalone code that shows the usage of libCEED (with no external 634dependencies) to apply the Laplace operator, `ex1`; an `mfem` example to perform BP1 635(with the application of the mass operator); and a `petsc` example to perform BP1 636(with the application of the mass operator). 637 638Backends available in this release: 639 640| CEED resource (`-ceed`) | Backend | 641|-------------------------|---------------------------------| 642| `/cpu/self` | Serial reference implementation | 643| `/cpu/occa` | Serial OCCA kernels | 644| `/gpu/occa` | CUDA OCCA kernels | 645| `/omp/occa` | OpenMP OCCA kernels | 646 647Examples available in this release: 648 649| User code | Example | 650|-----------------------|-----------------------------------| 651| `ceed` | ex1 (scalar Laplace operator) | 652| `mfem` | BP1 (scalar mass operator) | 653| `petsc` | BP1 (scalar mass operator) | 654``` 655